MRI_brain_v3_CONV3D_colab.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "name": "MRI_brain_v3_CONV3D_colab.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.7.9"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "78J78_UthZU7"
      },
      "source": [
        "* todo: add augmentation\n",
        "* todo: seperate the validation_generator from train_genarator after augmentation"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "P10DNOwlhe1u",
        "outputId": "0cf1a2f6-b1dd-4a2e-cd19-b58ca4a67266"
      },
      "source": [
        "!wget http://biomedic.doc.ic.ac.uk/brain-development/downloads/IXI/IXI-T1.tar"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "URL transformed to HTTPS due to an HSTS policy\n",
            "--2021-06-04 11:56:00--  https://biomedic.doc.ic.ac.uk/brain-development/downloads/IXI/IXI-T1.tar\n",
            "Resolving biomedic.doc.ic.ac.uk (biomedic.doc.ic.ac.uk)... 146.169.22.31\n",
            "Connecting to biomedic.doc.ic.ac.uk (biomedic.doc.ic.ac.uk)|146.169.22.31|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 4840816640 (4.5G) [application/x-tar]\n",
            "Saving to: ‘IXI-T1.tar’\n",
            "\n",
            "IXI-T1.tar          100%[===================>]   4.51G  21.4MB/s    in 3m 37s  \n",
            "\n",
            "2021-06-04 11:59:37 (21.3 MB/s) - ‘IXI-T1.tar’ saved [4840816640/4840816640]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yP38dkXDhgaF",
        "outputId": "8107d272-6787-4a68-d53a-f3a9df0ac2df"
      },
      "source": [
        "!wget http://biomedic.doc.ic.ac.uk/brain-development/downloads/IXI/IXI.xls"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "URL transformed to HTTPS due to an HSTS policy\n",
            "--2021-06-04 11:59:37--  https://biomedic.doc.ic.ac.uk/brain-development/downloads/IXI/IXI.xls\n",
            "Resolving biomedic.doc.ic.ac.uk (biomedic.doc.ic.ac.uk)... 146.169.22.31\n",
            "Connecting to biomedic.doc.ic.ac.uk (biomedic.doc.ic.ac.uk)|146.169.22.31|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 210432 (206K) [application/vnd.ms-excel]\n",
            "Saving to: ‘IXI.xls.1’\n",
            "\n",
            "IXI.xls.1           100%[===================>] 205.50K   536KB/s    in 0.4s    \n",
            "\n",
            "2021-06-04 11:59:39 (536 KB/s) - ‘IXI.xls.1’ saved [210432/210432]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5zFn5SC5hhs2",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "3a530530-6eb3-4149-cc9a-7fe9a9704158"
      },
      "source": [
        "!mkdir dataset"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "mkdir: cannot create directory ‘dataset’: File exists\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jcG9eSBthitG",
        "outputId": "e9993731-671f-48b7-f99c-334dd902de10"
      },
      "source": [
        "!tar -xvf  'IXI-T1.tar' -C 'dataset'"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "IXI002-Guys-0828-T1.nii.gz\n",
            "IXI012-HH-1211-T1.nii.gz\n",
            "IXI013-HH-1212-T1.nii.gz\n",
            "IXI014-HH-1236-T1.nii.gz\n",
            "IXI015-HH-1258-T1.nii.gz\n",
            "IXI016-Guys-0697-T1.nii.gz\n",
            "IXI017-Guys-0698-T1.nii.gz\n",
            "IXI019-Guys-0702-T1.nii.gz\n",
            "IXI020-Guys-0700-T1.nii.gz\n",
            "IXI021-Guys-0703-T1.nii.gz\n",
            "IXI022-Guys-0701-T1.nii.gz\n",
            "IXI023-Guys-0699-T1.nii.gz\n",
            "IXI024-Guys-0705-T1.nii.gz\n",
            "IXI025-Guys-0852-T1.nii.gz\n",
            "IXI026-Guys-0696-T1.nii.gz\n",
            "IXI027-Guys-0710-T1.nii.gz\n",
            "IXI028-Guys-1038-T1.nii.gz\n",
            "IXI029-Guys-0829-T1.nii.gz\n",
            "IXI030-Guys-0708-T1.nii.gz\n",
            "IXI031-Guys-0797-T1.nii.gz\n",
            "IXI033-HH-1259-T1.nii.gz\n",
            "IXI034-HH-1260-T1.nii.gz\n",
            "IXI035-IOP-0873-T1.nii.gz\n",
            "IXI036-Guys-0736-T1.nii.gz\n",
            "IXI037-Guys-0704-T1.nii.gz\n",
            "IXI038-Guys-0729-T1.nii.gz\n",
            "IXI039-HH-1261-T1.nii.gz\n",
            "IXI040-Guys-0724-T1.nii.gz\n",
            "IXI041-Guys-0706-T1.nii.gz\n",
            "IXI042-Guys-0725-T1.nii.gz\n",
            "IXI043-Guys-0714-T1.nii.gz\n",
            "IXI044-Guys-0712-T1.nii.gz\n",
            "IXI045-Guys-0713-T1.nii.gz\n",
            "IXI046-Guys-0824-T1.nii.gz\n",
            "IXI048-HH-1326-T1.nii.gz\n",
            "IXI049-HH-1358-T1.nii.gz\n",
            "IXI050-Guys-0711-T1.nii.gz\n",
            "IXI051-HH-1328-T1.nii.gz\n",
            "IXI052-HH-1343-T1.nii.gz\n",
            "IXI053-Guys-0727-T1.nii.gz\n",
            "IXI054-Guys-0707-T1.nii.gz\n",
            "IXI055-Guys-0730-T1.nii.gz\n",
            "IXI056-HH-1327-T1.nii.gz\n",
            "IXI057-HH-1342-T1.nii.gz\n",
            "IXI058-Guys-0726-T1.nii.gz\n",
            "IXI059-HH-1284-T1.nii.gz\n",
            "IXI060-Guys-0709-T1.nii.gz\n",
            "IXI061-Guys-0715-T1.nii.gz\n",
            "IXI062-Guys-0740-T1.nii.gz\n",
            "IXI063-Guys-0742-T1.nii.gz\n",
            "IXI064-Guys-0743-T1.nii.gz\n",
            "IXI065-Guys-0744-T1.nii.gz\n",
            "IXI066-Guys-0731-T1.nii.gz\n",
            "IXI067-HH-1356-T1.nii.gz\n",
            "IXI068-Guys-0756-T1.nii.gz\n",
            "IXI069-Guys-0769-T1.nii.gz\n",
            "IXI070-Guys-0767-T1.nii.gz\n",
            "IXI071-Guys-0770-T1.nii.gz\n",
            "IXI072-HH-2324-T1.nii.gz\n",
            "IXI073-Guys-0755-T1.nii.gz\n",
            "IXI074-Guys-0771-T1.nii.gz\n",
            "IXI075-Guys-0754-T1.nii.gz\n",
            "IXI076-Guys-0753-T1.nii.gz\n",
            "IXI077-Guys-0752-T1.nii.gz\n",
            "IXI078-Guys-0751-T1.nii.gz\n",
            "IXI079-HH-1388-T1.nii.gz\n",
            "IXI080-HH-1341-T1.nii.gz\n",
            "IXI081-Guys-0855-T1.nii.gz\n",
            "IXI083-HH-1357-T1.nii.gz\n",
            "IXI084-Guys-0741-T1.nii.gz\n",
            "IXI085-Guys-0759-T1.nii.gz\n",
            "IXI086-Guys-0728-T1.nii.gz\n",
            "IXI087-Guys-0768-T1.nii.gz\n",
            "IXI088-Guys-0758-T1.nii.gz\n",
            "IXI089-Guys-0757-T1.nii.gz\n",
            "IXI090-Guys-0800-T1.nii.gz\n",
            "IXI091-Guys-0762-T1.nii.gz\n",
            "IXI092-HH-1436-T1.nii.gz\n",
            "IXI093-HH-1359-T1.nii.gz\n",
            "IXI094-HH-1355-T1.nii.gz\n",
            "IXI095-HH-1390-T1.nii.gz\n",
            "IXI096-HH-1391-T1.nii.gz\n",
            "IXI097-HH-1619-T1.nii.gz\n",
            "IXI098-Guys-0745-T1.nii.gz\n",
            "IXI099-Guys-0748-T1.nii.gz\n",
            "IXI100-Guys-0747-T1.nii.gz\n",
            "IXI101-Guys-0749-T1.nii.gz\n",
            "IXI102-HH-1416-T1.nii.gz\n",
            "IXI103-Guys-0750-T1.nii.gz\n",
            "IXI104-HH-1450-T1.nii.gz\n",
            "IXI105-HH-1471-T1.nii.gz\n",
            "IXI106-Guys-0760-T1.nii.gz\n",
            "IXI107-Guys-0761-T1.nii.gz\n",
            "IXI108-Guys-0865-T1.nii.gz\n",
            "IXI109-Guys-0732-T1.nii.gz\n",
            "IXI110-Guys-0733-T1.nii.gz\n",
            "IXI111-Guys-0734-T1.nii.gz\n",
            "IXI112-Guys-0735-T1.nii.gz\n",
            "IXI113-Guys-0776-T1.nii.gz\n",
            "IXI114-Guys-0737-T1.nii.gz\n",
            "IXI115-Guys-0738-T1.nii.gz\n",
            "IXI116-Guys-0739-T1.nii.gz\n",
            "IXI117-Guys-0763-T1.nii.gz\n",
            "IXI118-Guys-0764-T1.nii.gz\n",
            "IXI119-Guys-0765-T1.nii.gz\n",
            "IXI120-Guys-0766-T1.nii.gz\n",
            "IXI121-Guys-0772-T1.nii.gz\n",
            "IXI122-Guys-0773-T1.nii.gz\n",
            "IXI123-Guys-0774-T1.nii.gz\n",
            "IXI126-HH-1437-T1.nii.gz\n",
            "IXI127-HH-1451-T1.nii.gz\n",
            "IXI128-HH-1470-T1.nii.gz\n",
            "IXI129-Guys-0775-T1.nii.gz\n",
            "IXI130-HH-1528-T1.nii.gz\n",
            "IXI131-HH-1527-T1.nii.gz\n",
            "IXI132-HH-1415-T1.nii.gz\n",
            "IXI134-Guys-0780-T1.nii.gz\n",
            "IXI135-Guys-0779-T1.nii.gz\n",
            "IXI136-HH-1452-T1.nii.gz\n",
            "IXI137-HH-1472-T1.nii.gz\n",
            "IXI138-Guys-0746-T1.nii.gz\n",
            "IXI139-Guys-0815-T1.nii.gz\n",
            "IXI140-Guys-0787-T1.nii.gz\n",
            "IXI141-Guys-0789-T1.nii.gz\n",
            "IXI142-Guys-0786-T1.nii.gz\n",
            "IXI143-Guys-0785-T1.nii.gz\n",
            "IXI144-Guys-0788-T1.nii.gz\n",
            "IXI145-Guys-0781-T1.nii.gz\n",
            "IXI146-HH-1389-T1.nii.gz\n",
            "IXI148-HH-1453-T1.nii.gz\n",
            "IXI150-HH-1550-T1.nii.gz\n",
            "IXI151-Guys-0793-T1.nii.gz\n",
            "IXI153-Guys-0782-T1.nii.gz\n",
            "IXI154-Guys-0821-T1.nii.gz\n",
            "IXI156-Guys-0837-T1.nii.gz\n",
            "IXI157-Guys-0816-T1.nii.gz\n",
            "IXI158-Guys-0783-T1.nii.gz\n",
            "IXI159-HH-1549-T1.nii.gz\n",
            "IXI160-HH-1637-T1.nii.gz\n",
            "IXI161-HH-2533-T1.nii.gz\n",
            "IXI162-HH-1548-T1.nii.gz\n",
            "IXI163-HH-1621-T1.nii.gz\n",
            "IXI164-Guys-0844-T1.nii.gz\n",
            "IXI165-HH-1589-T1.nii.gz\n",
            "IXI166-Guys-0846-T1.nii.gz\n",
            "IXI167-HH-1569-T1.nii.gz\n",
            "IXI168-HH-1607-T1.nii.gz\n",
            "IXI169-Guys-0842-T1.nii.gz\n",
            "IXI170-Guys-0843-T1.nii.gz\n",
            "IXI172-Guys-0982-T1.nii.gz\n",
            "IXI173-HH-1590-T1.nii.gz\n",
            "IXI174-HH-1571-T1.nii.gz\n",
            "IXI175-HH-1570-T1.nii.gz\n",
            "IXI176-HH-1604-T1.nii.gz\n",
            "IXI177-Guys-0831-T1.nii.gz\n",
            "IXI178-Guys-0778-T1.nii.gz\n",
            "IXI179-Guys-0777-T1.nii.gz\n",
            "IXI180-HH-1605-T1.nii.gz\n",
            "IXI181-Guys-0790-T1.nii.gz\n",
            "IXI182-Guys-0792-T1.nii.gz\n",
            "IXI183-Guys-0791-T1.nii.gz\n",
            "IXI184-Guys-0794-T1.nii.gz\n",
            "IXI185-Guys-0795-T1.nii.gz\n",
            "IXI186-Guys-0796-T1.nii.gz\n",
            "IXI188-Guys-0798-T1.nii.gz\n",
            "IXI189-Guys-0799-T1.nii.gz\n",
            "IXI191-Guys-0801-T1.nii.gz\n",
            "IXI192-Guys-0878-T1.nii.gz\n",
            "IXI193-Guys-0810-T1.nii.gz\n",
            "IXI194-Guys-0818-T1.nii.gz\n",
            "IXI195-HH-1620-T1.nii.gz\n",
            "IXI196-Guys-0805-T1.nii.gz\n",
            "IXI197-Guys-0811-T1.nii.gz\n",
            "IXI198-Guys-0803-T1.nii.gz\n",
            "IXI199-Guys-0802-T1.nii.gz\n",
            "IXI200-Guys-0812-T1.nii.gz\n",
            "IXI201-HH-1588-T1.nii.gz\n",
            "IXI202-HH-1526-T1.nii.gz\n",
            "IXI204-HH-1651-T1.nii.gz\n",
            "IXI205-HH-1649-T1.nii.gz\n",
            "IXI206-HH-1650-T1.nii.gz\n",
            "IXI207-Guys-0809-T1.nii.gz\n",
            "IXI208-Guys-0808-T1.nii.gz\n",
            "IXI209-Guys-0804-T1.nii.gz\n",
            "IXI210-Guys-0856-T1.nii.gz\n",
            "IXI211-HH-1568-T1.nii.gz\n",
            "IXI212-HH-1643-T1.nii.gz\n",
            "IXI213-HH-1642-T1.nii.gz\n",
            "IXI214-HH-1636-T1.nii.gz\n",
            "IXI216-HH-1635-T1.nii.gz\n",
            "IXI217-HH-1638-T1.nii.gz\n",
            "IXI218-HH-1815-T1.nii.gz\n",
            "IXI219-Guys-0894-T1.nii.gz\n",
            "IXI221-HH-1606-T1.nii.gz\n",
            "IXI222-Guys-0819-T1.nii.gz\n",
            "IXI223-Guys-0830-T1.nii.gz\n",
            "IXI224-Guys-0823-T1.nii.gz\n",
            "IXI225-Guys-0832-T1.nii.gz\n",
            "IXI226-HH-1618-T1.nii.gz\n",
            "IXI227-Guys-0813-T1.nii.gz\n",
            "IXI228-Guys-0822-T1.nii.gz\n",
            "IXI229-Guys-0980-T1.nii.gz\n",
            "IXI230-IOP-0869-T1.nii.gz\n",
            "IXI231-IOP-0866-T1.nii.gz\n",
            "IXI232-IOP-0898-T1.nii.gz\n",
            "IXI233-IOP-0875-T1.nii.gz\n",
            "IXI234-IOP-0870-T1.nii.gz\n",
            "IXI236-Guys-0817-T1.nii.gz\n",
            "IXI237-Guys-1049-T1.nii.gz\n",
            "IXI238-IOP-0883-T1.nii.gz\n",
            "IXI239-HH-2296-T1.nii.gz\n",
            "IXI240-Guys-0834-T1.nii.gz\n",
            "IXI241-Guys-0833-T1.nii.gz\n",
            "IXI242-HH-1722-T1.nii.gz\n",
            "IXI244-Guys-0841-T1.nii.gz\n",
            "IXI246-Guys-0840-T1.nii.gz\n",
            "IXI247-Guys-0838-T1.nii.gz\n",
            "IXI248-HH-1972-T1.nii.gz\n",
            "IXI249-Guys-1072-T1.nii.gz\n",
            "IXI250-Guys-0836-T1.nii.gz\n",
            "IXI251-Guys-1055-T1.nii.gz\n",
            "IXI252-HH-1693-T1.nii.gz\n",
            "IXI253-HH-1694-T1.nii.gz\n",
            "IXI254-HH-1705-T1.nii.gz\n",
            "IXI255-HH-1882-T1.nii.gz\n",
            "IXI256-HH-1723-T1.nii.gz\n",
            "IXI257-HH-1724-T1.nii.gz\n",
            "IXI258-HH-1769-T1.nii.gz\n",
            "IXI259-HH-1804-T1.nii.gz\n",
            "IXI260-HH-1805-T1.nii.gz\n",
            "IXI261-HH-1704-T1.nii.gz\n",
            "IXI262-HH-1861-T1.nii.gz\n",
            "IXI263-HH-1684-T1.nii.gz\n",
            "IXI264-Guys-0854-T1.nii.gz\n",
            "IXI265-Guys-0845-T1.nii.gz\n",
            "IXI266-Guys-0853-T1.nii.gz\n",
            "IXI267-HH-1772-T1.nii.gz\n",
            "IXI268-Guys-0858-T1.nii.gz\n",
            "IXI269-Guys-0839-T1.nii.gz\n",
            "IXI270-Guys-0847-T1.nii.gz\n",
            "IXI274-HH-2294-T1.nii.gz\n",
            "IXI275-HH-1803-T1.nii.gz\n",
            "IXI276-HH-1840-T1.nii.gz\n",
            "IXI277-HH-1770-T1.nii.gz\n",
            "IXI278-HH-1771-T1.nii.gz\n",
            "IXI279-Guys-1044-T1.nii.gz\n",
            "IXI280-HH-1860-T1.nii.gz\n",
            "IXI282-HH-2025-T1.nii.gz\n",
            "IXI284-HH-2354-T1.nii.gz\n",
            "IXI285-Guys-0857-T1.nii.gz\n",
            "IXI286-Guys-0859-T1.nii.gz\n",
            "IXI287-Guys-0863-T1.nii.gz\n",
            "IXI288-Guys-0879-T1.nii.gz\n",
            "IXI289-Guys-0864-T1.nii.gz\n",
            "IXI290-IOP-0874-T1.nii.gz\n",
            "IXI291-IOP-0882-T1.nii.gz\n",
            "IXI292-IOP-0877-T1.nii.gz\n",
            "IXI293-IOP-0876-T1.nii.gz\n",
            "IXI294-IOP-0868-T1.nii.gz\n",
            "IXI295-HH-1814-T1.nii.gz\n",
            "IXI296-HH-1970-T1.nii.gz\n",
            "IXI297-Guys-0886-T1.nii.gz\n",
            "IXI298-Guys-0861-T1.nii.gz\n",
            "IXI299-Guys-0893-T1.nii.gz\n",
            "IXI300-Guys-0880-T1.nii.gz\n",
            "IXI302-HH-1883-T1.nii.gz\n",
            "IXI303-IOP-0968-T1.nii.gz\n",
            "IXI304-Guys-0862-T1.nii.gz\n",
            "IXI305-IOP-0871-T1.nii.gz\n",
            "IXI306-IOP-0867-T1.nii.gz\n",
            "IXI307-IOP-0872-T1.nii.gz\n",
            "IXI308-Guys-0884-T1.nii.gz\n",
            "IXI309-IOP-0897-T1.nii.gz\n",
            "IXI310-IOP-0890-T1.nii.gz\n",
            "IXI311-Guys-0885-T1.nii.gz\n",
            "IXI312-Guys-0887-T1.nii.gz\n",
            "IXI313-HH-2241-T1.nii.gz\n",
            "IXI314-IOP-0889-T1.nii.gz\n",
            "IXI315-IOP-0888-T1.nii.gz\n",
            "IXI316-HH-1862-T1.nii.gz\n",
            "IXI317-Guys-0896-T1.nii.gz\n",
            "IXI318-Guys-0895-T1.nii.gz\n",
            "IXI319-Guys-0901-T1.nii.gz\n",
            "IXI320-Guys-0902-T1.nii.gz\n",
            "IXI321-Guys-0903-T1.nii.gz\n",
            "IXI322-IOP-0891-T1.nii.gz\n",
            "IXI324-Guys-0922-T1.nii.gz\n",
            "IXI325-Guys-0911-T1.nii.gz\n",
            "IXI326-Guys-0907-T1.nii.gz\n",
            "IXI327-HH-1999-T1.nii.gz\n",
            "IXI328-HH-2295-T1.nii.gz\n",
            "IXI329-HH-1908-T1.nii.gz\n",
            "IXI330-Guys-0881-T1.nii.gz\n",
            "IXI331-IOP-0892-T1.nii.gz\n",
            "IXI332-IOP-1134-T1.nii.gz\n",
            "IXI333-IOP-0926-T1.nii.gz\n",
            "IXI334-HH-1907-T1.nii.gz\n",
            "IXI335-HH-1906-T1.nii.gz\n",
            "IXI336-Guys-0904-T1.nii.gz\n",
            "IXI337-IOP-0929-T1.nii.gz\n",
            "IXI338-HH-1971-T1.nii.gz\n",
            "IXI340-IOP-0915-T1.nii.gz\n",
            "IXI341-Guys-0906-T1.nii.gz\n",
            "IXI342-Guys-0909-T1.nii.gz\n",
            "IXI344-Guys-0905-T1.nii.gz\n",
            "IXI345-IOP-0928-T1.nii.gz\n",
            "IXI347-IOP-0927-T1.nii.gz\n",
            "IXI348-Guys-0910-T1.nii.gz\n",
            "IXI350-Guys-0908-T1.nii.gz\n",
            "IXI351-Guys-0914-T1.nii.gz\n",
            "IXI353-HH-1996-T1.nii.gz\n",
            "IXI354-HH-2024-T1.nii.gz\n",
            "IXI356-HH-2049-T1.nii.gz\n",
            "IXI357-HH-2076-T1.nii.gz\n",
            "IXI358-Guys-0919-T1.nii.gz\n",
            "IXI359-Guys-0918-T1.nii.gz\n",
            "IXI360-Guys-0912-T1.nii.gz\n",
            "IXI361-Guys-0913-T1.nii.gz\n",
            "IXI362-HH-2051-T1.nii.gz\n",
            "IXI363-HH-2050-T1.nii.gz\n",
            "IXI364-Guys-0920-T1.nii.gz\n",
            "IXI365-Guys-0923-T1.nii.gz\n",
            "IXI367-Guys-0917-T1.nii.gz\n",
            "IXI368-Guys-0916-T1.nii.gz\n",
            "IXI369-Guys-0924-T1.nii.gz\n",
            "IXI370-Guys-0921-T1.nii.gz\n",
            "IXI371-IOP-0970-T1.nii.gz\n",
            "IXI372-IOP-0971-T1.nii.gz\n",
            "IXI373-IOP-0967-T1.nii.gz\n",
            "IXI375-Guys-0925-T1.nii.gz\n",
            "IXI376-Guys-0938-T1.nii.gz\n",
            "IXI377-Guys-0937-T1.nii.gz\n",
            "IXI378-IOP-0972-T1.nii.gz\n",
            "IXI379-Guys-0943-T1.nii.gz\n",
            "IXI380-Guys-0944-T1.nii.gz\n",
            "IXI381-Guys-1024-T1.nii.gz\n",
            "IXI382-IOP-1135-T1.nii.gz\n",
            "IXI383-HH-2099-T1.nii.gz\n",
            "IXI384-HH-2100-T1.nii.gz\n",
            "IXI385-HH-2078-T1.nii.gz\n",
            "IXI386-HH-2077-T1.nii.gz\n",
            "IXI387-HH-2101-T1.nii.gz\n",
            "IXI388-IOP-0973-T1.nii.gz\n",
            "IXI389-Guys-0930-T1.nii.gz\n",
            "IXI390-Guys-0931-T1.nii.gz\n",
            "IXI391-Guys-0934-T1.nii.gz\n",
            "IXI392-Guys-1064-T1.nii.gz\n",
            "IXI393-Guys-0941-T1.nii.gz\n",
            "IXI394-Guys-0940-T1.nii.gz\n",
            "IXI395-IOP-0969-T1.nii.gz\n",
            "IXI396-HH-2115-T1.nii.gz\n",
            "IXI397-Guys-0953-T1.nii.gz\n",
            "IXI398-Guys-0952-T1.nii.gz\n",
            "IXI399-Guys-0966-T1.nii.gz\n",
            "IXI400-Guys-0977-T1.nii.gz\n",
            "IXI401-Guys-0978-T1.nii.gz\n",
            "IXI402-Guys-0961-T1.nii.gz\n",
            "IXI403-Guys-0965-T1.nii.gz\n",
            "IXI404-Guys-0950-T1.nii.gz\n",
            "IXI405-Guys-0948-T1.nii.gz\n",
            "IXI406-Guys-0963-T1.nii.gz\n",
            "IXI407-Guys-0964-T1.nii.gz\n",
            "IXI408-Guys-0962-T1.nii.gz\n",
            "IXI409-Guys-0960-T1.nii.gz\n",
            "IXI410-Guys-0958-T1.nii.gz\n",
            "IXI411-Guys-0959-T1.nii.gz\n",
            "IXI412-Guys-0949-T1.nii.gz\n",
            "IXI413-Guys-0955-T1.nii.gz\n",
            "IXI414-Guys-0957-T1.nii.gz\n",
            "IXI415-Guys-0942-T1.nii.gz\n",
            "IXI416-Guys-1051-T1.nii.gz\n",
            "IXI417-Guys-0939-T1.nii.gz\n",
            "IXI418-Guys-0956-T1.nii.gz\n",
            "IXI419-Guys-1027-T1.nii.gz\n",
            "IXI420-Guys-1028-T1.nii.gz\n",
            "IXI422-Guys-1071-T1.nii.gz\n",
            "IXI423-IOP-0974-T1.nii.gz\n",
            "IXI424-IOP-0991-T1.nii.gz\n",
            "IXI425-IOP-0988-T1.nii.gz\n",
            "IXI426-IOP-1011-T1.nii.gz\n",
            "IXI427-IOP-1012-T1.nii.gz\n",
            "IXI428-Guys-0996-T1.nii.gz\n",
            "IXI429-Guys-0997-T1.nii.gz\n",
            "IXI430-IOP-0990-T1.nii.gz\n",
            "IXI431-Guys-0986-T1.nii.gz\n",
            "IXI432-Guys-0987-T1.nii.gz\n",
            "IXI433-IOP-0989-T1.nii.gz\n",
            "IXI434-IOP-1010-T1.nii.gz\n",
            "IXI435-IOP-1040-T1.nii.gz\n",
            "IXI436-HH-2153-T1.nii.gz\n",
            "IXI437-HH-2152-T1.nii.gz\n",
            "IXI438-HH-2155-T1.nii.gz\n",
            "IXI439-HH-2114-T1.nii.gz\n",
            "IXI440-HH-2127-T1.nii.gz\n",
            "IXI441-HH-2154-T1.nii.gz\n",
            "IXI442-IOP-1041-T1.nii.gz\n",
            "IXI443-HH-2215-T1.nii.gz\n",
            "IXI444-HH-2270-T1.nii.gz\n",
            "IXI445-HH-2269-T1.nii.gz\n",
            "IXI446-HH-2325-T1.nii.gz\n",
            "IXI447-Guys-0979-T1.nii.gz\n",
            "IXI448-HH-2393-T1.nii.gz\n",
            "IXI449-Guys-1082-T1.nii.gz\n",
            "IXI450-Guys-1093-T1.nii.gz\n",
            "IXI451-HH-2212-T1.nii.gz\n",
            "IXI452-HH-2213-T1.nii.gz\n",
            "IXI453-HH-2214-T1.nii.gz\n",
            "IXI454-Guys-1083-T1.nii.gz\n",
            "IXI455-Guys-0981-T1.nii.gz\n",
            "IXI456-Guys-1019-T1.nii.gz\n",
            "IXI457-Guys-0984-T1.nii.gz\n",
            "IXI458-Guys-0993-T1.nii.gz\n",
            "IXI459-Guys-0992-T1.nii.gz\n",
            "IXI460-Guys-0999-T1.nii.gz\n",
            "IXI461-Guys-0998-T1.nii.gz\n",
            "IXI462-IOP-1042-T1.nii.gz\n",
            "IXI463-IOP-1043-T1.nii.gz\n",
            "IXI464-IOP-1029-T1.nii.gz\n",
            "IXI465-HH-2176-T1.nii.gz\n",
            "IXI467-Guys-0983-T1.nii.gz\n",
            "IXI468-Guys-0985-T1.nii.gz\n",
            "IXI469-IOP-1136-T1.nii.gz\n",
            "IXI470-IOP-1030-T1.nii.gz\n",
            "IXI473-IOP-1137-T1.nii.gz\n",
            "IXI474-IOP-1138-T1.nii.gz\n",
            "IXI475-IOP-1139-T1.nii.gz\n",
            "IXI476-IOP-1140-T1.nii.gz\n",
            "IXI477-IOP-1141-T1.nii.gz\n",
            "IXI478-IOP-1142-T1.nii.gz\n",
            "IXI479-Guys-1034-T1.nii.gz\n",
            "IXI480-Guys-1033-T1.nii.gz\n",
            "IXI481-HH-2175-T1.nii.gz\n",
            "IXI482-HH-2178-T1.nii.gz\n",
            "IXI483-HH-2177-T1.nii.gz\n",
            "IXI484-HH-2179-T1.nii.gz\n",
            "IXI485-HH-2180-T1.nii.gz\n",
            "IXI486-Guys-1005-T1.nii.gz\n",
            "IXI487-Guys-1037-T1.nii.gz\n",
            "IXI488-Guys-1015-T1.nii.gz\n",
            "IXI489-Guys-1014-T1.nii.gz\n",
            "IXI490-Guys-1018-T1.nii.gz\n",
            "IXI491-Guys-1032-T1.nii.gz\n",
            "IXI492-Guys-1022-T1.nii.gz\n",
            "IXI493-Guys-1007-T1.nii.gz\n",
            "IXI494-Guys-1008-T1.nii.gz\n",
            "IXI495-Guys-1009-T1.nii.gz\n",
            "IXI496-Guys-1045-T1.nii.gz\n",
            "IXI497-Guys-1002-T1.nii.gz\n",
            "IXI498-Guys-1050-T1.nii.gz\n",
            "IXI499-Guys-1004-T1.nii.gz\n",
            "IXI500-Guys-1017-T1.nii.gz\n",
            "IXI501-Guys-1016-T1.nii.gz\n",
            "IXI502-Guys-1020-T1.nii.gz\n",
            "IXI503-Guys-1021-T1.nii.gz\n",
            "IXI504-Guys-1025-T1.nii.gz\n",
            "IXI505-Guys-1026-T1.nii.gz\n",
            "IXI506-Guys-1035-T1.nii.gz\n",
            "IXI507-Guys-1036-T1.nii.gz\n",
            "IXI508-HH-2268-T1.nii.gz\n",
            "IXI510-IOP-1143-T1.nii.gz\n",
            "IXI511-HH-2238-T1.nii.gz\n",
            "IXI512-Guys-1054-T1.nii.gz\n",
            "IXI515-HH-2377-T1.nii.gz\n",
            "IXI516-HH-2297-T1.nii.gz\n",
            "IXI517-IOP-1144-T1.nii.gz\n",
            "IXI518-HH-2239-T1.nii.gz\n",
            "IXI519-HH-2240-T1.nii.gz\n",
            "IXI521-HH-2353-T1.nii.gz\n",
            "IXI522-HH-2453-T1.nii.gz\n",
            "IXI523-Guys-1056-T1.nii.gz\n",
            "IXI524-HH-2412-T1.nii.gz\n",
            "IXI525-HH-2413-T1.nii.gz\n",
            "IXI526-HH-2392-T1.nii.gz\n",
            "IXI527-HH-2376-T1.nii.gz\n",
            "IXI528-Guys-1073-T1.nii.gz\n",
            "IXI531-Guys-1057-T1.nii.gz\n",
            "IXI532-IOP-1145-T1.nii.gz\n",
            "IXI533-Guys-1066-T1.nii.gz\n",
            "IXI534-Guys-1062-T1.nii.gz\n",
            "IXI535-Guys-1061-T1.nii.gz\n",
            "IXI536-Guys-1059-T1.nii.gz\n",
            "IXI537-HH-2378-T1.nii.gz\n",
            "IXI538-HH-2411-T1.nii.gz\n",
            "IXI539-Guys-1067-T1.nii.gz\n",
            "IXI541-IOP-1146-T1.nii.gz\n",
            "IXI542-IOP-1147-T1.nii.gz\n",
            "IXI543-IOP-1148-T1.nii.gz\n",
            "IXI544-HH-2395-T1.nii.gz\n",
            "IXI546-HH-2450-T1.nii.gz\n",
            "IXI547-IOP-1149-T1.nii.gz\n",
            "IXI548-IOP-1150-T1.nii.gz\n",
            "IXI549-Guys-1046-T1.nii.gz\n",
            "IXI550-Guys-1069-T1.nii.gz\n",
            "IXI551-Guys-1065-T1.nii.gz\n",
            "IXI552-Guys-1063-T1.nii.gz\n",
            "IXI553-IOP-1151-T1.nii.gz\n",
            "IXI554-Guys-1068-T1.nii.gz\n",
            "IXI555-Guys-1074-T1.nii.gz\n",
            "IXI556-HH-2452-T1.nii.gz\n",
            "IXI558-Guys-1079-T1.nii.gz\n",
            "IXI559-HH-2394-T1.nii.gz\n",
            "IXI560-Guys-1070-T1.nii.gz\n",
            "IXI561-IOP-1152-T1.nii.gz\n",
            "IXI562-Guys-1131-T1.nii.gz\n",
            "IXI563-IOP-1153-T1.nii.gz\n",
            "IXI565-HH-2534-T1.nii.gz\n",
            "IXI566-HH-2535-T1.nii.gz\n",
            "IXI567-HH-2536-T1.nii.gz\n",
            "IXI568-HH-2607-T1.nii.gz\n",
            "IXI569-Guys-1101-T1.nii.gz\n",
            "IXI571-IOP-1154-T1.nii.gz\n",
            "IXI572-HH-2605-T1.nii.gz\n",
            "IXI573-IOP-1155-T1.nii.gz\n",
            "IXI574-IOP-1156-T1.nii.gz\n",
            "IXI575-HH-2658-T1.nii.gz\n",
            "IXI576-Guys-1077-T1.nii.gz\n",
            "IXI577-HH-2661-T1.nii.gz\n",
            "IXI578-Guys-1078-T1.nii.gz\n",
            "IXI579-Guys-1126-T1.nii.gz\n",
            "IXI582-Guys-1127-T1.nii.gz\n",
            "IXI584-Guys-1129-T1.nii.gz\n",
            "IXI585-Guys-1130-T1.nii.gz\n",
            "IXI586-HH-2451-T1.nii.gz\n",
            "IXI587-Guys-1128-T1.nii.gz\n",
            "IXI588-IOP-1158-T1.nii.gz\n",
            "IXI589-Guys-1080-T1.nii.gz\n",
            "IXI591-Guys-1084-T1.nii.gz\n",
            "IXI592-Guys-1085-T1.nii.gz\n",
            "IXI593-Guys-1109-T1.nii.gz\n",
            "IXI594-Guys-1089-T1.nii.gz\n",
            "IXI595-IOP-1159-T1.nii.gz\n",
            "IXI596-IOP-1160-T1.nii.gz\n",
            "IXI597-IOP-1161-T1.nii.gz\n",
            "IXI598-HH-2606-T1.nii.gz\n",
            "IXI599-HH-2659-T1.nii.gz\n",
            "IXI600-HH-2660-T1.nii.gz\n",
            "IXI601-HH-2700-T1.nii.gz\n",
            "IXI603-HH-2701-T1.nii.gz\n",
            "IXI605-HH-2598-T1.nii.gz\n",
            "IXI606-HH-2601-T1.nii.gz\n",
            "IXI607-Guys-1097-T1.nii.gz\n",
            "IXI608-HH-2599-T1.nii.gz\n",
            "IXI609-HH-2600-T1.nii.gz\n",
            "IXI610-HH-2649-T1.nii.gz\n",
            "IXI611-HH-2650-T1.nii.gz\n",
            "IXI612-HH-2688-T1.nii.gz\n",
            "IXI613-HH-2734-T1.nii.gz\n",
            "IXI614-HH-2735-T1.nii.gz\n",
            "IXI616-Guys-1092-T1.nii.gz\n",
            "IXI617-Guys-1090-T1.nii.gz\n",
            "IXI618-Guys-1091-T1.nii.gz\n",
            "IXI619-Guys-1099-T1.nii.gz\n",
            "IXI621-Guys-1100-T1.nii.gz\n",
            "IXI622-Guys-1102-T1.nii.gz\n",
            "IXI623-Guys-1076-T1.nii.gz\n",
            "IXI625-Guys-1098-T1.nii.gz\n",
            "IXI626-Guys-1094-T1.nii.gz\n",
            "IXI627-Guys-1103-T1.nii.gz\n",
            "IXI629-Guys-1095-T1.nii.gz\n",
            "IXI630-Guys-1108-T1.nii.gz\n",
            "IXI631-HH-2651-T1.nii.gz\n",
            "IXI632-HH-2652-T1.nii.gz\n",
            "IXI633-HH-2689-T1.nii.gz\n",
            "IXI634-HH-2690-T1.nii.gz\n",
            "IXI635-HH-2691-T1.nii.gz\n",
            "IXI636-HH-2733-T1.nii.gz\n",
            "IXI637-HH-2785-T1.nii.gz\n",
            "IXI638-HH-2786-T1.nii.gz\n",
            "IXI639-Guys-1088-T1.nii.gz\n",
            "IXI640-Guys-1106-T1.nii.gz\n",
            "IXI641-Guys-1105-T1.nii.gz\n",
            "IXI642-Guys-1104-T1.nii.gz\n",
            "IXI643-HH-2787-T1.nii.gz\n",
            "IXI644-Guys-1121-T1.nii.gz\n",
            "IXI646-HH-2653-T1.nii.gz\n",
            "IXI648-Guys-1107-T1.nii.gz\n",
            "IXI651-Guys-1118-T1.nii.gz\n",
            "IXI652-Guys-1116-T1.nii.gz\n",
            "IXI653-Guys-1122-T1.nii.gz\n",
            "IXI661-HH-2788-T1.nii.gz\n",
            "IXI662-Guys-1120-T1.nii.gz\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ChM0l6oJWjrw"
      },
      "source": [
        "!rm 'IXI-T1.tar'"
      ],
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "wxdhaHAOhj3e",
        "outputId": "db16e6cd-c3a6-4155-8db3-1a1bc30559a1"
      },
      "source": [
        "!pip install dltk"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Processing /root/.cache/pip/wheels/9d/72/c6/8d739d38edc23d5f5e58aeaeff23df0236bb2449edccf93898/dltk-0.2.1-py2.py3-none-any.whl\n",
            "Requirement already satisfied: scipy>=0.19.0 in /usr/local/lib/python3.7/dist-packages (from dltk) (1.4.1)\n",
            "Requirement already satisfied: xlrd>=1.1.0 in /usr/local/lib/python3.7/dist-packages (from dltk) (1.1.0)\n",
            "Requirement already satisfied: matplotlib>=1.5.3 in /usr/local/lib/python3.7/dist-packages (from dltk) (3.2.2)\n",
            "Requirement already satisfied: scikit-image>=0.13.0 in /usr/local/lib/python3.7/dist-packages (from dltk) (0.16.2)\n",
            "Requirement already satisfied: pandas>=0.19.0 in /usr/local/lib/python3.7/dist-packages (from dltk) (1.1.5)\n",
            "Requirement already satisfied: jupyter>=1.0.0 in /usr/local/lib/python3.7/dist-packages (from dltk) (1.0.0)\n",
            "Collecting argparse\n",
            "  Using cached https://files.pythonhosted.org/packages/f2/94/3af39d34be01a24a6e65433d19e107099374224905f1e0cc6bbe1fd22a2f/argparse-1.4.0-py2.py3-none-any.whl\n",
            "Collecting SimpleITK>=1.0.0\n",
            "  Using cached https://files.pythonhosted.org/packages/9c/6b/85df5eb3a8059b23a53a9f224476e75473f9bcc0a8583ed1a9c34619f372/SimpleITK-2.0.2-cp37-cp37m-manylinux2010_x86_64.whl\n",
            "Requirement already satisfied: future>=0.16.0 in /usr/local/lib/python3.7/dist-packages (from dltk) (0.16.0)\n",
            "Requirement already satisfied: numpy>=1.14.0 in /usr/local/lib/python3.7/dist-packages (from dltk) (1.19.5)\n",
            "Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.7/dist-packages (from matplotlib>=1.5.3->dltk) (0.10.0)\n",
            "Requirement already satisfied: python-dateutil>=2.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib>=1.5.3->dltk) (2.8.1)\n",
            "Requirement already satisfied: kiwisolver>=1.0.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib>=1.5.3->dltk) (1.3.1)\n",
            "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib>=1.5.3->dltk) (2.4.7)\n",
            "Requirement already satisfied: PyWavelets>=0.4.0 in /usr/local/lib/python3.7/dist-packages (from scikit-image>=0.13.0->dltk) (1.1.1)\n",
            "Requirement already satisfied: pillow>=4.3.0 in /usr/local/lib/python3.7/dist-packages (from scikit-image>=0.13.0->dltk) (7.1.2)\n",
            "Requirement already satisfied: networkx>=2.0 in /usr/local/lib/python3.7/dist-packages (from scikit-image>=0.13.0->dltk) (2.5.1)\n",
            "Requirement already satisfied: imageio>=2.3.0 in /usr/local/lib/python3.7/dist-packages (from scikit-image>=0.13.0->dltk) (2.4.1)\n",
            "Requirement already satisfied: pytz>=2017.2 in /usr/local/lib/python3.7/dist-packages (from pandas>=0.19.0->dltk) (2018.9)\n",
            "Requirement already satisfied: qtconsole in /usr/local/lib/python3.7/dist-packages (from jupyter>=1.0.0->dltk) (5.1.0)\n",
            "Requirement already satisfied: ipywidgets in /usr/local/lib/python3.7/dist-packages (from jupyter>=1.0.0->dltk) (7.6.3)\n",
            "Requirement already satisfied: ipykernel in /usr/local/lib/python3.7/dist-packages (from jupyter>=1.0.0->dltk) (4.10.1)\n",
            "Requirement already satisfied: jupyter-console in /usr/local/lib/python3.7/dist-packages (from jupyter>=1.0.0->dltk) (5.2.0)\n",
            "Requirement already satisfied: nbconvert in /usr/local/lib/python3.7/dist-packages (from jupyter>=1.0.0->dltk) (5.6.1)\n",
            "Requirement already satisfied: notebook in /usr/local/lib/python3.7/dist-packages (from jupyter>=1.0.0->dltk) (5.3.1)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.7/dist-packages (from cycler>=0.10->matplotlib>=1.5.3->dltk) (1.15.0)\n",
            "Requirement already satisfied: decorator<5,>=4.3 in /usr/local/lib/python3.7/dist-packages (from networkx>=2.0->scikit-image>=0.13.0->dltk) (4.4.2)\n",
            "Requirement already satisfied: pyzmq>=17.1 in /usr/local/lib/python3.7/dist-packages (from qtconsole->jupyter>=1.0.0->dltk) (22.0.3)\n",
            "Requirement already satisfied: qtpy in /usr/local/lib/python3.7/dist-packages (from qtconsole->jupyter>=1.0.0->dltk) (1.9.0)\n",
            "Requirement already satisfied: ipython-genutils in /usr/local/lib/python3.7/dist-packages (from qtconsole->jupyter>=1.0.0->dltk) (0.2.0)\n",
            "Requirement already satisfied: pygments in /usr/local/lib/python3.7/dist-packages (from qtconsole->jupyter>=1.0.0->dltk) (2.6.1)\n",
            "Requirement already satisfied: jupyter-client>=4.1 in /usr/local/lib/python3.7/dist-packages (from qtconsole->jupyter>=1.0.0->dltk) (5.3.5)\n",
            "Requirement already satisfied: jupyter-core in /usr/local/lib/python3.7/dist-packages (from qtconsole->jupyter>=1.0.0->dltk) (4.7.1)\n",
            "Requirement already satisfied: traitlets in /usr/local/lib/python3.7/dist-packages (from qtconsole->jupyter>=1.0.0->dltk) (5.0.5)\n",
            "Requirement already satisfied: widgetsnbextension~=3.5.0 in /usr/local/lib/python3.7/dist-packages (from ipywidgets->jupyter>=1.0.0->dltk) (3.5.1)\n",
            "Requirement already satisfied: jupyterlab-widgets>=1.0.0; python_version >= \"3.6\" in /usr/local/lib/python3.7/dist-packages (from ipywidgets->jupyter>=1.0.0->dltk) (1.0.0)\n",
            "Requirement already satisfied: nbformat>=4.2.0 in /usr/local/lib/python3.7/dist-packages (from ipywidgets->jupyter>=1.0.0->dltk) (5.1.3)\n",
            "Requirement already satisfied: ipython>=4.0.0; python_version >= \"3.3\" in /usr/local/lib/python3.7/dist-packages (from ipywidgets->jupyter>=1.0.0->dltk) (5.5.0)\n",
            "Requirement already satisfied: tornado>=4.0 in /usr/local/lib/python3.7/dist-packages (from ipykernel->jupyter>=1.0.0->dltk) (5.1.1)\n",
            "Requirement already satisfied: prompt-toolkit<2.0.0,>=1.0.0 in /usr/local/lib/python3.7/dist-packages (from jupyter-console->jupyter>=1.0.0->dltk) (1.0.18)\n",
            "Requirement already satisfied: pandocfilters>=1.4.1 in /usr/local/lib/python3.7/dist-packages (from nbconvert->jupyter>=1.0.0->dltk) (1.4.3)\n",
            "Requirement already satisfied: jinja2>=2.4 in /usr/local/lib/python3.7/dist-packages (from nbconvert->jupyter>=1.0.0->dltk) (2.11.3)\n",
            "Requirement already satisfied: entrypoints>=0.2.2 in /usr/local/lib/python3.7/dist-packages (from nbconvert->jupyter>=1.0.0->dltk) (0.3)\n",
            "Requirement already satisfied: defusedxml in /usr/local/lib/python3.7/dist-packages (from nbconvert->jupyter>=1.0.0->dltk) (0.7.1)\n",
            "Requirement already satisfied: testpath in /usr/local/lib/python3.7/dist-packages (from nbconvert->jupyter>=1.0.0->dltk) (0.5.0)\n",
            "Requirement already satisfied: bleach in /usr/local/lib/python3.7/dist-packages (from nbconvert->jupyter>=1.0.0->dltk) (3.3.0)\n",
            "Requirement already satisfied: mistune<2,>=0.8.1 in /usr/local/lib/python3.7/dist-packages (from nbconvert->jupyter>=1.0.0->dltk) (0.8.4)\n",
            "Requirement already satisfied: Send2Trash in /usr/local/lib/python3.7/dist-packages (from notebook->jupyter>=1.0.0->dltk) (1.5.0)\n",
            "Requirement already satisfied: terminado>=0.8.1 in /usr/local/lib/python3.7/dist-packages (from notebook->jupyter>=1.0.0->dltk) (0.10.0)\n",
            "Requirement already satisfied: jsonschema!=2.5.0,>=2.4 in /usr/local/lib/python3.7/dist-packages (from nbformat>=4.2.0->ipywidgets->jupyter>=1.0.0->dltk) (2.6.0)\n",
            "Requirement already satisfied: pickleshare in /usr/local/lib/python3.7/dist-packages (from ipython>=4.0.0; python_version >= \"3.3\"->ipywidgets->jupyter>=1.0.0->dltk) (0.7.5)\n",
            "Requirement already satisfied: simplegeneric>0.8 in /usr/local/lib/python3.7/dist-packages (from ipython>=4.0.0; python_version >= \"3.3\"->ipywidgets->jupyter>=1.0.0->dltk) (0.8.1)\n",
            "Requirement already satisfied: setuptools>=18.5 in /usr/local/lib/python3.7/dist-packages (from ipython>=4.0.0; python_version >= \"3.3\"->ipywidgets->jupyter>=1.0.0->dltk) (57.0.0)\n",
            "Requirement already satisfied: pexpect; sys_platform != \"win32\" in /usr/local/lib/python3.7/dist-packages (from ipython>=4.0.0; python_version >= \"3.3\"->ipywidgets->jupyter>=1.0.0->dltk) (4.8.0)\n",
            "Requirement already satisfied: wcwidth in /usr/local/lib/python3.7/dist-packages (from prompt-toolkit<2.0.0,>=1.0.0->jupyter-console->jupyter>=1.0.0->dltk) (0.2.5)\n",
            "Requirement already satisfied: MarkupSafe>=0.23 in /usr/local/lib/python3.7/dist-packages (from jinja2>=2.4->nbconvert->jupyter>=1.0.0->dltk) (2.0.1)\n",
            "Requirement already satisfied: packaging in /usr/local/lib/python3.7/dist-packages (from bleach->nbconvert->jupyter>=1.0.0->dltk) (20.9)\n",
            "Requirement already satisfied: webencodings in /usr/local/lib/python3.7/dist-packages (from bleach->nbconvert->jupyter>=1.0.0->dltk) (0.5.1)\n",
            "Requirement already satisfied: ptyprocess; os_name != \"nt\" in /usr/local/lib/python3.7/dist-packages (from terminado>=0.8.1->notebook->jupyter>=1.0.0->dltk) (0.7.0)\n",
            "Installing collected packages: argparse, SimpleITK, dltk\n",
            "Successfully installed SimpleITK-2.0.2 argparse-1.4.0 dltk-0.2.1\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.colab-display-data+json": {
              "pip_warning": {
                "packages": [
                  "argparse"
                ]
              }
            }
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xWo9N2W6g076",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "4f74b7e3-bd81-4de9-af18-8bfc599539ac"
      },
      "source": [
        "!mkdir 'weights'"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "mkdir: cannot create directory ‘weights’: File exists\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "uJ6jHv2ahgwR",
        "outputId": "9bc661ea-1b7b-4fe9-9f9b-8a25936b9a8c"
      },
      "source": [
        "!git clone 'http://github.com/GalDude33/DenseNetFCN-3D.git'\n",
        "%cd DenseNetFCN-3D\n",
        "!pip install git+https://www.github.com/keras-team/keras-contrib.git\n",
        "from DenseNet3D import DenseNet3DImageNet121, DenseNet3DImageNet169\n",
        "%cd ../\n",
        "!ls"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "fatal: destination path 'DenseNetFCN-3D' already exists and is not an empty directory.\n",
            "/content/DenseNetFCN-3D\n",
            "Collecting git+https://www.github.com/keras-team/keras-contrib.git\n",
            "  Cloning https://www.github.com/keras-team/keras-contrib.git to /tmp/pip-req-build-qyh7kudp\n",
            "  Running command git clone -q https://www.github.com/keras-team/keras-contrib.git /tmp/pip-req-build-qyh7kudp\n",
            "Requirement already satisfied (use --upgrade to upgrade): keras-contrib==2.0.8 from git+https://www.github.com/keras-team/keras-contrib.git in /usr/local/lib/python3.7/dist-packages\n",
            "Requirement already satisfied: keras in /usr/local/lib/python3.7/dist-packages (from keras-contrib==2.0.8) (2.4.3)\n",
            "Requirement already satisfied: numpy>=1.9.1 in /usr/local/lib/python3.7/dist-packages (from keras->keras-contrib==2.0.8) (1.19.5)\n",
            "Requirement already satisfied: scipy>=0.14 in /usr/local/lib/python3.7/dist-packages (from keras->keras-contrib==2.0.8) (1.4.1)\n",
            "Requirement already satisfied: pyyaml in /usr/local/lib/python3.7/dist-packages (from keras->keras-contrib==2.0.8) (3.13)\n",
            "Requirement already satisfied: h5py in /usr/local/lib/python3.7/dist-packages (from keras->keras-contrib==2.0.8) (3.1.0)\n",
            "Requirement already satisfied: cached-property; python_version < \"3.8\" in /usr/local/lib/python3.7/dist-packages (from h5py->keras->keras-contrib==2.0.8) (1.5.2)\n",
            "Building wheels for collected packages: keras-contrib\n",
            "  Building wheel for keras-contrib (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for keras-contrib: filename=keras_contrib-2.0.8-cp37-none-any.whl size=101078 sha256=551267bf3010c25de9687b86803d3b7d7bbbb8466aa4766777e4e54f20bf98e2\n",
            "  Stored in directory: /tmp/pip-ephem-wheel-cache-wo_9pf7j/wheels/11/27/c8/4ed56de7b55f4f61244e2dc6ef3cdbaff2692527a2ce6502ba\n",
            "Successfully built keras-contrib\n",
            "/content\n",
            "dataset  DenseNetFCN-3D  IXI.xls  IXI.xls.1  sample_data  weights\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Kf0l4B66ggXz"
      },
      "source": [
        "# !git clone https://github.com/Moeinh77/keras-resnet3d.git\n",
        "# %cd keras-resnet3d\n",
        "# from resnet3d import Resnet3DBuilder\n",
        "# !python setup.py build\n",
        "# %cd ../\n",
        "# !ls"
      ],
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qdtFRwT2hqVm"
      },
      "source": [
        ""
      ],
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3lkge0LBhZU9"
      },
      "source": [
        "import os\n",
        "import matplotlib.pyplot as plt\n",
        "import glob\n",
        "import h5py\n",
        "import shutil\n",
        "import numpy as np # linear algebra\n",
        "import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n",
        "import seaborn as sns\n",
        "from os import listdir, makedirs, getcwd, remove\n",
        "from os.path import isfile, join, abspath, exists, isdir, expanduser\n",
        "from PIL import Image\n",
        "from pathlib import Path\n",
        "from keras.preprocessing.image import ImageDataGenerator, load_img, img_to_array\n",
        "from keras.optimizers import Adam\n",
        "from keras.callbacks import ModelCheckpoint, Callback, EarlyStopping\n",
        "from sklearn.model_selection import train_test_split\n",
        "from mlxtend.plotting import plot_confusion_matrix\n",
        "from sklearn.metrics import confusion_matrix\n",
        "import cv2\n",
        "import keras\n",
        "import pandas as pd\n",
        "from keras.callbacks import ReduceLROnPlateau\n",
        "from keras.callbacks import EarlyStopping"
      ],
      "execution_count": 21,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ylbVCxz3hZU-",
        "outputId": "c58f321d-4dd9-4a65-f352-c49f496d1991"
      },
      "source": [
        "import numpy as np\n",
        "import sys\n",
        "import os\n",
        "import PIL\n",
        "import PIL.Image as Image\n",
        "import seaborn as sns\n",
        "import warnings\n",
        "import nibabel as nib\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import tensorflow as tf\n",
        "from dltk.io.preprocessing import whitening\n",
        "from dltk.io.augmentation import flip\n",
        "import time\n",
        "print('All imported!')"
      ],
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "All imported!\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 333
        },
        "id": "qL90xbPChZU_",
        "outputId": "8f181195-3c1d-4da0-f092-c1860ccc81f1"
      },
      "source": [
        "img_directory = 'dataset'\n",
        "dataset_info = pd.read_excel('IXI.xls')\n",
        "dataset_info.head()"
      ],
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>IXI_ID</th>\n",
              "      <th>SEX_ID (1=m, 2=f)</th>\n",
              "      <th>HEIGHT</th>\n",
              "      <th>WEIGHT</th>\n",
              "      <th>ETHNIC_ID</th>\n",
              "      <th>MARITAL_ID</th>\n",
              "      <th>OCCUPATION_ID</th>\n",
              "      <th>QUALIFICATION_ID</th>\n",
              "      <th>DOB</th>\n",
              "      <th>DATE_AVAILABLE</th>\n",
              "      <th>STUDY_DATE</th>\n",
              "      <th>AGE</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>170</td>\n",
              "      <td>80</td>\n",
              "      <td>2</td>\n",
              "      <td>3</td>\n",
              "      <td>5</td>\n",
              "      <td>2</td>\n",
              "      <td>1968-02-22</td>\n",
              "      <td>0</td>\n",
              "      <td>NaT</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>2</td>\n",
              "      <td>2</td>\n",
              "      <td>164</td>\n",
              "      <td>58</td>\n",
              "      <td>1</td>\n",
              "      <td>4</td>\n",
              "      <td>1</td>\n",
              "      <td>5</td>\n",
              "      <td>1970-01-30</td>\n",
              "      <td>1</td>\n",
              "      <td>2005-11-18</td>\n",
              "      <td>35.800137</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>12</td>\n",
              "      <td>1</td>\n",
              "      <td>175</td>\n",
              "      <td>70</td>\n",
              "      <td>1</td>\n",
              "      <td>2</td>\n",
              "      <td>1</td>\n",
              "      <td>5</td>\n",
              "      <td>1966-08-20</td>\n",
              "      <td>1</td>\n",
              "      <td>2005-06-01</td>\n",
              "      <td>38.781656</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>13</td>\n",
              "      <td>1</td>\n",
              "      <td>182</td>\n",
              "      <td>70</td>\n",
              "      <td>1</td>\n",
              "      <td>2</td>\n",
              "      <td>1</td>\n",
              "      <td>5</td>\n",
              "      <td>1958-09-15</td>\n",
              "      <td>1</td>\n",
              "      <td>2005-06-01</td>\n",
              "      <td>46.710472</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>14</td>\n",
              "      <td>2</td>\n",
              "      <td>163</td>\n",
              "      <td>65</td>\n",
              "      <td>1</td>\n",
              "      <td>4</td>\n",
              "      <td>1</td>\n",
              "      <td>5</td>\n",
              "      <td>1971-03-15</td>\n",
              "      <td>1</td>\n",
              "      <td>2005-06-09</td>\n",
              "      <td>34.236824</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   IXI_ID  SEX_ID (1=m, 2=f)  HEIGHT  ...  DATE_AVAILABLE  STUDY_DATE        AGE\n",
              "0       1                  1     170  ...               0         NaT        NaN\n",
              "1       2                  2     164  ...               1  2005-11-18  35.800137\n",
              "2      12                  1     175  ...               1  2005-06-01  38.781656\n",
              "3      13                  1     182  ...               1  2005-06-01  46.710472\n",
              "4      14                  2     163  ...               1  2005-06-09  34.236824\n",
              "\n",
              "[5 rows x 12 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 23
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QLD5SYRAhZVB"
      },
      "source": [
        "### Making the dataset generator "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "lQHT8qU9hZVB",
        "outputId": "8acbf1f7-152b-4212-fcef-448e71a6ed13"
      },
      "source": [
        "# reading the MRI images with nibabel and\n",
        "# identifying file names that are not mentioned in the EXCEL file\n",
        "\n",
        "image_files = []\n",
        "labels = []\n",
        "not_included = []\n",
        "for filename in os.listdir(img_directory):\n",
        "    try:\n",
        "        # reading the label from Excel\n",
        "        img_id = int(filename[3:6])\n",
        "        sex = np.array(dataset_info[dataset_info.IXI_ID == img_id])[0][1]\n",
        "        labels.append(sex-1)\n",
        "        # reading images\n",
        "        img = nib.load(img_directory+\"/\"+filename)\n",
        "        image_files.append(img)\n",
        "    except Exception as e:\n",
        "        print(e)\n",
        "        not_included.append(filename)\n",
        "        \n",
        "len(not_included)"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "index 0 is out of bounds for axis 0 with size 0\n",
            "index 0 is out of bounds for axis 0 with size 0\n",
            "index 0 is out of bounds for axis 0 with size 0\n",
            "index 0 is out of bounds for axis 0 with size 0\n",
            "index 0 is out of bounds for axis 0 with size 0\n",
            "index 0 is out of bounds for axis 0 with size 0\n",
            "index 0 is out of bounds for axis 0 with size 0\n",
            "index 0 is out of bounds for axis 0 with size 0\n",
            "index 0 is out of bounds for axis 0 with size 0\n",
            "index 0 is out of bounds for axis 0 with size 0\n",
            "index 0 is out of bounds for axis 0 with size 0\n",
            "index 0 is out of bounds for axis 0 with size 0\n",
            "index 0 is out of bounds for axis 0 with size 0\n",
            "index 0 is out of bounds for axis 0 with size 0\n",
            "index 0 is out of bounds for axis 0 with size 0\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "15"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "sHuYrVfJAM4G",
        "outputId": "ff4e1561-a85e-476f-ee4a-6837ca69cbe1"
      },
      "source": [
        "len(image_files), len(labels)"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(566, 566)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 25
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "q0mHyqNhUUbl",
        "outputId": "2a39329d-c1f6-4ebb-cdcc-0f35ea13e38a"
      },
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "X_train, X_test, y_train, y_test = train_test_split(image_files, labels, test_size=0.2,\n",
        "                                                    random_state =1, stratify = labels )\n",
        "\n",
        "len(X_train), len(X_test)"
      ],
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(452, 114)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 26
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Q6W2RKM7AM4H"
      },
      "source": [
        "def count_class(labels):\n",
        "    zero = 0\n",
        "    one = 0\n",
        "    for l in labels:\n",
        "        if l == 0:\n",
        "             zero += 1\n",
        "        else:\n",
        "            one += 1\n",
        "    return zero, one, zero/ one"
      ],
      "execution_count": 27,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 701
        },
        "id": "XtK3NP2JAM4H",
        "outputId": "3da80302-ad12-480a-ab93-224b9c30fd00"
      },
      "source": [
        "from seaborn import countplot\n",
        "\n",
        "# Plot the results \n",
        "plt.figure(figsize=(4,4))\n",
        "c1 = countplot(y_train)\n",
        "plt.title('Train Data dist', fontsize=14)\n",
        "plt.ylabel('Count', fontsize=12)\n",
        "plt.xlabel('Class', fontsize=12)\n",
        "plt.show()\n",
        "print(count_class(y_train))\n",
        "\n",
        "plt.figure(figsize=(4,4))\n",
        "c2 = countplot(y_test)\n",
        "plt.title('Test Data dist', fontsize=14)\n",
        "plt.ylabel('Count', fontsize=12)\n",
        "plt.xlabel('Class', fontsize=12)\n",
        "plt.show()\n",
        "\n",
        "print(count_class(y_test))"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/seaborn/_decorators.py:43: FutureWarning: Pass the following variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.\n",
            "  FutureWarning\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 288x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/seaborn/_decorators.py:43: FutureWarning: Pass the following variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.\n",
            "  FutureWarning\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "(202, 250, 0.808)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 288x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "(51, 63, 0.8095238095238095)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5Fk_YX0JZmrt",
        "outputId": "5956bdbc-e272-4c3d-c1d5-85c394ca6e86"
      },
      "source": [
        "from sklearn.utils import class_weight\n",
        "weight = class_weight.compute_class_weight('balanced', np.unique(y_train), y_train)\n",
        "weights = {i : weight[i] for i in range(2)}\n",
        "weights"
      ],
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{0: 1.118811881188119, 1: 0.904}"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 29
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4nAwa_lQAM4I"
      },
      "source": [
        "# y_train = np.array(y_train).reshape(-1, 1)\n",
        "# y_test = np.array(y_test).reshape(-1, 1)\n",
        "\n",
        "# y_train.shape, y_test.shape"
      ],
      "execution_count": 30,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AVflavVYhZVC"
      },
      "source": [
        "# # making a list out of the files that are in the EXCEL file\n",
        "# filenames = []\n",
        "# for filename in os.listdir(img_directory):\n",
        "#     if filename not in not_included:\n",
        "#         filenames.append(filename)\n",
        "        \n",
        "# len(filenames), len(not_included)"
      ],
      "execution_count": 31,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZbjHrKhwhZVC"
      },
      "source": [
        "### Defining the generator"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "saECXVAjhZVC"
      },
      "source": [
        "bs = 4"
      ],
      "execution_count": 32,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GEaXmv5HhZVD"
      },
      "source": [
        "def my_data_gen(images, labels, batch_size= 16, method = None,\n",
        "                channels = 150, test_run = False, train_ = True, flip_= False):#, resize= False, resize_scale = 2):\n",
        "    \n",
        "    data = []\n",
        "    labels_in_func = []\n",
        "    i = 0 # for the batch\n",
        "    j = 0 # for selecting the image\n",
        "    \n",
        "    while True:\n",
        "        try:\n",
        "            if j >= len(images):\n",
        "                j = 0\n",
        "            if test_run:\n",
        "                print('j=',j)\n",
        "                print()\n",
        "                # continue\n",
        "            \n",
        "            labels_in_func.append(labels[j])\n",
        "\n",
        "            img = images[j]\n",
        "            j += 1\n",
        "\n",
        "##             rescaling the image, although the problem is the output images does not have the same size\n",
        "#             if resize:\n",
        "#                 img = resample_img(img, target_affine=np.eye(3)*resize_scale, interpolation='nearest')\n",
        "#                 print(img.shape)\n",
        "            \n",
        "            img = np.array(img.dataobj)\n",
        "            dx , dy = img.shape[0], img.shape[1]\n",
        "            \n",
        "            # add zero channels to images with fewer channels than others\n",
        "            if method == 'add_channels':\n",
        "                dz = 150\n",
        "                if img.shape[2] != 150:\n",
        "                    img = np.concatenate((img, np.zeros((img.shape[0],img.shape[1],150-img.shape[2]))), axis =2)\n",
        "            elif method == 'del_channels':\n",
        "                dz = channels\n",
        "                if dz%2 == 0:\n",
        "                    img = img[:,:,75 - (channels//2) :75 + (channels//2)]\n",
        "                else:\n",
        "                    print(74 - (channels//2) +1 ,76 + (channels//2) )\n",
        "                    img = img[:,:,74 - (channels//2) +1 :76 + (channels//2)]\n",
        "\n",
        "            \n",
        "            img = whitening(img)\n",
        "\n",
        "            if train_ and flip_:\n",
        "                img = flip(img, axis=1)\n",
        "                \n",
        "#             img_id = int(filename[3:6])\n",
        "#             sex = np.array(dataset_info[dataset_info.IXI_ID == img_id])[0][1]\n",
        "            \n",
        "            data.append(img)\n",
        "            \n",
        "            i += 1\n",
        "            if i == batch_size:\n",
        "                i = 0\n",
        "                yield np.array(data).reshape(batch_size, dx, dy, dz, 1), np.array(labels_in_func).reshape(-1, 1)\n",
        "                data = []\n",
        "                labels_in_func = []\n",
        "                \n",
        "        except Exception as e:\n",
        "            print(filename)\n",
        "            print(e)"
      ],
      "execution_count": 33,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WT1waw4chZVE"
      },
      "source": [
        "method = 'del_channels'; in_dim = 80\n",
        "# channels = 'add_channels'; in_dim = 150\n",
        "data_gen_train = my_data_gen(X_train, y_train, batch_size = bs, method =method, channels=in_dim, train_ = True)\n",
        "data_gen_test = my_data_gen(X_test, y_test, batch_size = bs, method =method, channels=in_dim, train_ =  False)"
      ],
      "execution_count": 34,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AeE7Nh-SKksA"
      },
      "source": [
        "# for i in my_data_gen(X_test, y_test, batch_size = bs, method ='add_channels', test_run= True):\n",
        "#     print(i[0].shape, i[1].shape)"
      ],
      "execution_count": 35,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "t-TM09D5hZVE"
      },
      "source": [
        "# i1, l1 = next(data_gen_train)\n",
        "# i1.shape, l1.shape"
      ],
      "execution_count": 36,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nLAQNLIWE5aY"
      },
      "source": [
        "# i2, l2 = next(data_gen_test)\n",
        "# i2.shape, l2.shape"
      ],
      "execution_count": 37,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "AcWsR2wXFFfo"
      },
      "source": [
        "### Defining the model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qGToRb2LmScc"
      },
      "source": [
        "in_size = 256"
      ],
      "execution_count": 38,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HLIVYn7ehZVF"
      },
      "source": [
        "# model = Resnet3DBuilder.build_resnet_18((in_size, in_size, in_dim, 1), num_outputs = 1, reg_factor= 0.0001)\n",
        "\n",
        "# opt = Adam(3e-3) \n",
        "# model.compile(loss= 'binary_crossentropy', metrics=['accuracy'], optimizer=opt)\n",
        "\n",
        "# model.summary()"
      ],
      "execution_count": 39,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "AbwVgi--mScd",
        "outputId": "1b60e3fb-e202-4cc6-87d2-5d1151b32ae8"
      },
      "source": [
        "model = DenseNet3DImageNet121(input_shape=(in_size, in_size, in_dim, 1),\n",
        "                          bottleneck=True,\n",
        "                          reduction=0.5,\n",
        "                          dropout_rate=0.,\n",
        "                          weight_decay=0.,\n",
        "                          include_top=True, # since i have no layer after this model\n",
        "                          input_tensor=None,\n",
        "                          pooling='avg',\n",
        "                          classes=1,\n",
        "                          activation='sigmoid')\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": 40,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"densenet\"\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_1 (InputLayer)            [(None, 256, 256, 80 0                                            \n",
            "__________________________________________________________________________________________________\n",
            "initial_Conv3D (Conv3D)         (None, 128, 128, 40, 21952       input_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "initial_bn (BatchNormalization) (None, 128, 128, 40, 256         initial_Conv3D[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation (Activation)         (None, 128, 128, 40, 0           initial_bn[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "max_pooling3d (MaxPooling3D)    (None, 64, 64, 20, 6 0           activation[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_0_bn (BatchNormalizatio (None, 64, 64, 20, 6 256         max_pooling3d[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "activation_1 (Activation)       (None, 64, 64, 20, 6 0           dense_0_0_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_0_bottleneck_Conv3D (Co (None, 64, 64, 20, 1 8192        activation_1[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_0_bottleneck_bn (BatchN (None, 64, 64, 20, 1 512         dense_0_0_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_2 (Activation)       (None, 64, 64, 20, 1 0           dense_0_0_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_0_Conv3D (Conv3D)       (None, 64, 64, 20, 3 110592      activation_2[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "concatenate (Concatenate)       (None, 64, 64, 20, 9 0           max_pooling3d[0][0]              \n",
            "                                                                 dense_0_0_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_1_bn (BatchNormalizatio (None, 64, 64, 20, 9 384         concatenate[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "activation_3 (Activation)       (None, 64, 64, 20, 9 0           dense_0_1_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_1_bottleneck_Conv3D (Co (None, 64, 64, 20, 1 12288       activation_3[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_1_bottleneck_bn (BatchN (None, 64, 64, 20, 1 512         dense_0_1_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_4 (Activation)       (None, 64, 64, 20, 1 0           dense_0_1_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_1_Conv3D (Conv3D)       (None, 64, 64, 20, 3 110592      activation_4[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_1 (Concatenate)     (None, 64, 64, 20, 1 0           concatenate[0][0]                \n",
            "                                                                 dense_0_1_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_2_bn (BatchNormalizatio (None, 64, 64, 20, 1 512         concatenate_1[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "activation_5 (Activation)       (None, 64, 64, 20, 1 0           dense_0_2_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_2_bottleneck_Conv3D (Co (None, 64, 64, 20, 1 16384       activation_5[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_2_bottleneck_bn (BatchN (None, 64, 64, 20, 1 512         dense_0_2_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_6 (Activation)       (None, 64, 64, 20, 1 0           dense_0_2_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_2_Conv3D (Conv3D)       (None, 64, 64, 20, 3 110592      activation_6[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_2 (Concatenate)     (None, 64, 64, 20, 1 0           concatenate_1[0][0]              \n",
            "                                                                 dense_0_2_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_3_bn (BatchNormalizatio (None, 64, 64, 20, 1 640         concatenate_2[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "activation_7 (Activation)       (None, 64, 64, 20, 1 0           dense_0_3_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_3_bottleneck_Conv3D (Co (None, 64, 64, 20, 1 20480       activation_7[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_3_bottleneck_bn (BatchN (None, 64, 64, 20, 1 512         dense_0_3_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_8 (Activation)       (None, 64, 64, 20, 1 0           dense_0_3_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_3_Conv3D (Conv3D)       (None, 64, 64, 20, 3 110592      activation_8[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_3 (Concatenate)     (None, 64, 64, 20, 1 0           concatenate_2[0][0]              \n",
            "                                                                 dense_0_3_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_4_bn (BatchNormalizatio (None, 64, 64, 20, 1 768         concatenate_3[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "activation_9 (Activation)       (None, 64, 64, 20, 1 0           dense_0_4_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_4_bottleneck_Conv3D (Co (None, 64, 64, 20, 1 24576       activation_9[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_4_bottleneck_bn (BatchN (None, 64, 64, 20, 1 512         dense_0_4_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_10 (Activation)      (None, 64, 64, 20, 1 0           dense_0_4_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_4_Conv3D (Conv3D)       (None, 64, 64, 20, 3 110592      activation_10[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_4 (Concatenate)     (None, 64, 64, 20, 2 0           concatenate_3[0][0]              \n",
            "                                                                 dense_0_4_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_5_bn (BatchNormalizatio (None, 64, 64, 20, 2 896         concatenate_4[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "activation_11 (Activation)      (None, 64, 64, 20, 2 0           dense_0_5_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_5_bottleneck_Conv3D (Co (None, 64, 64, 20, 1 28672       activation_11[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_5_bottleneck_bn (BatchN (None, 64, 64, 20, 1 512         dense_0_5_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_12 (Activation)      (None, 64, 64, 20, 1 0           dense_0_5_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_0_5_Conv3D (Conv3D)       (None, 64, 64, 20, 3 110592      activation_12[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_5 (Concatenate)     (None, 64, 64, 20, 2 0           concatenate_4[0][0]              \n",
            "                                                                 dense_0_5_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "tr_0_bn (BatchNormalization)    (None, 64, 64, 20, 2 1024        concatenate_5[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "activation_13 (Activation)      (None, 64, 64, 20, 2 0           tr_0_bn[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "tr_0_Conv3D (Conv3D)            (None, 64, 64, 20, 1 32768       activation_13[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "average_pooling3d (AveragePooli (None, 32, 32, 10, 1 0           tr_0_Conv3D[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_0_bn (BatchNormalizatio (None, 32, 32, 10, 1 512         average_pooling3d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "activation_14 (Activation)      (None, 32, 32, 10, 1 0           dense_1_0_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_0_bottleneck_Conv3D (Co (None, 32, 32, 10, 1 16384       activation_14[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_0_bottleneck_bn (BatchN (None, 32, 32, 10, 1 512         dense_1_0_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_15 (Activation)      (None, 32, 32, 10, 1 0           dense_1_0_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_0_Conv3D (Conv3D)       (None, 32, 32, 10, 3 110592      activation_15[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_6 (Concatenate)     (None, 32, 32, 10, 1 0           average_pooling3d[0][0]          \n",
            "                                                                 dense_1_0_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_1_bn (BatchNormalizatio (None, 32, 32, 10, 1 640         concatenate_6[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "activation_16 (Activation)      (None, 32, 32, 10, 1 0           dense_1_1_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_1_bottleneck_Conv3D (Co (None, 32, 32, 10, 1 20480       activation_16[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_1_bottleneck_bn (BatchN (None, 32, 32, 10, 1 512         dense_1_1_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_17 (Activation)      (None, 32, 32, 10, 1 0           dense_1_1_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_1_Conv3D (Conv3D)       (None, 32, 32, 10, 3 110592      activation_17[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_7 (Concatenate)     (None, 32, 32, 10, 1 0           concatenate_6[0][0]              \n",
            "                                                                 dense_1_1_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_2_bn (BatchNormalizatio (None, 32, 32, 10, 1 768         concatenate_7[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "activation_18 (Activation)      (None, 32, 32, 10, 1 0           dense_1_2_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_2_bottleneck_Conv3D (Co (None, 32, 32, 10, 1 24576       activation_18[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_2_bottleneck_bn (BatchN (None, 32, 32, 10, 1 512         dense_1_2_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_19 (Activation)      (None, 32, 32, 10, 1 0           dense_1_2_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_2_Conv3D (Conv3D)       (None, 32, 32, 10, 3 110592      activation_19[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_8 (Concatenate)     (None, 32, 32, 10, 2 0           concatenate_7[0][0]              \n",
            "                                                                 dense_1_2_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_3_bn (BatchNormalizatio (None, 32, 32, 10, 2 896         concatenate_8[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "activation_20 (Activation)      (None, 32, 32, 10, 2 0           dense_1_3_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_3_bottleneck_Conv3D (Co (None, 32, 32, 10, 1 28672       activation_20[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_3_bottleneck_bn (BatchN (None, 32, 32, 10, 1 512         dense_1_3_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_21 (Activation)      (None, 32, 32, 10, 1 0           dense_1_3_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_3_Conv3D (Conv3D)       (None, 32, 32, 10, 3 110592      activation_21[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_9 (Concatenate)     (None, 32, 32, 10, 2 0           concatenate_8[0][0]              \n",
            "                                                                 dense_1_3_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_4_bn (BatchNormalizatio (None, 32, 32, 10, 2 1024        concatenate_9[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "activation_22 (Activation)      (None, 32, 32, 10, 2 0           dense_1_4_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_4_bottleneck_Conv3D (Co (None, 32, 32, 10, 1 32768       activation_22[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_4_bottleneck_bn (BatchN (None, 32, 32, 10, 1 512         dense_1_4_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_23 (Activation)      (None, 32, 32, 10, 1 0           dense_1_4_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_4_Conv3D (Conv3D)       (None, 32, 32, 10, 3 110592      activation_23[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_10 (Concatenate)    (None, 32, 32, 10, 2 0           concatenate_9[0][0]              \n",
            "                                                                 dense_1_4_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_5_bn (BatchNormalizatio (None, 32, 32, 10, 2 1152        concatenate_10[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_24 (Activation)      (None, 32, 32, 10, 2 0           dense_1_5_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_5_bottleneck_Conv3D (Co (None, 32, 32, 10, 1 36864       activation_24[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_5_bottleneck_bn (BatchN (None, 32, 32, 10, 1 512         dense_1_5_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_25 (Activation)      (None, 32, 32, 10, 1 0           dense_1_5_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_5_Conv3D (Conv3D)       (None, 32, 32, 10, 3 110592      activation_25[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_11 (Concatenate)    (None, 32, 32, 10, 3 0           concatenate_10[0][0]             \n",
            "                                                                 dense_1_5_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_6_bn (BatchNormalizatio (None, 32, 32, 10, 3 1280        concatenate_11[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_26 (Activation)      (None, 32, 32, 10, 3 0           dense_1_6_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_6_bottleneck_Conv3D (Co (None, 32, 32, 10, 1 40960       activation_26[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_6_bottleneck_bn (BatchN (None, 32, 32, 10, 1 512         dense_1_6_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_27 (Activation)      (None, 32, 32, 10, 1 0           dense_1_6_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_6_Conv3D (Conv3D)       (None, 32, 32, 10, 3 110592      activation_27[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_12 (Concatenate)    (None, 32, 32, 10, 3 0           concatenate_11[0][0]             \n",
            "                                                                 dense_1_6_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_7_bn (BatchNormalizatio (None, 32, 32, 10, 3 1408        concatenate_12[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_28 (Activation)      (None, 32, 32, 10, 3 0           dense_1_7_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_7_bottleneck_Conv3D (Co (None, 32, 32, 10, 1 45056       activation_28[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_7_bottleneck_bn (BatchN (None, 32, 32, 10, 1 512         dense_1_7_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_29 (Activation)      (None, 32, 32, 10, 1 0           dense_1_7_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_7_Conv3D (Conv3D)       (None, 32, 32, 10, 3 110592      activation_29[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_13 (Concatenate)    (None, 32, 32, 10, 3 0           concatenate_12[0][0]             \n",
            "                                                                 dense_1_7_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_8_bn (BatchNormalizatio (None, 32, 32, 10, 3 1536        concatenate_13[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_30 (Activation)      (None, 32, 32, 10, 3 0           dense_1_8_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_8_bottleneck_Conv3D (Co (None, 32, 32, 10, 1 49152       activation_30[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_8_bottleneck_bn (BatchN (None, 32, 32, 10, 1 512         dense_1_8_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_31 (Activation)      (None, 32, 32, 10, 1 0           dense_1_8_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_8_Conv3D (Conv3D)       (None, 32, 32, 10, 3 110592      activation_31[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_14 (Concatenate)    (None, 32, 32, 10, 4 0           concatenate_13[0][0]             \n",
            "                                                                 dense_1_8_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_9_bn (BatchNormalizatio (None, 32, 32, 10, 4 1664        concatenate_14[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_32 (Activation)      (None, 32, 32, 10, 4 0           dense_1_9_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_9_bottleneck_Conv3D (Co (None, 32, 32, 10, 1 53248       activation_32[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_9_bottleneck_bn (BatchN (None, 32, 32, 10, 1 512         dense_1_9_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_33 (Activation)      (None, 32, 32, 10, 1 0           dense_1_9_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_9_Conv3D (Conv3D)       (None, 32, 32, 10, 3 110592      activation_33[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_15 (Concatenate)    (None, 32, 32, 10, 4 0           concatenate_14[0][0]             \n",
            "                                                                 dense_1_9_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_10_bn (BatchNormalizati (None, 32, 32, 10, 4 1792        concatenate_15[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_34 (Activation)      (None, 32, 32, 10, 4 0           dense_1_10_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_10_bottleneck_Conv3D (C (None, 32, 32, 10, 1 57344       activation_34[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_10_bottleneck_bn (Batch (None, 32, 32, 10, 1 512         dense_1_10_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_35 (Activation)      (None, 32, 32, 10, 1 0           dense_1_10_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_10_Conv3D (Conv3D)      (None, 32, 32, 10, 3 110592      activation_35[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_16 (Concatenate)    (None, 32, 32, 10, 4 0           concatenate_15[0][0]             \n",
            "                                                                 dense_1_10_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_11_bn (BatchNormalizati (None, 32, 32, 10, 4 1920        concatenate_16[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_36 (Activation)      (None, 32, 32, 10, 4 0           dense_1_11_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_11_bottleneck_Conv3D (C (None, 32, 32, 10, 1 61440       activation_36[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_11_bottleneck_bn (Batch (None, 32, 32, 10, 1 512         dense_1_11_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_37 (Activation)      (None, 32, 32, 10, 1 0           dense_1_11_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_1_11_Conv3D (Conv3D)      (None, 32, 32, 10, 3 110592      activation_37[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_17 (Concatenate)    (None, 32, 32, 10, 5 0           concatenate_16[0][0]             \n",
            "                                                                 dense_1_11_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "tr_1_bn (BatchNormalization)    (None, 32, 32, 10, 5 2048        concatenate_17[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_38 (Activation)      (None, 32, 32, 10, 5 0           tr_1_bn[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "tr_1_Conv3D (Conv3D)            (None, 32, 32, 10, 2 131072      activation_38[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "average_pooling3d_1 (AveragePoo (None, 16, 16, 5, 25 0           tr_1_Conv3D[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_0_bn (BatchNormalizatio (None, 16, 16, 5, 25 1024        average_pooling3d_1[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "activation_39 (Activation)      (None, 16, 16, 5, 25 0           dense_2_0_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_0_bottleneck_Conv3D (Co (None, 16, 16, 5, 12 32768       activation_39[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_0_bottleneck_bn (BatchN (None, 16, 16, 5, 12 512         dense_2_0_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_40 (Activation)      (None, 16, 16, 5, 12 0           dense_2_0_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_0_Conv3D (Conv3D)       (None, 16, 16, 5, 32 110592      activation_40[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_18 (Concatenate)    (None, 16, 16, 5, 28 0           average_pooling3d_1[0][0]        \n",
            "                                                                 dense_2_0_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_1_bn (BatchNormalizatio (None, 16, 16, 5, 28 1152        concatenate_18[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_41 (Activation)      (None, 16, 16, 5, 28 0           dense_2_1_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_1_bottleneck_Conv3D (Co (None, 16, 16, 5, 12 36864       activation_41[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_1_bottleneck_bn (BatchN (None, 16, 16, 5, 12 512         dense_2_1_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_42 (Activation)      (None, 16, 16, 5, 12 0           dense_2_1_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_1_Conv3D (Conv3D)       (None, 16, 16, 5, 32 110592      activation_42[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_19 (Concatenate)    (None, 16, 16, 5, 32 0           concatenate_18[0][0]             \n",
            "                                                                 dense_2_1_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_2_bn (BatchNormalizatio (None, 16, 16, 5, 32 1280        concatenate_19[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_43 (Activation)      (None, 16, 16, 5, 32 0           dense_2_2_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_2_bottleneck_Conv3D (Co (None, 16, 16, 5, 12 40960       activation_43[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_2_bottleneck_bn (BatchN (None, 16, 16, 5, 12 512         dense_2_2_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_44 (Activation)      (None, 16, 16, 5, 12 0           dense_2_2_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_2_Conv3D (Conv3D)       (None, 16, 16, 5, 32 110592      activation_44[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_20 (Concatenate)    (None, 16, 16, 5, 35 0           concatenate_19[0][0]             \n",
            "                                                                 dense_2_2_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_3_bn (BatchNormalizatio (None, 16, 16, 5, 35 1408        concatenate_20[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_45 (Activation)      (None, 16, 16, 5, 35 0           dense_2_3_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_3_bottleneck_Conv3D (Co (None, 16, 16, 5, 12 45056       activation_45[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_3_bottleneck_bn (BatchN (None, 16, 16, 5, 12 512         dense_2_3_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_46 (Activation)      (None, 16, 16, 5, 12 0           dense_2_3_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_3_Conv3D (Conv3D)       (None, 16, 16, 5, 32 110592      activation_46[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_21 (Concatenate)    (None, 16, 16, 5, 38 0           concatenate_20[0][0]             \n",
            "                                                                 dense_2_3_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_4_bn (BatchNormalizatio (None, 16, 16, 5, 38 1536        concatenate_21[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_47 (Activation)      (None, 16, 16, 5, 38 0           dense_2_4_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_4_bottleneck_Conv3D (Co (None, 16, 16, 5, 12 49152       activation_47[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_4_bottleneck_bn (BatchN (None, 16, 16, 5, 12 512         dense_2_4_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_48 (Activation)      (None, 16, 16, 5, 12 0           dense_2_4_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_4_Conv3D (Conv3D)       (None, 16, 16, 5, 32 110592      activation_48[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_22 (Concatenate)    (None, 16, 16, 5, 41 0           concatenate_21[0][0]             \n",
            "                                                                 dense_2_4_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_5_bn (BatchNormalizatio (None, 16, 16, 5, 41 1664        concatenate_22[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_49 (Activation)      (None, 16, 16, 5, 41 0           dense_2_5_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_5_bottleneck_Conv3D (Co (None, 16, 16, 5, 12 53248       activation_49[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_5_bottleneck_bn (BatchN (None, 16, 16, 5, 12 512         dense_2_5_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_50 (Activation)      (None, 16, 16, 5, 12 0           dense_2_5_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_5_Conv3D (Conv3D)       (None, 16, 16, 5, 32 110592      activation_50[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_23 (Concatenate)    (None, 16, 16, 5, 44 0           concatenate_22[0][0]             \n",
            "                                                                 dense_2_5_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_6_bn (BatchNormalizatio (None, 16, 16, 5, 44 1792        concatenate_23[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_51 (Activation)      (None, 16, 16, 5, 44 0           dense_2_6_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_6_bottleneck_Conv3D (Co (None, 16, 16, 5, 12 57344       activation_51[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_6_bottleneck_bn (BatchN (None, 16, 16, 5, 12 512         dense_2_6_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_52 (Activation)      (None, 16, 16, 5, 12 0           dense_2_6_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_6_Conv3D (Conv3D)       (None, 16, 16, 5, 32 110592      activation_52[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_24 (Concatenate)    (None, 16, 16, 5, 48 0           concatenate_23[0][0]             \n",
            "                                                                 dense_2_6_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_7_bn (BatchNormalizatio (None, 16, 16, 5, 48 1920        concatenate_24[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_53 (Activation)      (None, 16, 16, 5, 48 0           dense_2_7_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_7_bottleneck_Conv3D (Co (None, 16, 16, 5, 12 61440       activation_53[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_7_bottleneck_bn (BatchN (None, 16, 16, 5, 12 512         dense_2_7_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_54 (Activation)      (None, 16, 16, 5, 12 0           dense_2_7_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_7_Conv3D (Conv3D)       (None, 16, 16, 5, 32 110592      activation_54[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_25 (Concatenate)    (None, 16, 16, 5, 51 0           concatenate_24[0][0]             \n",
            "                                                                 dense_2_7_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_8_bn (BatchNormalizatio (None, 16, 16, 5, 51 2048        concatenate_25[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_55 (Activation)      (None, 16, 16, 5, 51 0           dense_2_8_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_8_bottleneck_Conv3D (Co (None, 16, 16, 5, 12 65536       activation_55[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_8_bottleneck_bn (BatchN (None, 16, 16, 5, 12 512         dense_2_8_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_56 (Activation)      (None, 16, 16, 5, 12 0           dense_2_8_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_8_Conv3D (Conv3D)       (None, 16, 16, 5, 32 110592      activation_56[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_26 (Concatenate)    (None, 16, 16, 5, 54 0           concatenate_25[0][0]             \n",
            "                                                                 dense_2_8_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_9_bn (BatchNormalizatio (None, 16, 16, 5, 54 2176        concatenate_26[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_57 (Activation)      (None, 16, 16, 5, 54 0           dense_2_9_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_9_bottleneck_Conv3D (Co (None, 16, 16, 5, 12 69632       activation_57[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_9_bottleneck_bn (BatchN (None, 16, 16, 5, 12 512         dense_2_9_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_58 (Activation)      (None, 16, 16, 5, 12 0           dense_2_9_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_9_Conv3D (Conv3D)       (None, 16, 16, 5, 32 110592      activation_58[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_27 (Concatenate)    (None, 16, 16, 5, 57 0           concatenate_26[0][0]             \n",
            "                                                                 dense_2_9_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_10_bn (BatchNormalizati (None, 16, 16, 5, 57 2304        concatenate_27[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_59 (Activation)      (None, 16, 16, 5, 57 0           dense_2_10_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_10_bottleneck_Conv3D (C (None, 16, 16, 5, 12 73728       activation_59[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_10_bottleneck_bn (Batch (None, 16, 16, 5, 12 512         dense_2_10_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_60 (Activation)      (None, 16, 16, 5, 12 0           dense_2_10_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_10_Conv3D (Conv3D)      (None, 16, 16, 5, 32 110592      activation_60[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_28 (Concatenate)    (None, 16, 16, 5, 60 0           concatenate_27[0][0]             \n",
            "                                                                 dense_2_10_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_11_bn (BatchNormalizati (None, 16, 16, 5, 60 2432        concatenate_28[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_61 (Activation)      (None, 16, 16, 5, 60 0           dense_2_11_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_11_bottleneck_Conv3D (C (None, 16, 16, 5, 12 77824       activation_61[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_11_bottleneck_bn (Batch (None, 16, 16, 5, 12 512         dense_2_11_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_62 (Activation)      (None, 16, 16, 5, 12 0           dense_2_11_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_11_Conv3D (Conv3D)      (None, 16, 16, 5, 32 110592      activation_62[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_29 (Concatenate)    (None, 16, 16, 5, 64 0           concatenate_28[0][0]             \n",
            "                                                                 dense_2_11_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_12_bn (BatchNormalizati (None, 16, 16, 5, 64 2560        concatenate_29[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_63 (Activation)      (None, 16, 16, 5, 64 0           dense_2_12_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_12_bottleneck_Conv3D (C (None, 16, 16, 5, 12 81920       activation_63[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_12_bottleneck_bn (Batch (None, 16, 16, 5, 12 512         dense_2_12_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_64 (Activation)      (None, 16, 16, 5, 12 0           dense_2_12_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_12_Conv3D (Conv3D)      (None, 16, 16, 5, 32 110592      activation_64[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_30 (Concatenate)    (None, 16, 16, 5, 67 0           concatenate_29[0][0]             \n",
            "                                                                 dense_2_12_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_13_bn (BatchNormalizati (None, 16, 16, 5, 67 2688        concatenate_30[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_65 (Activation)      (None, 16, 16, 5, 67 0           dense_2_13_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_13_bottleneck_Conv3D (C (None, 16, 16, 5, 12 86016       activation_65[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_13_bottleneck_bn (Batch (None, 16, 16, 5, 12 512         dense_2_13_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_66 (Activation)      (None, 16, 16, 5, 12 0           dense_2_13_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_13_Conv3D (Conv3D)      (None, 16, 16, 5, 32 110592      activation_66[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_31 (Concatenate)    (None, 16, 16, 5, 70 0           concatenate_30[0][0]             \n",
            "                                                                 dense_2_13_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_14_bn (BatchNormalizati (None, 16, 16, 5, 70 2816        concatenate_31[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_67 (Activation)      (None, 16, 16, 5, 70 0           dense_2_14_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_14_bottleneck_Conv3D (C (None, 16, 16, 5, 12 90112       activation_67[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_14_bottleneck_bn (Batch (None, 16, 16, 5, 12 512         dense_2_14_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_68 (Activation)      (None, 16, 16, 5, 12 0           dense_2_14_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_14_Conv3D (Conv3D)      (None, 16, 16, 5, 32 110592      activation_68[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_32 (Concatenate)    (None, 16, 16, 5, 73 0           concatenate_31[0][0]             \n",
            "                                                                 dense_2_14_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_15_bn (BatchNormalizati (None, 16, 16, 5, 73 2944        concatenate_32[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_69 (Activation)      (None, 16, 16, 5, 73 0           dense_2_15_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_15_bottleneck_Conv3D (C (None, 16, 16, 5, 12 94208       activation_69[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_15_bottleneck_bn (Batch (None, 16, 16, 5, 12 512         dense_2_15_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_70 (Activation)      (None, 16, 16, 5, 12 0           dense_2_15_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_15_Conv3D (Conv3D)      (None, 16, 16, 5, 32 110592      activation_70[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_33 (Concatenate)    (None, 16, 16, 5, 76 0           concatenate_32[0][0]             \n",
            "                                                                 dense_2_15_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_16_bn (BatchNormalizati (None, 16, 16, 5, 76 3072        concatenate_33[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_71 (Activation)      (None, 16, 16, 5, 76 0           dense_2_16_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_16_bottleneck_Conv3D (C (None, 16, 16, 5, 12 98304       activation_71[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_16_bottleneck_bn (Batch (None, 16, 16, 5, 12 512         dense_2_16_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_72 (Activation)      (None, 16, 16, 5, 12 0           dense_2_16_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_16_Conv3D (Conv3D)      (None, 16, 16, 5, 32 110592      activation_72[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_34 (Concatenate)    (None, 16, 16, 5, 80 0           concatenate_33[0][0]             \n",
            "                                                                 dense_2_16_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_17_bn (BatchNormalizati (None, 16, 16, 5, 80 3200        concatenate_34[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_73 (Activation)      (None, 16, 16, 5, 80 0           dense_2_17_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_17_bottleneck_Conv3D (C (None, 16, 16, 5, 12 102400      activation_73[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_17_bottleneck_bn (Batch (None, 16, 16, 5, 12 512         dense_2_17_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_74 (Activation)      (None, 16, 16, 5, 12 0           dense_2_17_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_17_Conv3D (Conv3D)      (None, 16, 16, 5, 32 110592      activation_74[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_35 (Concatenate)    (None, 16, 16, 5, 83 0           concatenate_34[0][0]             \n",
            "                                                                 dense_2_17_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_18_bn (BatchNormalizati (None, 16, 16, 5, 83 3328        concatenate_35[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_75 (Activation)      (None, 16, 16, 5, 83 0           dense_2_18_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_18_bottleneck_Conv3D (C (None, 16, 16, 5, 12 106496      activation_75[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_18_bottleneck_bn (Batch (None, 16, 16, 5, 12 512         dense_2_18_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_76 (Activation)      (None, 16, 16, 5, 12 0           dense_2_18_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_18_Conv3D (Conv3D)      (None, 16, 16, 5, 32 110592      activation_76[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_36 (Concatenate)    (None, 16, 16, 5, 86 0           concatenate_35[0][0]             \n",
            "                                                                 dense_2_18_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_19_bn (BatchNormalizati (None, 16, 16, 5, 86 3456        concatenate_36[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_77 (Activation)      (None, 16, 16, 5, 86 0           dense_2_19_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_19_bottleneck_Conv3D (C (None, 16, 16, 5, 12 110592      activation_77[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_19_bottleneck_bn (Batch (None, 16, 16, 5, 12 512         dense_2_19_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_78 (Activation)      (None, 16, 16, 5, 12 0           dense_2_19_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_19_Conv3D (Conv3D)      (None, 16, 16, 5, 32 110592      activation_78[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_37 (Concatenate)    (None, 16, 16, 5, 89 0           concatenate_36[0][0]             \n",
            "                                                                 dense_2_19_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_20_bn (BatchNormalizati (None, 16, 16, 5, 89 3584        concatenate_37[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_79 (Activation)      (None, 16, 16, 5, 89 0           dense_2_20_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_20_bottleneck_Conv3D (C (None, 16, 16, 5, 12 114688      activation_79[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_20_bottleneck_bn (Batch (None, 16, 16, 5, 12 512         dense_2_20_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_80 (Activation)      (None, 16, 16, 5, 12 0           dense_2_20_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_20_Conv3D (Conv3D)      (None, 16, 16, 5, 32 110592      activation_80[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_38 (Concatenate)    (None, 16, 16, 5, 92 0           concatenate_37[0][0]             \n",
            "                                                                 dense_2_20_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_21_bn (BatchNormalizati (None, 16, 16, 5, 92 3712        concatenate_38[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_81 (Activation)      (None, 16, 16, 5, 92 0           dense_2_21_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_21_bottleneck_Conv3D (C (None, 16, 16, 5, 12 118784      activation_81[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_21_bottleneck_bn (Batch (None, 16, 16, 5, 12 512         dense_2_21_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_82 (Activation)      (None, 16, 16, 5, 12 0           dense_2_21_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_21_Conv3D (Conv3D)      (None, 16, 16, 5, 32 110592      activation_82[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_39 (Concatenate)    (None, 16, 16, 5, 96 0           concatenate_38[0][0]             \n",
            "                                                                 dense_2_21_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_22_bn (BatchNormalizati (None, 16, 16, 5, 96 3840        concatenate_39[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_83 (Activation)      (None, 16, 16, 5, 96 0           dense_2_22_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_22_bottleneck_Conv3D (C (None, 16, 16, 5, 12 122880      activation_83[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_22_bottleneck_bn (Batch (None, 16, 16, 5, 12 512         dense_2_22_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_84 (Activation)      (None, 16, 16, 5, 12 0           dense_2_22_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_22_Conv3D (Conv3D)      (None, 16, 16, 5, 32 110592      activation_84[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_40 (Concatenate)    (None, 16, 16, 5, 99 0           concatenate_39[0][0]             \n",
            "                                                                 dense_2_22_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_23_bn (BatchNormalizati (None, 16, 16, 5, 99 3968        concatenate_40[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_85 (Activation)      (None, 16, 16, 5, 99 0           dense_2_23_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_23_bottleneck_Conv3D (C (None, 16, 16, 5, 12 126976      activation_85[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_23_bottleneck_bn (Batch (None, 16, 16, 5, 12 512         dense_2_23_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_86 (Activation)      (None, 16, 16, 5, 12 0           dense_2_23_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_2_23_Conv3D (Conv3D)      (None, 16, 16, 5, 32 110592      activation_86[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_41 (Concatenate)    (None, 16, 16, 5, 10 0           concatenate_40[0][0]             \n",
            "                                                                 dense_2_23_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "tr_2_bn (BatchNormalization)    (None, 16, 16, 5, 10 4096        concatenate_41[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_87 (Activation)      (None, 16, 16, 5, 10 0           tr_2_bn[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "tr_2_Conv3D (Conv3D)            (None, 16, 16, 5, 51 524288      activation_87[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "average_pooling3d_2 (AveragePoo (None, 8, 8, 2, 512) 0           tr_2_Conv3D[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_0_bn (BatchNormalizatio (None, 8, 8, 2, 512) 2048        average_pooling3d_2[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "activation_88 (Activation)      (None, 8, 8, 2, 512) 0           dense_3_0_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_0_bottleneck_Conv3D (Co (None, 8, 8, 2, 128) 65536       activation_88[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_0_bottleneck_bn (BatchN (None, 8, 8, 2, 128) 512         dense_3_0_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_89 (Activation)      (None, 8, 8, 2, 128) 0           dense_3_0_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_0_Conv3D (Conv3D)       (None, 8, 8, 2, 32)  110592      activation_89[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_42 (Concatenate)    (None, 8, 8, 2, 544) 0           average_pooling3d_2[0][0]        \n",
            "                                                                 dense_3_0_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_1_bn (BatchNormalizatio (None, 8, 8, 2, 544) 2176        concatenate_42[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_90 (Activation)      (None, 8, 8, 2, 544) 0           dense_3_1_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_1_bottleneck_Conv3D (Co (None, 8, 8, 2, 128) 69632       activation_90[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_1_bottleneck_bn (BatchN (None, 8, 8, 2, 128) 512         dense_3_1_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_91 (Activation)      (None, 8, 8, 2, 128) 0           dense_3_1_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_1_Conv3D (Conv3D)       (None, 8, 8, 2, 32)  110592      activation_91[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_43 (Concatenate)    (None, 8, 8, 2, 576) 0           concatenate_42[0][0]             \n",
            "                                                                 dense_3_1_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_2_bn (BatchNormalizatio (None, 8, 8, 2, 576) 2304        concatenate_43[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_92 (Activation)      (None, 8, 8, 2, 576) 0           dense_3_2_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_2_bottleneck_Conv3D (Co (None, 8, 8, 2, 128) 73728       activation_92[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_2_bottleneck_bn (BatchN (None, 8, 8, 2, 128) 512         dense_3_2_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_93 (Activation)      (None, 8, 8, 2, 128) 0           dense_3_2_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_2_Conv3D (Conv3D)       (None, 8, 8, 2, 32)  110592      activation_93[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_44 (Concatenate)    (None, 8, 8, 2, 608) 0           concatenate_43[0][0]             \n",
            "                                                                 dense_3_2_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_3_bn (BatchNormalizatio (None, 8, 8, 2, 608) 2432        concatenate_44[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_94 (Activation)      (None, 8, 8, 2, 608) 0           dense_3_3_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_3_bottleneck_Conv3D (Co (None, 8, 8, 2, 128) 77824       activation_94[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_3_bottleneck_bn (BatchN (None, 8, 8, 2, 128) 512         dense_3_3_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_95 (Activation)      (None, 8, 8, 2, 128) 0           dense_3_3_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_3_Conv3D (Conv3D)       (None, 8, 8, 2, 32)  110592      activation_95[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_45 (Concatenate)    (None, 8, 8, 2, 640) 0           concatenate_44[0][0]             \n",
            "                                                                 dense_3_3_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_4_bn (BatchNormalizatio (None, 8, 8, 2, 640) 2560        concatenate_45[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_96 (Activation)      (None, 8, 8, 2, 640) 0           dense_3_4_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_4_bottleneck_Conv3D (Co (None, 8, 8, 2, 128) 81920       activation_96[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_4_bottleneck_bn (BatchN (None, 8, 8, 2, 128) 512         dense_3_4_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_97 (Activation)      (None, 8, 8, 2, 128) 0           dense_3_4_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_4_Conv3D (Conv3D)       (None, 8, 8, 2, 32)  110592      activation_97[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_46 (Concatenate)    (None, 8, 8, 2, 672) 0           concatenate_45[0][0]             \n",
            "                                                                 dense_3_4_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_5_bn (BatchNormalizatio (None, 8, 8, 2, 672) 2688        concatenate_46[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_98 (Activation)      (None, 8, 8, 2, 672) 0           dense_3_5_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_5_bottleneck_Conv3D (Co (None, 8, 8, 2, 128) 86016       activation_98[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_5_bottleneck_bn (BatchN (None, 8, 8, 2, 128) 512         dense_3_5_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_99 (Activation)      (None, 8, 8, 2, 128) 0           dense_3_5_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_5_Conv3D (Conv3D)       (None, 8, 8, 2, 32)  110592      activation_99[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_47 (Concatenate)    (None, 8, 8, 2, 704) 0           concatenate_46[0][0]             \n",
            "                                                                 dense_3_5_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_6_bn (BatchNormalizatio (None, 8, 8, 2, 704) 2816        concatenate_47[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_100 (Activation)     (None, 8, 8, 2, 704) 0           dense_3_6_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_6_bottleneck_Conv3D (Co (None, 8, 8, 2, 128) 90112       activation_100[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_6_bottleneck_bn (BatchN (None, 8, 8, 2, 128) 512         dense_3_6_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_101 (Activation)     (None, 8, 8, 2, 128) 0           dense_3_6_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_6_Conv3D (Conv3D)       (None, 8, 8, 2, 32)  110592      activation_101[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_48 (Concatenate)    (None, 8, 8, 2, 736) 0           concatenate_47[0][0]             \n",
            "                                                                 dense_3_6_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_7_bn (BatchNormalizatio (None, 8, 8, 2, 736) 2944        concatenate_48[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_102 (Activation)     (None, 8, 8, 2, 736) 0           dense_3_7_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_7_bottleneck_Conv3D (Co (None, 8, 8, 2, 128) 94208       activation_102[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_7_bottleneck_bn (BatchN (None, 8, 8, 2, 128) 512         dense_3_7_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_103 (Activation)     (None, 8, 8, 2, 128) 0           dense_3_7_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_7_Conv3D (Conv3D)       (None, 8, 8, 2, 32)  110592      activation_103[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_49 (Concatenate)    (None, 8, 8, 2, 768) 0           concatenate_48[0][0]             \n",
            "                                                                 dense_3_7_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_8_bn (BatchNormalizatio (None, 8, 8, 2, 768) 3072        concatenate_49[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_104 (Activation)     (None, 8, 8, 2, 768) 0           dense_3_8_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_8_bottleneck_Conv3D (Co (None, 8, 8, 2, 128) 98304       activation_104[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_8_bottleneck_bn (BatchN (None, 8, 8, 2, 128) 512         dense_3_8_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_105 (Activation)     (None, 8, 8, 2, 128) 0           dense_3_8_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_8_Conv3D (Conv3D)       (None, 8, 8, 2, 32)  110592      activation_105[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_50 (Concatenate)    (None, 8, 8, 2, 800) 0           concatenate_49[0][0]             \n",
            "                                                                 dense_3_8_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_9_bn (BatchNormalizatio (None, 8, 8, 2, 800) 3200        concatenate_50[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_106 (Activation)     (None, 8, 8, 2, 800) 0           dense_3_9_bn[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_9_bottleneck_Conv3D (Co (None, 8, 8, 2, 128) 102400      activation_106[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_9_bottleneck_bn (BatchN (None, 8, 8, 2, 128) 512         dense_3_9_bottleneck_Conv3D[0][0]\n",
            "__________________________________________________________________________________________________\n",
            "activation_107 (Activation)     (None, 8, 8, 2, 128) 0           dense_3_9_bottleneck_bn[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_9_Conv3D (Conv3D)       (None, 8, 8, 2, 32)  110592      activation_107[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_51 (Concatenate)    (None, 8, 8, 2, 832) 0           concatenate_50[0][0]             \n",
            "                                                                 dense_3_9_Conv3D[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_10_bn (BatchNormalizati (None, 8, 8, 2, 832) 3328        concatenate_51[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_108 (Activation)     (None, 8, 8, 2, 832) 0           dense_3_10_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_10_bottleneck_Conv3D (C (None, 8, 8, 2, 128) 106496      activation_108[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_10_bottleneck_bn (Batch (None, 8, 8, 2, 128) 512         dense_3_10_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_109 (Activation)     (None, 8, 8, 2, 128) 0           dense_3_10_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_10_Conv3D (Conv3D)      (None, 8, 8, 2, 32)  110592      activation_109[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_52 (Concatenate)    (None, 8, 8, 2, 864) 0           concatenate_51[0][0]             \n",
            "                                                                 dense_3_10_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_11_bn (BatchNormalizati (None, 8, 8, 2, 864) 3456        concatenate_52[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_110 (Activation)     (None, 8, 8, 2, 864) 0           dense_3_11_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_11_bottleneck_Conv3D (C (None, 8, 8, 2, 128) 110592      activation_110[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_11_bottleneck_bn (Batch (None, 8, 8, 2, 128) 512         dense_3_11_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_111 (Activation)     (None, 8, 8, 2, 128) 0           dense_3_11_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_11_Conv3D (Conv3D)      (None, 8, 8, 2, 32)  110592      activation_111[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_53 (Concatenate)    (None, 8, 8, 2, 896) 0           concatenate_52[0][0]             \n",
            "                                                                 dense_3_11_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_12_bn (BatchNormalizati (None, 8, 8, 2, 896) 3584        concatenate_53[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_112 (Activation)     (None, 8, 8, 2, 896) 0           dense_3_12_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_12_bottleneck_Conv3D (C (None, 8, 8, 2, 128) 114688      activation_112[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_12_bottleneck_bn (Batch (None, 8, 8, 2, 128) 512         dense_3_12_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_113 (Activation)     (None, 8, 8, 2, 128) 0           dense_3_12_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_12_Conv3D (Conv3D)      (None, 8, 8, 2, 32)  110592      activation_113[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_54 (Concatenate)    (None, 8, 8, 2, 928) 0           concatenate_53[0][0]             \n",
            "                                                                 dense_3_12_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_13_bn (BatchNormalizati (None, 8, 8, 2, 928) 3712        concatenate_54[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_114 (Activation)     (None, 8, 8, 2, 928) 0           dense_3_13_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_13_bottleneck_Conv3D (C (None, 8, 8, 2, 128) 118784      activation_114[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_13_bottleneck_bn (Batch (None, 8, 8, 2, 128) 512         dense_3_13_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_115 (Activation)     (None, 8, 8, 2, 128) 0           dense_3_13_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_13_Conv3D (Conv3D)      (None, 8, 8, 2, 32)  110592      activation_115[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_55 (Concatenate)    (None, 8, 8, 2, 960) 0           concatenate_54[0][0]             \n",
            "                                                                 dense_3_13_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_14_bn (BatchNormalizati (None, 8, 8, 2, 960) 3840        concatenate_55[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_116 (Activation)     (None, 8, 8, 2, 960) 0           dense_3_14_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_14_bottleneck_Conv3D (C (None, 8, 8, 2, 128) 122880      activation_116[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_14_bottleneck_bn (Batch (None, 8, 8, 2, 128) 512         dense_3_14_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_117 (Activation)     (None, 8, 8, 2, 128) 0           dense_3_14_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_14_Conv3D (Conv3D)      (None, 8, 8, 2, 32)  110592      activation_117[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_56 (Concatenate)    (None, 8, 8, 2, 992) 0           concatenate_55[0][0]             \n",
            "                                                                 dense_3_14_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_15_bn (BatchNormalizati (None, 8, 8, 2, 992) 3968        concatenate_56[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_118 (Activation)     (None, 8, 8, 2, 992) 0           dense_3_15_bn[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_15_bottleneck_Conv3D (C (None, 8, 8, 2, 128) 126976      activation_118[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_15_bottleneck_bn (Batch (None, 8, 8, 2, 128) 512         dense_3_15_bottleneck_Conv3D[0][0\n",
            "__________________________________________________________________________________________________\n",
            "activation_119 (Activation)     (None, 8, 8, 2, 128) 0           dense_3_15_bottleneck_bn[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "dense_3_15_Conv3D (Conv3D)      (None, 8, 8, 2, 32)  110592      activation_119[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_57 (Concatenate)    (None, 8, 8, 2, 1024 0           concatenate_56[0][0]             \n",
            "                                                                 dense_3_15_Conv3D[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "final_bn (BatchNormalization)   (None, 8, 8, 2, 1024 4096        concatenate_57[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_120 (Activation)     (None, 8, 8, 2, 1024 0           final_bn[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling3d (Globa (None, 1024)         0           activation_120[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dense (Dense)                   (None, 1)            1025        global_average_pooling3d[0][0]   \n",
            "==================================================================================================\n",
            "Total params: 11,327,297\n",
            "Trainable params: 11,243,649\n",
            "Non-trainable params: 83,648\n",
            "__________________________________________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KEGPBzzVFajv"
      },
      "source": [
        "steps = len(X_train) // bs\n",
        "valid_steps = len(X_test) // (bs)\n",
        "\n",
        "reduce_lr = ReduceLROnPlateau(\n",
        "    monitor='val_loss', factor=0.1, patience=3, verbose=1,\n",
        "    mode='auto', min_delta=0.001, cooldown=0, min_lr=1e-8)\n",
        "\n",
        "\n",
        "chkpt = ModelCheckpoint(filepath='weights/', save_best_only=True,\n",
        "                        save_weights_only=True, monitor= 'val_accuracy', mode='max')\n",
        "\n",
        "initial_learning_rate = 3e-3\n",
        "lr_schedule = keras.optimizers.schedules.ExponentialDecay(\n",
        "    initial_learning_rate,\n",
        "    decay_steps=steps,\n",
        "    decay_rate=0.9,\n",
        "    staircase=True)\n",
        "\n",
        "def get_lr_metric(optimizer):\n",
        "    def lr(y_true, y_pred):\n",
        "        return optimizer.lr\n",
        "    return lr\n",
        "\n",
        "opt = Adam(learning_rate=lr_schedule) \n",
        "\n",
        "lr_track = get_lr_metric(opt)\n",
        "\n",
        "model.compile(loss= 'binary_crossentropy', metrics=['accuracy'], optimizer=opt)"
      ],
      "execution_count": 41,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "M-hQlgPBhZVG",
        "scrolled": true,
        "outputId": "63442ab7-e583-4328-dea7-1df93d6ed15f"
      },
      "source": [
        "start = time.time()\n",
        "\n",
        "nb_epochs = 10 \n",
        "\n",
        "# Fit the model\n",
        "history= model.fit(data_gen_train, epochs= nb_epochs, steps_per_epoch= steps,\n",
        "                              validation_data=data_gen_test, validation_steps=valid_steps,\n",
        "                              callbacks=[chkpt], #, [chkpt, clr, es , reduce_lr]\n",
        "                              class_weight=weights)\n",
        "end = time.time()\n",
        "\n",
        "print('*'*40)\n",
        "print('Took {:.2f} minutes'. format((end-start)/60))"
      ],
      "execution_count": 42,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/array_ops.py:5049: calling gather (from tensorflow.python.ops.array_ops) with validate_indices is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "The `validate_indices` argument has no effect. Indices are always validated on CPU and never validated on GPU.\n",
            "Epoch 1/10\n",
            "113/113 [==============================] - 404s 3s/step - loss: 0.7091 - accuracy: 0.6870 - val_loss: 49.3630 - val_accuracy: 0.4464\n",
            "Epoch 2/10\n",
            "113/113 [==============================] - 303s 3s/step - loss: 0.5691 - accuracy: 0.7249 - val_loss: 4.9408 - val_accuracy: 0.4464\n",
            "Epoch 3/10\n",
            "113/113 [==============================] - 304s 3s/step - loss: 0.5037 - accuracy: 0.7709 - val_loss: 0.9184 - val_accuracy: 0.7768\n",
            "Epoch 4/10\n",
            "113/113 [==============================] - 304s 3s/step - loss: 0.4846 - accuracy: 0.7740 - val_loss: 0.8584 - val_accuracy: 0.6964\n",
            "Epoch 5/10\n",
            "113/113 [==============================] - 288s 3s/step - loss: 0.4972 - accuracy: 0.7601 - val_loss: 0.9456 - val_accuracy: 0.5625\n",
            "Epoch 6/10\n",
            "113/113 [==============================] - 288s 3s/step - loss: 0.4104 - accuracy: 0.8090 - val_loss: 0.2719 - val_accuracy: 0.9018\n",
            "Epoch 7/10\n",
            "113/113 [==============================] - 288s 3s/step - loss: 0.3748 - accuracy: 0.8620 - val_loss: 0.2401 - val_accuracy: 0.9196\n",
            "Epoch 8/10\n",
            "113/113 [==============================] - 288s 3s/step - loss: 0.3421 - accuracy: 0.8581 - val_loss: 0.3074 - val_accuracy: 0.8839\n",
            "Epoch 9/10\n",
            "113/113 [==============================] - 303s 3s/step - loss: 0.3142 - accuracy: 0.8795 - val_loss: 1.1767 - val_accuracy: 0.4732\n",
            "Epoch 10/10\n",
            "113/113 [==============================] - 303s 3s/step - loss: 0.3280 - accuracy: 0.8566 - val_loss: 0.4537 - val_accuracy: 0.7768\n",
            "****************************************\n",
            "Took 51.35 minutes\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        },
        "id": "mLm1DCsTtadL",
        "outputId": "019d3ba5-a8ae-4d8e-e1b2-4040f5384ae7"
      },
      "source": [
        "print('Best train acc: {} | Best val acc: {} '.format(max(history.history['accuracy']),\n",
        "                                                      max(history.history['val_accuracy'])))\n",
        "plt.plot(history.history['accuracy'],'ro-')\n",
        "plt.plot(history.history['val_accuracy'],'bx-')\n",
        "plt.legend(['accuracy','val_accuracy'])\n",
        "plt.show()"
      ],
      "execution_count": 43,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Best train acc: 0.8871681690216064 | Best val acc: 0.9196428656578064 \n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "lQjiOl5vF6T4",
        "outputId": "3321c035-046f-4a82-d443-d00abed88ebd"
      },
      "source": [
        "start = time.time()\n",
        "\n",
        "nb_epochs = 5 \n",
        "\n",
        "# Fit the model\n",
        "history= model.fit(data_gen_train, epochs= nb_epochs, steps_per_epoch= steps,\n",
        "                              validation_data=data_gen_test, validation_steps=valid_steps,\n",
        "                              callbacks=[chkpt], #, [chkpt, clr, es , reduce_lr]\n",
        "                              class_weight=weights)\n",
        "end = time.time()\n",
        "\n",
        "print('*'*40)\n",
        "print('Took {:.2f} minutes'. format((end-start)/60))"
      ],
      "execution_count": 46,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/5\n",
            "113/113 [==============================] - 288s 3s/step - loss: 0.2525 - accuracy: 0.9004 - val_loss: 0.2821 - val_accuracy: 0.9107\n",
            "Epoch 2/5\n",
            "113/113 [==============================] - 304s 3s/step - loss: 0.2290 - accuracy: 0.9181 - val_loss: 0.2561 - val_accuracy: 0.9107\n",
            "Epoch 3/5\n",
            "113/113 [==============================] - 289s 3s/step - loss: 0.2214 - accuracy: 0.9137 - val_loss: 0.3952 - val_accuracy: 0.8036\n",
            "Epoch 4/5\n",
            "113/113 [==============================] - 288s 3s/step - loss: 0.2038 - accuracy: 0.9248 - val_loss: 0.2276 - val_accuracy: 0.9375\n",
            "Epoch 5/5\n",
            "113/113 [==============================] - 288s 3s/step - loss: 0.1950 - accuracy: 0.9292 - val_loss: 0.2112 - val_accuracy: 0.9375\n",
            "****************************************\n",
            "Took 24.31 minutes\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        },
        "id": "GU-Wmfs_MStt",
        "outputId": "754045ab-a47b-49a0-9a9c-b2abc2bc0a95"
      },
      "source": [
        "print('Best train acc: {} | Best val acc: {} '.format(max(history.history['accuracy']),\n",
        "                                                      max(history.history['val_accuracy'])))\n",
        "plt.plot(history.history['accuracy'],'ro-')\n",
        "plt.plot(history.history['val_accuracy'],'bx-')\n",
        "plt.legend(['accuracy','val_accuracy'])\n",
        "plt.show()"
      ],
      "execution_count": 48,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Best train acc: 0.9292035102844238 | Best val acc: 0.9375 \n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "H1hp_ek_MXgc",
        "outputId": "5a603b0c-3fdd-419c-ecf2-fa7308c33ced"
      },
      "source": [
        "start = time.time()\n",
        "\n",
        "nb_epochs = 3 \n",
        "\n",
        "# Fit the model\n",
        "history= model.fit(data_gen_train, epochs= nb_epochs, steps_per_epoch= steps,\n",
        "                              validation_data=data_gen_test, validation_steps=valid_steps,\n",
        "                              callbacks=[chkpt], #, [chkpt, clr, es , reduce_lr]\n",
        "                              class_weight=weights)\n",
        "end = time.time()\n",
        "\n",
        "print('*'*40)\n",
        "print('Took {:.2f} minutes'. format((end-start)/60))"
      ],
      "execution_count": 49,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/3\n",
            "113/113 [==============================] - 288s 3s/step - loss: 0.2027 - accuracy: 0.9270 - val_loss: 0.2459 - val_accuracy: 0.9018\n",
            "Epoch 2/3\n",
            "113/113 [==============================] - 303s 3s/step - loss: 0.1770 - accuracy: 0.9381 - val_loss: 0.1959 - val_accuracy: 0.9286\n",
            "Epoch 3/3\n",
            "113/113 [==============================] - 304s 3s/step - loss: 0.1740 - accuracy: 0.9314 - val_loss: 0.1911 - val_accuracy: 0.9375\n",
            "****************************************\n",
            "Took 15.38 minutes\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        },
        "id": "yqcxYX9KP9kE",
        "outputId": "eabe71c5-63e5-4b03-c929-d7d218df5aea"
      },
      "source": [
        "print('Best train acc: {} | Best val acc: {} '.format(max(history.history['accuracy']),\n",
        "                                                      max(history.history['val_accuracy'])))\n",
        "plt.plot(history.history['accuracy'],'ro-')\n",
        "plt.plot(history.history['val_accuracy'],'bx-')\n",
        "plt.legend(['accuracy','val_accuracy'])\n",
        "plt.show()"
      ],
      "execution_count": 51,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Best train acc: 0.9380530714988708 | Best val acc: 0.9375 \n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "n2u3CuqLAu_x"
      },
      "source": [
        "from IPython.display import Audio\n",
        "sound_file = 'complete.m4a'\n",
        "Audio(sound_file, autoplay=True)"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}