CNN_2.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "\n",
    "## first run \n",
    "##pip install h5py==2.8.0\n",
    "##pip install tensorflow==1.5.0 (I'm not kidding)\n",
    "\n",
    "\n",
    "\n",
    "from keras.layers import Dense,Dropout, Activation, Flatten, Conv3D, MaxPooling3D \n",
    "#from keras.regularizers import Regularizer\n",
    "from keras.regularizers import l1, l2, l1_l2\n",
    "from keras.models import Sequential\n",
    "from keras.optimizers import Adadelta\n",
    "from keras.models import load_model\n",
    "import pandas as pd\n",
    "\n",
    "import os\n",
    "import h5py"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_model(model_para = [6, 12, 18, 18, 64, 0.01, 0.4,(5, 5, 5)],foldname = os.path.abspath('.') ):\n",
    "    # input image dimensions\n",
    "    \n",
    "    #img_rows, img_cols, img_depth = 49,39,38\n",
    "    # number of convolutional filters to use\n",
    "    conv_l2 = 0.008\n",
    "    \n",
    "    full_l2  = 0.3\n",
    "    # convolution kernel size\n",
    "    kernel_size = (3,3,3)\n",
    "    # size of pooling area for max pooling\n",
    "    pool_size = (2, 2, 2)\n",
    "\n",
    "    drop_out = (model_para[5], model_para[6])\n",
    "    \n",
    "    act_function = 'tanh'\n",
    "    \n",
    "    full_connect = model_para[4]\n",
    "    \n",
    "    nb_filters = (model_para[0], model_para[1], model_para[2], model_para[3])\n",
    "#    nb_filters = (5, 10, 15, 15)\n",
    "    l1_regularizer = 0.01\n",
    "    \n",
    "    l2_regularizer = full_l2\n",
    "    \n",
    "    nb_classes = 2\n",
    "    \n",
    "    #input_shape = (1, 40, 48, 48)\n",
    "    input_shape = (40, 48, 48, 1)\n",
    "    \n",
    "    #wr = WeightRegularizer(l1=l1_regularizer,l2=l2_regularizer)\n",
    "    kr = l1_l2(l1=l1_regularizer, l2=l2_regularizer)\n",
    "    \n",
    "    \n",
    "    \n",
    "    #create cnn model\n",
    "    model = Sequential()\n",
    "    \n",
    "    \n",
    "    #keras.layers.Conv3D(filters, kernel_size, strides=(1, 1, 1), padding='valid', \n",
    "    #data_format=None, dilation_rate=(1, 1, 1), activation=None, use_bias=True, \n",
    "    #kernel_initializer='glorot_uniform', bias_initializer='zeros', kernel_regularizer=None, \n",
    "    #bias_regularizer=None, activity_regularizer=None, kernel_constraint=None, bias_constraint=None)\n",
    "   \n",
    "    #model.add(Conv3D(filters = 1,kernel_size = 3, strides = 1, activation = 'tanh', input_shape = (50, 41, 40, 1)))\n",
    "    \n",
    "    # need to specify data_format='channels_first' \n",
    "    \n",
    "    model.add(Conv3D(nb_filters[0], kernel_size[0], kernel_size[1],kernel_size[2],W_regularizer = l2(conv_l2),\n",
    "                                activation = act_function , input_shape=input_shape, data_format='channels_last'))\n",
    "    \n",
    "    model.add(MaxPooling3D(pool_size=pool_size)) \n",
    "    \n",
    "    model.add(Dropout(drop_out[0]))\n",
    "\n",
    "    model.add(Conv3D(nb_filters[1], kernel_size[0], kernel_size[1], kernel_size[2],W_regularizer = l2(conv_l2),\n",
    "                                   activation = act_function))\n",
    "    \n",
    "    model.add(MaxPooling3D(pool_size=pool_size))\n",
    "    \n",
    "    model.add(Dropout(drop_out[0]))\n",
    "    \n",
    "    model.add(Conv3D(nb_filters[2], kernel_size[0], kernel_size[1], kernel_size[2],W_regularizer = l2(conv_l2),\n",
    "                                    activation = act_function))  \n",
    "    \n",
    "    model.add(MaxPooling3D(pool_size=pool_size))    \n",
    "\n",
    "    model.add(Dropout(drop_out[0]))\n",
    "    \n",
    "    \n",
    "    model.add(Conv3D(nb_filters[3], kernel_size[0], kernel_size[1], kernel_size[2],W_regularizer = l2(conv_l2),\n",
    "                            activation = act_function))\n",
    "    #model.add(MaxPooling3D(pool_size=pool_size))    \n",
    "\n",
    "    model.add(Dropout(drop_out[1]/2))\n",
    "\n",
    "    model.add(Flatten())\n",
    "    \n",
    "    #model.add(Dense(full_connect, W_regularizer = wr,activation = act_function))\n",
    "\n",
    "    model.add(Dropout(drop_out[1]))\n",
    "    \n",
    "    model.add(Dense(nb_classes,activation = act_function))\n",
    "    \n",
    "    \n",
    "    \n",
    "    model.add(Activation('softmax'))\n",
    "    #model.add(Activation(act_function))\n",
    "    model.summary()\n",
    "\n",
    "    ADA = Adadelta(lr = 2.0, rho=0.95)\n",
    "\n",
    "    model.compile(loss= 'categorical_crossentropy',\n",
    "              optimizer= ADA,\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "    \n",
    "\n",
    "##    save parameters of cnn model to .txt             \n",
    "    sname = 'model_parameter.txt'\n",
    "    full_namem = os.path.join(foldname,sname)\n",
    "    fm = open(full_namem,'w')\n",
    "    fm.write('************CNN model parameter************ '+'\\n')\n",
    "    fm.write('Number of Convolution layer :     '+str(len(nb_filters))+'\\n')\n",
    "    fm.write('Input shape :                     '+str(input_shape)+'\\n')\n",
    "    fm.write('Number of kernal per layer ï¼?    '+str(nb_filters)+'\\n')\n",
    "    fm.write('Kernel size per layer :           '+str(kernel_size)+'\\n')\n",
    "    fm.write('Pool size per layer :             '+str(pool_size)+'\\n')\n",
    "    fm.write('Activation function per layer :   '+act_function+'\\n')\n",
    "#    fm.write('Dropout rate :                    '+str(drop_out)+'\\n')\n",
    "    fm.write('Number of full-connect layer :    '+str(full_connect)+'\\n')\n",
    "    fm.write('Coefficient of L1 regularizer :   '+str(l1_regularizer)+'\\n')\n",
    "    fm.write('Coefficient of L2 regularizer :   '+str(l2_regularizer)+'\\n')\n",
    "    fm.write('Output :                          '+str(nb_classes)+' classes'+'\\n')\n",
    "    fm.close()\n",
    "\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# do the file traversal\n",
    "# do the downsample\n",
    "# do the fit\n",
    "\n",
    "import os\n",
    "import nibabel as nib\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "\n",
    "#C:\\Users\\Reid\\Desktop\\dataSciPrinciples\\finalProj\\ADNI_Preproc\\ADNI\\AD\\036_S_1001\\2006-11-20_09_21_45.0.nii\n",
    "\n",
    "\n",
    "#img = nib.load(adni_file_path)\n",
    "#img_data = img.get_fdata()\n",
    "\n",
    "#print(img.shape)\n",
    "#print(type(img))\n",
    "#img_np = np.array(img.dataobj)\n",
    "#print(img_np.shape)\n",
    "#print(type(img_np))\n",
    "\n",
    "\n",
    "#img_np_ds = img_np[::2,::2,::2]\n",
    "#print(img_np_ds.shape)\n",
    "\n",
    "def get_patch(input_3D, patch_idx):\n",
    "    if (patch_idx < 0 | patch_idx > 26):\n",
    "        print(\"You have passed an incorrect patch index to get patch: \"+ str(patch_idx))\n",
    "    dim1_start_idxs = [0, 20, 40] \n",
    "    dim1_end_idxs =[40, 60, 80]\n",
    "    dim2_start_idxs = [0, 24, 48] \n",
    "    dim2_end_idxs =[48, 72, 96]\n",
    "    dim3_start_idxs = [0, 24, 48]\n",
    "    dim3_end_idxs = [48, 72, 96]\n",
    "    ii = int(patch_idx / 9)\n",
    "    jj = int( (patch_idx %9)  / 3)\n",
    "    kk = int(patch_idx % 3)\n",
    "    return input_3D[dim1_start_idxs[ii]:dim1_end_idxs[ii],dim2_start_idxs[jj]:dim2_end_idxs[jj],dim3_start_idxs[kk]:dim3_end_idxs[kk]]\n",
    "\n",
    "\n",
    "def get_sampleXY(path, patch_idx):\n",
    "    img = nib.load(path)\n",
    "    img_data = img.get_fdata()\n",
    "    img_np = np.array(img.dataobj)\n",
    "    img_np_ds = img_np[::2,::2,::2]\n",
    "    x = get_patch(img_np_ds, patch_idx)\n",
    "    #(batches, x, y, z, channels)\n",
    "    x = x.reshape((1, 40, 48, 48, 1))\n",
    "    isAD = \"\\\\AD\\\\\" in path\n",
    "    y = np.array([[int(isAD), 1-int(isAD)]])\n",
    "    return x, y\n",
    "    \n",
    "    \n",
    "\n",
    "\n",
    "\n",
    "#patch_0 = get_patch(img_np_ds, 0)\n",
    "#patch_0 = patch_0.reshape((1, 40, 48, 48, 1))\n",
    "#print(patch_0.shape)\n",
    "\n",
    "#patch_label = np.array([[1,0]])\n",
    "#print(patch_label)\n",
    "#print(patch_label.shape)\n",
    "#result = model.predict(patch_0)\n",
    "\n",
    "\n",
    "#sample_x, sample_y = get_sampleXY(adni_file_path, 0)\n",
    "\n",
    "#model.fit(x=sample_x, y=sample_y, epochs=25) \n",
    "#model.fit(x= sample_x, y=sample_y, epochs=1)\n",
    "\n",
    "#result = model.predict(sample_x)\n",
    "\n",
    "#print(result)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_ad_nii_IDs():\n",
    "    ad_nii_ids = []\n",
    "    patients_path = os.getcwd() + \"\\\\..\\\\..\\\\ADNI\\\\AD\\\\\" \n",
    "    for patient_folder in os.listdir(patients_path):\n",
    "        nii_path = os.path.join(patients_path, patient_folder)\n",
    "        for nii in os.listdir(nii_path):\n",
    "            file_path = os.path.join(nii_path, nii)\n",
    "            ad_nii_ids.append(file_path)\n",
    "    #print(len(ad_nii_ids))\n",
    "    #print(ad_nii_ids)\n",
    "    return ad_nii_ids\n",
    "    \n",
    "ad_nii_IDs = get_ad_nii_IDs()    \n",
    "#print(len(ad_nii_IDs))\n",
    "#print(ad_nii_IDs)\n",
    "\n",
    "\n",
    "def get_nc_nii_IDs():\n",
    "    nc_nii_ids = []\n",
    "    patients_path = os.getcwd() + \"\\\\..\\\\..\\\\ADNI\\\\CN\\\\\" \n",
    "    for patient_folder in os.listdir(patients_path):\n",
    "        nii_path = os.path.join(patients_path, patient_folder)\n",
    "        for nii in os.listdir(nii_path):\n",
    "            file_path = os.path.join(nii_path, nii)\n",
    "            nc_nii_ids.append(file_path)\n",
    "    #print(len(ad_nii_ids))\n",
    "    #print(ad_nii_ids)\n",
    "    return nc_nii_ids\n",
    "\n",
    "nc_nii_IDs = get_nc_nii_IDs()    \n",
    "#print(len(nc_nii_IDs))\n",
    "#print(nc_nii_IDs)\n",
    "\n",
    "#list(set(ad_nii_IDs).intersection(set(nc_nii_IDs)))\n",
    "\n",
    "\n",
    "def create_train_test_split_nii_ids(ad_nii_IDs, nc_nii_IDs):\n",
    "    ad_randVec = np.random.rand( len(ad_nii_IDs), 1) \n",
    "    #decide 80% for training, 20% for testing\n",
    "\n",
    "    #print(type(train_indices))\n",
    "    #print(len(train_indices))\n",
    "    #print(train_indices[0])\n",
    "    #print(train_indices[1])\n",
    "    \n",
    "    ad_train_indices = np.where(ad_randVec < 0.8)[0]\n",
    "    ad_test_indices = np.where(ad_randVec >= 0.8)[0]\n",
    "    \n",
    "    nc_randVec = np.random.rand(len(nc_nii_IDs), 1) \n",
    "    \n",
    "    nc_train_indices = np.where(nc_randVec < 0.8)[0]\n",
    "    nc_test_indices = np.where(nc_randVec >= 0.8)[0]\n",
    "    \n",
    "    train_ad_IDs= []\n",
    "    train_nc_IDs= []\n",
    "    test_ad_IDs= []\n",
    "    test_nc_IDs= []\n",
    "    \n",
    "    for index in ad_train_indices:\n",
    "        train_ad_IDs.append(ad_nii_IDs[index])\n",
    "    \n",
    "    for index in nc_train_indices:\n",
    "        train_nc_IDs.append(nc_nii_IDs[index])\n",
    "    \n",
    "    for index in ad_test_indices:\n",
    "        test_ad_IDs.append(ad_nii_IDs[index])\n",
    "    \n",
    "    for index in nc_test_indices:\n",
    "        test_nc_IDs.append(nc_nii_IDs[index])\n",
    "        \n",
    "        \n",
    "    train_IDs = {'AD':train_ad_IDs, 'NC': train_nc_IDs }\n",
    "    test_IDs = {'AD':test_ad_IDs, 'NC': test_nc_IDs }\n",
    "    \n",
    "    return train_IDs, test_IDs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "150"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_IDs = ['C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\003_S_1059\\\\2008-12-23_09_15_12.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\011_S_0053\\\\2007-11-29_12_31_01.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\011_S_0183\\\\2007-04-03_12_25_27.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\011_S_0183\\\\2007-04-03_12_42_38.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\012_S_0712\\\\2007-07-25_09_20_03.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\018_S_0286\\\\2007-05-15_09_24_51.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\018_S_0286\\\\2007-05-15_09_32_40.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\020_S_0213\\\\2006-03-22_13_33_31.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\020_S_0213\\\\2007-06-04_13_21_27.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_0084\\\\2007-01-24_09_47_00.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_0084\\\\2006-01-04_08_51_47.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_0084\\\\2008-01-22_08_09_44.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_0916\\\\2007-09-28_10_07_05.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_1262\\\\2008-04-02_08_41_46.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_1262\\\\2009-03-13_08_38_41.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_1262\\\\2007-02-09_08_36_16.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_1289\\\\2007-02-12_11_41_49.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\035_S_0341\\\\2008-04-24_14_23_37.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\003_S_0931\\\\2006-10-11_11_49_21.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\010_S_0420\\\\2006-06-15_15_41_24.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\010_S_0420\\\\2006-06-15_16_05_36.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\011_S_0002\\\\2005-08-26_08_45_00.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\012_S_1133\\\\2007-01-12_11_30_59.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\012_S_1133\\\\2007-01-12_11_21_49.0.nii.gz',  'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\016_S_0538\\\\2006-12-15_16_54_13.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\016_S_0538\\\\2006-08-02_11_02_27.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\020_S_0097\\\\2006-01-12_10_30_14.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\023_S_0926\\\\2006-09-27_12_09_35.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\032_S_1169\\\\2006-12-21_08_59_25.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\032_S_1169\\\\2006-12-21_08_50_09.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\036_S_0813\\\\2006-08-25_10_11_42.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\036_S_0813\\\\2010-10-05_15_31_10.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\041_S_0898\\\\2006-10-18_10_33_26.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\057_S_0779\\\\2006-08-16_13_05_14.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\057_S_0934\\\\2006-10-04_14_02_01.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\003_S_1059\\\\2006-11-09_17_28_06.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\003_S_1257\\\\2007-03-01_10_20_36.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\011_S_0183\\\\2006-03-03_12_21_01.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\011_S_0183\\\\2008-04-09_12_54_26.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\011_S_0053\\\\2005-11-14_09_04_44.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\011_S_0053\\\\2006-11-28_12_33_23.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\011_S_0053\\\\2007-11-29_12_56_08.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\032_S_1101\\\\2006-11-21_11_41_21.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\032_S_1101\\\\2006-11-21_11_50_37.0.nii.gz']\n",
    "train_IDs = ['C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\003_S_1257\\\\2008-05-15_13_41_31.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\010_S_0829\\\\2007-02-08_14_25_09.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\010_S_0829\\\\2007-02-08_14_33_41.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\011_S_0003\\\\2005-09-01_10_33_55.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\011_S_0003\\\\2006-09-12_12_24_48.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\011_S_0003\\\\2007-09-12_12_48_32.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\011_S_0010\\\\2005-11-07_08_19_30.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\011_S_0010\\\\2006-11-09_12_01_54.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\011_S_0010\\\\2007-11-07_12_23_57.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\012_S_0689\\\\2007-07-19_10_48_16.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\012_S_0689\\\\2007-07-19_10_57_26.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\012_S_0712\\\\2007-07-25_09_02_14.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\012_S_0712\\\\2007-07-25_09_11_22.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\012_S_0720\\\\2007-08-21_10_35_07.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\012_S_0720\\\\2007-08-21_10_43_51.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\012_S_0803\\\\2006-09-22_07_58_57.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\012_S_0803\\\\2006-09-22_08_07_49.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\013_S_0592\\\\2006-08-17_15_07_00.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\013_S_0699\\\\2006-08-28_15_00_26.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\013_S_0996\\\\2006-11-06_14_53_40.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\013_S_0996\\\\2008-11-17_13_37_17.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\013_S_1161\\\\2006-12-20_14_34_30.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\013_S_1205\\\\2008-01-24_12_48_08.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\013_S_1205\\\\2009-03-16_16_09_31.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\013_S_1205\\\\2007-01-11_13_55_08.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\016_S_0991\\\\2006-11-01_14_40_40.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\016_S_0991\\\\2007-12-13_14_26_09.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\016_S_0991\\\\2009-04-22_16_02_55.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\016_S_1263\\\\2007-02-27_13_43_25.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\018_S_0633\\\\2008-09-25_11_37_32.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\018_S_0682\\\\2008-07-22_16_11_12.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\022_S_0007\\\\2005-09-13_11_30_12.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\022_S_0129\\\\2006-02-06_11_54_32.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\022_S_0129\\\\2008-02-19_10_33_04.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\022_S_0219\\\\2006-03-17_13_47_47.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\022_S_0543\\\\2006-05-22_09_27_24.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_0083\\\\2007-01-08_12_50_12.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_0083\\\\2008-01-31_09_08_41.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_0083\\\\2005-12-14_08_30_26.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_0093\\\\2006-01-03_08_54_30.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_0093\\\\2007-02-13_10_37_18.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_0093\\\\2008-01-28_15_05_27.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_0139\\\\2006-01-24_11_12_57.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_0139\\\\2007-02-09_09_34_44.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_0139\\\\2007-02-09_09_55_08.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_0139\\\\2008-03-04_07_32_59.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_0916\\\\2006-09-22_14_14_58.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\023_S_0916\\\\2008-09-30_09_00_52.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\032_S_0147\\\\2007-02-19_15_06_33.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\032_S_0147\\\\2007-02-19_15_15_49.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\032_S_0400\\\\2006-04-17_11_44_52.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\032_S_0400\\\\2007-05-18_14_55_05.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\032_S_0400\\\\2007-05-18_15_04_21.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\032_S_0400\\\\2006-04-17_11_54_09.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\032_S_1037\\\\2006-11-02_10_31_19.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\032_S_1037\\\\2006-11-02_10_40_36.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\035_S_0341\\\\2006-04-06_10_28_57.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\035_S_0341\\\\2007-06-21_14_30_50.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\036_S_0577\\\\2006-05-26_12_09_33.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\036_S_0577\\\\2007-06-06_14_40_25.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\036_S_0577\\\\2008-05-29_13_12_44.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\036_S_0759\\\\2006-08-22_11_08_33.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\036_S_0759\\\\2007-08-30_12_08_53.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\036_S_0759\\\\2008-09-10_13_26_02.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\036_S_0760\\\\2006-08-25_11_11_22.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\036_S_0760\\\\2007-09-27_11_57_36.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\036_S_0760\\\\2008-10-07_12_46_02.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\036_S_1001\\\\2006-11-20_09_21_45.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\036_S_1001\\\\2006-11-20_09_47_56.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\036_S_1001\\\\2007-11-30_09_59_20.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\041_S_1368\\\\2007-03-05_10_29_58.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\041_S_1368\\\\2008-03-24_11_34_50.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\AD\\\\041_S_1368\\\\2009-03-16_09_28_42.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\003_S_0907\\\\2006-09-11_11_38_13.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\003_S_1021\\\\2006-11-01_11_29_25.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\010_S_0472\\\\2006-11-22_11_19_28.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\010_S_0472\\\\2006-11-22_11_30_27.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\011_S_0005\\\\2005-09-02_10_13_57.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\011_S_0008\\\\2005-09-13_13_26_10.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\011_S_0016\\\\2005-09-27_11_12_22.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\011_S_0021\\\\2005-10-10_13_12_17.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\011_S_0022\\\\2005-10-10_08_39_18.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\011_S_0023\\\\2005-10-31_09_18_49.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\011_S_0023\\\\2010-01-06_11_00_25.0.nii.gz','C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\012_S_1009\\\\2006-11-01_07_17_56.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\012_S_1009\\\\2006-11-01_07_26_40.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\012_S_1212\\\\2007-01-12_08_09_47.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\012_S_1212\\\\2007-01-12_08_18_28.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\013_S_0502\\\\2006-07-25_13_35_33.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\013_S_0575\\\\2006-08-14_14_59_59.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\013_S_1035\\\\2006-11-02_10_59_41.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\013_S_1276\\\\2007-03-05_13_31_20.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\016_S_0359\\\\2006-04-06_09_06_17.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\016_S_0359\\\\2010-11-04_11_04_56.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\018_S_0043\\\\2010-02-19_14_30_11.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\020_S_0883\\\\2006-09-15_13_03_24.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\020_S_0883\\\\2006-09-15_12_55_11.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\020_S_0899\\\\2006-09-25_19_44_35.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\020_S_1288\\\\2007-02-21_17_15_15.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\022_S_0014\\\\2005-09-29_15_36_16.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\022_S_0066\\\\2005-11-23_11_36_42.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\022_S_0096\\\\2006-01-18_12_32_57.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\022_S_0130\\\\2006-01-23_13_33_37.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\023_S_0031\\\\2005-10-12_15_48_09.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\023_S_0031\\\\2009-10-22_10_13_56.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\023_S_0058\\\\2005-11-30_09_36_10.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\023_S_0058\\\\2010-03-19_13_05_44.0.nii.gz',  'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\023_S_0061\\\\2005-11-29_08_54_05.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\023_S_0061\\\\2010-02-25_10_01_04.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\023_S_0081\\\\2006-01-10_08_27_21.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\023_S_0081\\\\2010-03-11_09_43_51.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\023_S_0926\\\\2010-10-27_12_19_05.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\023_S_0963\\\\2006-10-04_16_15_40.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\023_S_1190\\\\2007-01-08_16_43_31.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\023_S_1306\\\\2007-02-16_08_05_58.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\032_S_0479\\\\2006-05-15_12_17_39.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\032_S_0479\\\\2006-05-15_12_26_55.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\032_S_0479\\\\2010-07-07_09_10_00.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\032_S_0677\\\\2006-06-29_10_16_23.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\032_S_0677\\\\2006-06-29_10_25_39.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\032_S_0677\\\\2010-07-28_12_24_26.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\035_S_0048\\\\2005-11-22_13_31_20.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\035_S_0156\\\\2006-01-31_15_19_50.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\035_S_0156\\\\2010-02-22_16_54_56.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\035_S_0555\\\\2006-06-06_14_41_03.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\035_S_0555\\\\2006-11-14_07_22_47.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\035_S_0555\\\\2010-12-13_09_17_32.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\036_S_0576\\\\2006-06-01_16_25_20.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\036_S_0672\\\\2006-07-21_07_44_23.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\036_S_0672\\\\2010-12-06_11_59_55.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\036_S_1023\\\\2006-12-20_09_29_17.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\041_S_0125\\\\2006-01-13_13_00_31.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\041_S_0125\\\\2010-03-02_09_14_35.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\041_S_0125\\\\2010-03-02_09_40_09.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\041_S_0262\\\\2006-03-17_11_27_22.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\041_S_0262\\\\2006-03-17_11_17_57.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\041_S_0262\\\\2010-04-07_12_46_33.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\041_S_0262\\\\2010-04-07_13_12_49.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\041_S_1002\\\\2006-12-08_12_23_31.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\051_S_1123\\\\2006-12-15_10_20_28.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\057_S_0643\\\\2006-06-21_13_35_43.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\057_S_0818\\\\2006-09-06_13_54_55.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\062_S_0578\\\\2006-05-30_15_56_34.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\062_S_0768\\\\2006-08-02_09_18_44.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\062_S_1099\\\\2006-11-20_14_25_07.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\067_S_0019\\\\2005-10-12_10_40_27.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\067_S_0024\\\\2005-10-14_09_34_13.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\067_S_0056\\\\2005-11-09_09_25_54.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\067_S_0059\\\\2005-12-12_09_12_14.0.nii.gz', 'C:\\\\Users\\\\caleb\\\\Desktop\\\\Alz-Finders\\\\models\\\\..\\\\..\\\\ADNI\\\\CN\\\\067_S_0177\\\\2006-03-10_14_39_05.0.nii.gz']\n",
    "len(test_IDs)\n",
    "len(train_IDs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "****EPOCH NUMBER 23****\n",
      "NOW TRAINING NET NUMBER 0\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\022_S_0219\\2006-03-17_13_47_47.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 1s 687ms/step - loss: 0.7126 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\035_S_0156\\2006-01-31_15_19_50.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7081 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\062_S_1099\\2006-11-20_14_25_07.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7250 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\036_S_0759\\2008-09-10_13_26_02.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7052 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\023_S_0139\\2007-02-09_09_34_44.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7074 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\032_S_0677\\2010-07-28_12_24_26.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7146 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\013_S_1276\\2007-03-05_13_31_20.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.8610 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\023_S_0058\\2010-03-19_13_05_44.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7083 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\010_S_0472\\2006-11-22_11_30_27.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7075 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\032_S_0400\\2007-05-18_14_55_05.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7096 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\032_S_1037\\2006-11-02_10_31_19.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7067 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\010_S_0829\\2007-02-08_14_33_41.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7071 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\035_S_0555\\2006-11-14_07_22_47.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 0us/step - loss: 0.7316 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\035_S_0555\\2006-06-06_14_41_03.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7072 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\036_S_0576\\2006-06-01_16_25_20.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7079 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\013_S_1035\\2006-11-02_10_59_41.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7072 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\020_S_0883\\2006-09-15_13_03_24.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7074 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\032_S_0400\\2006-04-17_11_44_52.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 0us/step - loss: 0.7071 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\036_S_0672\\2010-12-06_11_59_55.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7091 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\041_S_1368\\2007-03-05_10_29_58.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7064 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\035_S_0048\\2005-11-22_13_31_20.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7075 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\012_S_1009\\2006-11-01_07_17_56.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7101 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\023_S_0081\\2010-03-11_09_43_51.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7067 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\013_S_1205\\2009-03-16_16_09_31.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7054 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\036_S_0577\\2008-05-29_13_12_44.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 0us/step - loss: 0.7068 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\041_S_0125\\2010-03-02_09_14_35.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7061 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\023_S_0139\\2006-01-24_11_12_57.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7078 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\041_S_0262\\2006-03-17_11_17_57.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7058 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\023_S_1306\\2007-02-16_08_05_58.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7062 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\016_S_0991\\2009-04-22_16_02_55.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7052 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\011_S_0010\\2007-11-07_12_23_57.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6970 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\041_S_0262\\2010-04-07_12_46_33.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7047 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\012_S_0720\\2007-08-21_10_35_07.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7051 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\018_S_0043\\2010-02-19_14_30_11.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7062 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\022_S_0543\\2006-05-22_09_27_24.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7053 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\023_S_0083\\2008-01-31_09_08_41.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7225 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\012_S_0803\\2006-09-22_08_07_49.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7043 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\010_S_0829\\2007-02-08_14_25_09.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6248 - acc: 1.0000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\036_S_1023\\2006-12-20_09_29_17.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7100 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\022_S_0066\\2005-11-23_11_36_42.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7053 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\032_S_0479\\2006-05-15_12_17_39.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7049 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\013_S_0575\\2006-08-14_14_59_59.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7038 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\022_S_0014\\2005-09-29_15_36_16.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 1.0019 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\032_S_0479\\2006-05-15_12_26_55.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7054 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\036_S_1001\\2006-11-20_09_21_45.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7051 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\057_S_0818\\2006-09-06_13_54_55.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7058 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\041_S_1368\\2008-03-24_11_34_50.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7000 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\032_S_0147\\2007-02-19_15_15_49.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7040 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\023_S_0061\\2005-11-29_08_54_05.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7073 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\020_S_0899\\2006-09-25_19_44_35.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7062 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\022_S_0129\\2006-02-06_11_54_32.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7082 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\067_S_0056\\2005-11-09_09_25_54.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7042 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\036_S_0760\\2008-10-07_12_46_02.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6645 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\041_S_0262\\2010-04-07_13_12_49.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6973 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\012_S_1212\\2007-01-12_08_09_47.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7091 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\032_S_0400\\2006-04-17_11_54_09.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7026 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\036_S_0577\\2007-06-06_14_40_25.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7005 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\023_S_1190\\2007-01-08_16_43_31.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7050 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\023_S_0963\\2006-10-04_16_15_40.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7034 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\011_S_0021\\2005-10-10_13_12_17.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7027 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\010_S_0472\\2006-11-22_11_19_28.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 31ms/step - loss: 0.7293 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\067_S_0177\\2006-03-10_14_39_05.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7032 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\AD\\012_S_0712\\2007-07-25_09_02_14.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7031 - acc: 1.0000\n",
      "NOW TRAINING NET NUMBER 0\n",
      "C:\\Users\\caleb\\Desktop\\Alz-Finders\\models\\..\\..\\ADNI\\CN\\011_S_0022\\2005-10-10_08_39_18.0.nii.gz\n",
      "Epoch 1/1\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7032 - acc: 0.0000e+00\n",
      "NOW TRAINING NET NUMBER 0\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-11-84e85a992d54>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[0;32m     13\u001b[0m         \u001b[1;32mfor\u001b[0m \u001b[0mnii_path\u001b[0m \u001b[1;32min\u001b[0m \u001b[0mtrain_IDs\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     14\u001b[0m             \u001b[0mprint\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m\"NOW TRAINING NET NUMBER \"\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0mstr\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mneural_net_idx\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 15\u001b[1;33m             \u001b[0mx\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mget_sampleXY\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mnii_path\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mpatch_idx\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mneural_net_idx\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;31m## one brain patch == one datum == one example\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     16\u001b[0m             \u001b[0mprint\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mnii_path\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     17\u001b[0m             \u001b[0mmodel\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0my\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m;\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m<ipython-input-5-18683df06649>\u001b[0m in \u001b[0;36mget_sampleXY\u001b[1;34m(path, patch_idx)\u001b[0m\n\u001b[0;32m     43\u001b[0m     \u001b[0mimg\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mnib\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mload\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mpath\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     44\u001b[0m     \u001b[0mimg_data\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mimg\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mget_fdata\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 45\u001b[1;33m     \u001b[0mimg_np\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0marray\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mimg\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mdataobj\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     46\u001b[0m     \u001b[0mimg_np_ds\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mimg_np\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;36m2\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;36m2\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;36m2\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     47\u001b[0m     \u001b[0mx\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mget_patch\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mimg_np_ds\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mpatch_idx\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\site-packages\\nibabel\\arrayproxy.py\u001b[0m in \u001b[0;36m__array__\u001b[1;34m(self)\u001b[0m\n\u001b[0;32m    354\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0m__array__\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    355\u001b[0m         \u001b[1;31m# Read array and scale\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 356\u001b[1;33m         \u001b[0mraw_data\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mget_unscaled\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    357\u001b[0m         \u001b[1;32mreturn\u001b[0m \u001b[0mapply_read_scaling\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mraw_data\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_slope\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_inter\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    358\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\site-packages\\nibabel\\arrayproxy.py\u001b[0m in \u001b[0;36mget_unscaled\u001b[1;34m(self)\u001b[0m\n\u001b[0;32m    349\u001b[0m                                        \u001b[0moffset\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_offset\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    350\u001b[0m                                        \u001b[0morder\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0morder\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 351\u001b[1;33m                                        mmap=self._mmap)\n\u001b[0m\u001b[0;32m    352\u001b[0m         \u001b[1;32mreturn\u001b[0m \u001b[0mraw_data\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    353\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\site-packages\\nibabel\\volumeutils.py\u001b[0m in \u001b[0;36marray_from_file\u001b[1;34m(shape, in_dtype, infile, offset, order, mmap)\u001b[0m\n\u001b[0;32m    526\u001b[0m     \u001b[1;32mif\u001b[0m \u001b[0mhasattr\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0minfile\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m'readinto'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    527\u001b[0m         \u001b[0mdata_bytes\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mbytearray\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mn_bytes\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 528\u001b[1;33m         \u001b[0mn_read\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0minfile\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mreadinto\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mdata_bytes\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    529\u001b[0m         \u001b[0mneeds_copy\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;32mFalse\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    530\u001b[0m     \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\gzip.py\u001b[0m in \u001b[0;36mread\u001b[1;34m(self, size)\u001b[0m\n\u001b[0;32m    274\u001b[0m             \u001b[1;32mimport\u001b[0m \u001b[0merrno\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    275\u001b[0m             \u001b[1;32mraise\u001b[0m \u001b[0mOSError\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0merrno\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mEBADF\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m\"read() on write-only GzipFile object\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 276\u001b[1;33m         \u001b[1;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_buffer\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mread\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0msize\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    277\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    278\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0mread1\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0msize\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;33m-\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\_compression.py\u001b[0m in \u001b[0;36mreadinto\u001b[1;34m(self, b)\u001b[0m\n\u001b[0;32m     66\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0mreadinto\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mb\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     67\u001b[0m         \u001b[1;32mwith\u001b[0m \u001b[0mmemoryview\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mb\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;32mas\u001b[0m \u001b[0mview\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mview\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mcast\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m\"B\"\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;32mas\u001b[0m \u001b[0mbyte_view\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 68\u001b[1;33m             \u001b[0mdata\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mread\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mlen\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbyte_view\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     69\u001b[0m             \u001b[0mbyte_view\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m:\u001b[0m\u001b[0mlen\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mdata\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     70\u001b[0m         \u001b[1;32mreturn\u001b[0m \u001b[0mlen\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\gzip.py\u001b[0m in \u001b[0;36mread\u001b[1;34m(self, size)\u001b[0m\n\u001b[0;32m    467\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    468\u001b[0m             \u001b[1;31m# Read a chunk of data from the file\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 469\u001b[1;33m             \u001b[0mbuf\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_fp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mread\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mio\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mDEFAULT_BUFFER_SIZE\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    470\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    471\u001b[0m             \u001b[0muncompress\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_decompressor\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mdecompress\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbuf\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0msize\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\gzip.py\u001b[0m in \u001b[0;36mread\u001b[1;34m(self, size)\u001b[0m\n\u001b[0;32m     80\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0mread\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0msize\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     81\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_read\u001b[0m \u001b[1;32mis\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 82\u001b[1;33m             \u001b[1;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mfile\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mread\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0msize\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     83\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_read\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0msize\u001b[0m \u001b[1;33m<=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_length\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     84\u001b[0m             \u001b[0mread\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_read\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "import random\n",
    "\n",
    "#print(train_IDs)\n",
    "\n",
    "for epoch_number in range(23, 30):\n",
    "    print(\"****EPOCH NUMBER \" + str(epoch_number) + \"****\")\n",
    "#epoch_number = epoch_number + 1\n",
    "    random.shuffle(train_IDs)\n",
    "    for neural_net_idx in range(27):\n",
    "        print(\"NOW TRAINING NET NUMBER \" + str(neural_net_idx))\n",
    "        model = load_model(os.getcwd() + \"\\\\model_weights\\\\saved_weights_net_\" + str(neural_net_idx) + \"_epoch_\" + str(epoch_number-1) + \".h5\")\n",
    "        #samples_path = os.getcwd() + \"\\\\..\\\\..\\\\ADNI\\\\AD\\\\\"\n",
    "        for nii_path in train_IDs:\n",
    "            print(\"NOW TRAINING NET NUMBER \" + str(neural_net_idx))\n",
    "            x, y = get_sampleXY(nii_path, patch_idx=neural_net_idx) ## one brain patch == one datum == one example\n",
    "            print(nii_path)\n",
    "            model.fit(x=x, y=y);\n",
    "        model.save(os.getcwd() + \"\\\\model_weights\\\\saved_weights_net_\" + str(neural_net_idx) + \"_epoch_\" + str(epoch_number) + \".h5\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Getting predictions for NN0\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-54-e8bb16a7ec31>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[0;32m      5\u001b[0m     \u001b[0mvector\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;33m[\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      6\u001b[0m     \u001b[0mprint\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m'Getting predictions for NN'\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0mstr\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mneural_net_idx\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 7\u001b[1;33m     \u001b[0mmodel\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mload_model\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mos\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mgetcwd\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m+\u001b[0m \u001b[1;34m\"\\\\model_weights\\\\saved_weights_net_\"\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0mstr\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mneural_net_idx\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m+\u001b[0m \u001b[1;34m\"_epoch_19.h5\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m      8\u001b[0m     \u001b[1;32mfor\u001b[0m \u001b[0mnii_path\u001b[0m \u001b[1;32min\u001b[0m \u001b[0mtrain_IDs\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      9\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\site-packages\\keras\\engine\\saving.py\u001b[0m in \u001b[0;36mload_model\u001b[1;34m(filepath, custom_objects, compile)\u001b[0m\n\u001b[0;32m    417\u001b[0m     \u001b[0mf\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mh5dict\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mfilepath\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m'r'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    418\u001b[0m     \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 419\u001b[1;33m         \u001b[0mmodel\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0m_deserialize_model\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mf\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mcustom_objects\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mcompile\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    420\u001b[0m     \u001b[1;32mfinally\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    421\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mopened_new_file\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\site-packages\\keras\\engine\\saving.py\u001b[0m in \u001b[0;36m_deserialize_model\u001b[1;34m(f, custom_objects, compile)\u001b[0m\n\u001b[0;32m    323\u001b[0m                                        optimizer_weight_names]\n\u001b[0;32m    324\u001b[0m             \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 325\u001b[1;33m                 \u001b[0mmodel\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0moptimizer\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mset_weights\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0moptimizer_weight_values\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    326\u001b[0m             \u001b[1;32mexcept\u001b[0m \u001b[0mValueError\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    327\u001b[0m                 warnings.warn('Error in loading the saved optimizer '\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\site-packages\\keras\\optimizers.py\u001b[0m in \u001b[0;36mset_weights\u001b[1;34m(self, weights)\u001b[0m\n\u001b[0;32m    124\u001b[0m                              'of the optimizer (' + str(len(params)) + ')')\n\u001b[0;32m    125\u001b[0m         \u001b[0mweight_value_tuples\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;33m[\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 126\u001b[1;33m         \u001b[0mparam_values\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mK\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mbatch_get_value\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mparams\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    127\u001b[0m         \u001b[1;32mfor\u001b[0m \u001b[0mpv\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mp\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mw\u001b[0m \u001b[1;32min\u001b[0m \u001b[0mzip\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mparam_values\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mparams\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mweights\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    128\u001b[0m             \u001b[1;32mif\u001b[0m \u001b[0mpv\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mshape\u001b[0m \u001b[1;33m!=\u001b[0m \u001b[0mw\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py\u001b[0m in \u001b[0;36mbatch_get_value\u001b[1;34m(ops)\u001b[0m\n\u001b[0;32m   2418\u001b[0m     \"\"\"\n\u001b[0;32m   2419\u001b[0m     \u001b[1;32mif\u001b[0m \u001b[0mops\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 2420\u001b[1;33m         \u001b[1;32mreturn\u001b[0m \u001b[0mget_session\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mrun\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mops\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   2421\u001b[0m     \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   2422\u001b[0m         \u001b[1;32mreturn\u001b[0m \u001b[1;33m[\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py\u001b[0m in \u001b[0;36mget_session\u001b[1;34m()\u001b[0m\n\u001b[0;32m    204\u001b[0m                     \u001b[0mv\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_keras_initialized\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;32mTrue\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    205\u001b[0m                 \u001b[1;32mif\u001b[0m \u001b[0muninitialized_vars\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 206\u001b[1;33m                     \u001b[0msession\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mrun\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mtf\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mvariables_initializer\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0muninitialized_vars\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    207\u001b[0m     \u001b[1;31m# hack for list_devices() function.\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    208\u001b[0m     \u001b[1;31m# list_devices() function is not available under tensorflow r1.3.\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\site-packages\\tensorflow\\python\\client\\session.py\u001b[0m in \u001b[0;36mrun\u001b[1;34m(self, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[0;32m    875\u001b[0m     \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    876\u001b[0m       result = self._run(None, fetches, feed_dict, options_ptr,\n\u001b[1;32m--> 877\u001b[1;33m                          run_metadata_ptr)\n\u001b[0m\u001b[0;32m    878\u001b[0m       \u001b[1;32mif\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    879\u001b[0m         \u001b[0mproto_data\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mtf_session\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mTF_GetBuffer\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mrun_metadata_ptr\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\site-packages\\tensorflow\\python\\client\\session.py\u001b[0m in \u001b[0;36m_run\u001b[1;34m(self, handle, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[0;32m   1098\u001b[0m     \u001b[1;32mif\u001b[0m \u001b[0mfinal_fetches\u001b[0m \u001b[1;32mor\u001b[0m \u001b[0mfinal_targets\u001b[0m \u001b[1;32mor\u001b[0m \u001b[1;33m(\u001b[0m\u001b[0mhandle\u001b[0m \u001b[1;32mand\u001b[0m \u001b[0mfeed_dict_tensor\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1099\u001b[0m       results = self._do_run(handle, final_targets, final_fetches,\n\u001b[1;32m-> 1100\u001b[1;33m                              feed_dict_tensor, options, run_metadata)\n\u001b[0m\u001b[0;32m   1101\u001b[0m     \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1102\u001b[0m       \u001b[0mresults\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;33m[\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\site-packages\\tensorflow\\python\\client\\session.py\u001b[0m in \u001b[0;36m_do_run\u001b[1;34m(self, handle, target_list, fetch_list, feed_dict, options, run_metadata)\u001b[0m\n\u001b[0;32m   1270\u001b[0m     \u001b[1;32mif\u001b[0m \u001b[0mhandle\u001b[0m \u001b[1;32mis\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1271\u001b[0m       return self._do_call(_run_fn, feeds, fetches, targets, options,\n\u001b[1;32m-> 1272\u001b[1;33m                            run_metadata)\n\u001b[0m\u001b[0;32m   1273\u001b[0m     \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1274\u001b[0m       \u001b[1;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_do_call\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0m_prun_fn\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mfeeds\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mfetches\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\site-packages\\tensorflow\\python\\client\\session.py\u001b[0m in \u001b[0;36m_do_call\u001b[1;34m(self, fn, *args)\u001b[0m\n\u001b[0;32m   1276\u001b[0m   \u001b[1;32mdef\u001b[0m \u001b[0m_do_call\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mfn\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1277\u001b[0m     \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1278\u001b[1;33m       \u001b[1;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1279\u001b[0m     \u001b[1;32mexcept\u001b[0m \u001b[0merrors\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mOpError\u001b[0m \u001b[1;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1280\u001b[0m       \u001b[0mmessage\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mcompat\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mas_text\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0me\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mmessage\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\site-packages\\tensorflow\\python\\client\\session.py\u001b[0m in \u001b[0;36m_run_fn\u001b[1;34m(feed_dict, fetch_list, target_list, options, run_metadata)\u001b[0m\n\u001b[0;32m   1259\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0m_run_fn\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mfeed_dict\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtarget_list\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0moptions\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1260\u001b[0m       \u001b[1;31m# Ensure any changes to the graph are reflected in the runtime.\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1261\u001b[1;33m       \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_extend_graph\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1262\u001b[0m       return self._call_tf_sessionrun(\n\u001b[0;32m   1263\u001b[0m           options, feed_dict, fetch_list, target_list, run_metadata)\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tensorflow\\lib\\site-packages\\tensorflow\\python\\client\\session.py\u001b[0m in \u001b[0;36m_extend_graph\u001b[1;34m(self)\u001b[0m\n\u001b[0;32m   1293\u001b[0m   \u001b[1;32mdef\u001b[0m \u001b[0m_extend_graph\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1294\u001b[0m     \u001b[1;32mwith\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_graph\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_session_run_lock\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m  \u001b[1;31m# pylint: disable=protected-access\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1295\u001b[1;33m       \u001b[0mtf_session\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mExtendSession\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_session\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1296\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1297\u001b[0m   \u001b[1;31m# The threshold to run garbage collection to delete dead tensors.\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "df = pd.DataFrame()\n",
    "df.to_csv(\"cnn_out_20_blind.csv\")\n",
    "labels = []\n",
    "for neural_net_idx in range(27):\n",
    "    vector = []\n",
    "    print('Getting predictions for NN' + str(neural_net_idx))\n",
    "    model = load_model(os.getcwd() + \"\\\\model_weights\\\\saved_weights_net_\" + str(neural_net_idx) + \"_epoch_19.h5\")\n",
    "    for nii_path in train_IDs:\n",
    "        \n",
    "        x, y = get_sampleXY(nii_path, patch_idx=neural_net_idx)\n",
    "        yhat = model.predict(x=x)\n",
    "        #print(yhat)\n",
    "        vector.append(yhat[0][0])\n",
    "        #print(vector)\n",
    "        if(neural_net_idx == 0 ):\n",
    "            labels.append(y[0][0])\n",
    "            \n",
    "    df[str(neural_net_idx)] = vector\n",
    "    #print(df)\n",
    "    \n",
    "df[\"Truth\"] = labels\n",
    "df.to_csv(\"cnn_out_20_blind.csv\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def report_best_scores(results, n_top=3):\n",
    "    for i in range(1, n_top + 1):\n",
    "        candidates = np.flatnonzero(results['rank_test_score'] == i)\n",
    "        for candidate in candidates:\n",
    "            print(\"Model with rank: {0}\".format(i))\n",
    "            print(\"Mean validation score: {0:.3f} (std: {1:.3f})\".format(\n",
    "                  results['mean_test_score'][candidate],\n",
    "                  results['std_test_score'][candidate]))\n",
    "            print(\"Parameters: {0}\".format(results['params'][candidate]))\n",
    "            print(\"\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "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>Unnamed: 0</th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>...</th>\n",
       "      <th>18</th>\n",
       "      <th>19</th>\n",
       "      <th>20</th>\n",
       "      <th>21</th>\n",
       "      <th>22</th>\n",
       "      <th>23</th>\n",
       "      <th>24</th>\n",
       "      <th>25</th>\n",
       "      <th>26</th>\n",
       "      <th>Truth</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0.509387</td>\n",
       "      <td>0.500629</td>\n",
       "      <td>0.499831</td>\n",
       "      <td>0.498434</td>\n",
       "      <td>0.500153</td>\n",
       "      <td>0.483391</td>\n",
       "      <td>0.788593</td>\n",
       "      <td>0.439439</td>\n",
       "      <td>0.500314</td>\n",
       "      <td>...</td>\n",
       "      <td>0.496945</td>\n",
       "      <td>0.501489</td>\n",
       "      <td>0.499929</td>\n",
       "      <td>0.513762</td>\n",
       "      <td>0.499473</td>\n",
       "      <td>0.499875</td>\n",
       "      <td>0.499608</td>\n",
       "      <td>0.500897</td>\n",
       "      <td>0.500032</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>0.509023</td>\n",
       "      <td>0.500618</td>\n",
       "      <td>0.499850</td>\n",
       "      <td>0.498435</td>\n",
       "      <td>0.500148</td>\n",
       "      <td>0.483310</td>\n",
       "      <td>0.631625</td>\n",
       "      <td>0.436612</td>\n",
       "      <td>0.500304</td>\n",
       "      <td>...</td>\n",
       "      <td>0.497171</td>\n",
       "      <td>0.501455</td>\n",
       "      <td>0.499915</td>\n",
       "      <td>0.514782</td>\n",
       "      <td>0.499460</td>\n",
       "      <td>0.499882</td>\n",
       "      <td>0.499597</td>\n",
       "      <td>0.500825</td>\n",
       "      <td>0.500023</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>0.508978</td>\n",
       "      <td>0.500617</td>\n",
       "      <td>0.499851</td>\n",
       "      <td>0.498426</td>\n",
       "      <td>0.500147</td>\n",
       "      <td>0.483343</td>\n",
       "      <td>0.698139</td>\n",
       "      <td>0.436643</td>\n",
       "      <td>0.500304</td>\n",
       "      <td>...</td>\n",
       "      <td>0.497177</td>\n",
       "      <td>0.501450</td>\n",
       "      <td>0.499915</td>\n",
       "      <td>0.514723</td>\n",
       "      <td>0.499458</td>\n",
       "      <td>0.499882</td>\n",
       "      <td>0.499601</td>\n",
       "      <td>0.500818</td>\n",
       "      <td>0.500023</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>0.509955</td>\n",
       "      <td>0.500635</td>\n",
       "      <td>0.499829</td>\n",
       "      <td>0.498217</td>\n",
       "      <td>0.500160</td>\n",
       "      <td>0.483222</td>\n",
       "      <td>0.821644</td>\n",
       "      <td>0.440179</td>\n",
       "      <td>0.500309</td>\n",
       "      <td>...</td>\n",
       "      <td>0.496646</td>\n",
       "      <td>0.501524</td>\n",
       "      <td>0.499929</td>\n",
       "      <td>0.515409</td>\n",
       "      <td>0.499468</td>\n",
       "      <td>0.499862</td>\n",
       "      <td>0.499612</td>\n",
       "      <td>0.500939</td>\n",
       "      <td>0.500021</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>0.509825</td>\n",
       "      <td>0.500632</td>\n",
       "      <td>0.499832</td>\n",
       "      <td>0.498315</td>\n",
       "      <td>0.500156</td>\n",
       "      <td>0.483408</td>\n",
       "      <td>0.801483</td>\n",
       "      <td>0.439780</td>\n",
       "      <td>0.500306</td>\n",
       "      <td>...</td>\n",
       "      <td>0.496682</td>\n",
       "      <td>0.501507</td>\n",
       "      <td>0.499928</td>\n",
       "      <td>0.514020</td>\n",
       "      <td>0.499468</td>\n",
       "      <td>0.499866</td>\n",
       "      <td>0.499603</td>\n",
       "      <td>0.500910</td>\n",
       "      <td>0.500021</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 29 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0         0         1         2         3         4         5  \\\n",
       "0           0  0.509387  0.500629  0.499831  0.498434  0.500153  0.483391   \n",
       "1           1  0.509023  0.500618  0.499850  0.498435  0.500148  0.483310   \n",
       "2           2  0.508978  0.500617  0.499851  0.498426  0.500147  0.483343   \n",
       "3           3  0.509955  0.500635  0.499829  0.498217  0.500160  0.483222   \n",
       "4           4  0.509825  0.500632  0.499832  0.498315  0.500156  0.483408   \n",
       "\n",
       "          6         7         8  ...          18        19        20  \\\n",
       "0  0.788593  0.439439  0.500314  ...    0.496945  0.501489  0.499929   \n",
       "1  0.631625  0.436612  0.500304  ...    0.497171  0.501455  0.499915   \n",
       "2  0.698139  0.436643  0.500304  ...    0.497177  0.501450  0.499915   \n",
       "3  0.821644  0.440179  0.500309  ...    0.496646  0.501524  0.499929   \n",
       "4  0.801483  0.439780  0.500306  ...    0.496682  0.501507  0.499928   \n",
       "\n",
       "         21        22        23        24        25        26  Truth  \n",
       "0  0.513762  0.499473  0.499875  0.499608  0.500897  0.500032      1  \n",
       "1  0.514782  0.499460  0.499882  0.499597  0.500825  0.500023      1  \n",
       "2  0.514723  0.499458  0.499882  0.499601  0.500818  0.500023      1  \n",
       "3  0.515409  0.499468  0.499862  0.499612  0.500939  0.500021      1  \n",
       "4  0.514020  0.499468  0.499866  0.499603  0.500910  0.500021      1  \n",
       "\n",
       "[5 rows x 29 columns]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "df = pd.read_csv('cnn_out_20_blind.csv')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 5 candidates, totalling 15 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=4)]: Using backend LokyBackend with 4 concurrent workers.\n",
      "[Parallel(n_jobs=4)]: Done  15 out of  15 | elapsed:    3.7s finished\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "RandomizedSearchCV(cv=<generator object _BaseKFold.split at 0x0000013D48F93D58>,\n",
       "          error_score='raise-deprecating',\n",
       "          estimator=XGBClassifier(base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
       "       colsample_bytree=1, gamma=0, learning_rate=0.02, max_delta_step=0,\n",
       "       max_depth=3, min_child_weight=1, missing=None, n_estimators=600,\n",
       "       n_jobs=1, nthread=1, objective='binary:logistic', random_state=0,\n",
       "       reg_alpha=0, reg_lambda=1, scale_pos_weight=1, seed=None,\n",
       "       silent=True, subsample=1),\n",
       "          fit_params=None, iid='warn', n_iter=5, n_jobs=4,\n",
       "          param_distributions={'min_child_weight': [1, 5, 10], 'gamma': [0.5, 1, 1.5, 2, 5], 'subsample': [0.6, 0.7, 0.8, 0.9, 1.0], 'colsample_bytree': [0.71, 0.72, 0.73, 0.74, 0.75, 0.76, 0.78, 0.79, 0.8], 'max_depth': [3, 4, 5]},\n",
       "          pre_dispatch='2*n_jobs', random_state=1001, refit=True,\n",
       "          return_train_score='warn', scoring='roc_auc', verbose=3)"
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import xgboost as xgb\n",
    "from xgboost import XGBClassifier\n",
    "from sklearn.model_selection import GridSearchCV, train_test_split, StratifiedKFold, RandomizedSearchCV\n",
    "\n",
    "y = df[\"Truth\"]\n",
    "x = df.drop('Truth', axis=1)\n",
    "X_train, X_test, y_train, y_test = train_test_split(x, y, test_size=0.33, random_state=42)\n",
    "xgb_model = xgb.XGBClassifier()\n",
    "\n",
    "parameters = {\n",
    "        'min_child_weight': [1, 5, 10],\n",
    "        'gamma': [0.5, 1, 1.5, 2, 5],\n",
    "        'subsample': [0.6, .7, 0.8, .9, 1.0],\n",
    "        'colsample_bytree': [.71, .72, .73, .74, .75, .76, .78, .79, .8],\n",
    "        'max_depth': [3, 4, 5]\n",
    "        }\n",
    "\n",
    "xgb = XGBClassifier(learning_rate=0.02, n_estimators=600, objective='binary:logistic',\n",
    "                    silent=True, nthread=1)\n",
    "\n",
    "folds = 3\n",
    "param_comb = 5\n",
    "\n",
    "skf = StratifiedKFold(n_splits=folds, shuffle = True, random_state = 1001)\n",
    "\n",
    "random_search = RandomizedSearchCV(xgb, param_distributions=parameters, n_iter=param_comb, scoring='roc_auc', n_jobs=4, cv=skf.split(x,y), verbose=3, random_state=1001 )\n",
    "random_search.fit(x, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model with rank: 1\n",
      "Mean validation score: 1.000 (std: 0.000)\n",
      "Parameters: {'subsample': 0.8, 'min_child_weight': 5, 'max_depth': 4, 'gamma': 5, 'colsample_bytree': 0.74}\n",
      "\n",
      "Model with rank: 1\n",
      "Mean validation score: 1.000 (std: 0.000)\n",
      "Parameters: {'subsample': 0.6, 'min_child_weight': 5, 'max_depth': 3, 'gamma': 0.5, 'colsample_bytree': 0.75}\n",
      "\n",
      "Model with rank: 1\n",
      "Mean validation score: 1.000 (std: 0.000)\n",
      "Parameters: {'subsample': 0.7, 'min_child_weight': 1, 'max_depth': 4, 'gamma': 1, 'colsample_bytree': 0.72}\n",
      "\n",
      "Model with rank: 1\n",
      "Mean validation score: 1.000 (std: 0.000)\n",
      "Parameters: {'subsample': 0.6, 'min_child_weight': 5, 'max_depth': 4, 'gamma': 5, 'colsample_bytree': 0.78}\n",
      "\n"
     ]
    }
   ],
   "source": [
    "report_best_scores(random_search.cv_results_, 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_auc_score\n",
    "import pickle\n",
    "\n",
    "xgb_model = XGBClassifier(objective='binary:logistic', random_state=42, eval_metric='auc',\n",
    "                             subsample=.8, min_child_weight=5, max_depth=4, gamma=5, colsample_bytree=.74)\n",
    "\n",
    "xgb_model.fit(x,y)\n",
    "pickle.dump(xgb_model, open(\"xgb_model_XX_blind.pickle.dat\", \"wb\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Getting predictions for NN0\n",
      "Getting predictions for NN1\n",
      "Getting predictions for NN2\n",
      "Getting predictions for NN3\n",
      "Getting predictions for NN4\n",
      "Getting predictions for NN5\n",
      "Getting predictions for NN6\n",
      "Getting predictions for NN7\n",
      "Getting predictions for NN8\n",
      "Getting predictions for NN9\n",
      "Getting predictions for NN10\n",
      "Getting predictions for NN11\n",
      "Getting predictions for NN12\n",
      "Getting predictions for NN13\n",
      "Getting predictions for NN14\n",
      "Getting predictions for NN15\n",
      "Getting predictions for NN16\n",
      "Getting predictions for NN17\n",
      "Getting predictions for NN18\n",
      "Getting predictions for NN19\n",
      "Getting predictions for NN20\n",
      "Getting predictions for NN21\n",
      "Getting predictions for NN22\n",
      "Getting predictions for NN23\n",
      "Getting predictions for NN24\n",
      "Getting predictions for NN25\n",
      "Getting predictions for NN26\n"
     ]
    }
   ],
   "source": [
    "#Predict test set\n",
    "import pandas as pd\n",
    "\n",
    "df = pd.DataFrame()\n",
    "df.to_csv(\"cnn_out_1_pred_blind.csv\")\n",
    "\n",
    "#all_test_IDs = test_IDs['AD'] + test_IDs['NC']\n",
    "labels = []\n",
    "for neural_net_idx in range(27):\n",
    "    vector = []\n",
    "    print('Getting predictions for NN' + str(neural_net_idx))\n",
    "    model = load_model(os.getcwd() + \"\\\\model_weights\\\\saved_weights_net_\" + str(neural_net_idx) + \"_epoch_17.h5\")\n",
    "    for nii_path in test_IDs:\n",
    "        \n",
    "        x, y = get_sampleXY(nii_path, patch_idx=neural_net_idx)\n",
    "        yhat = model.predict(x=x)\n",
    "        #print(yhat)\n",
    "        vector.append(yhat[0][0])\n",
    "        #print(vector)\n",
    "        if(neural_net_idx == 0 ):\n",
    "            labels.append(y[0][0])\n",
    "            \n",
    "    df[str(neural_net_idx)] = vector\n",
    "    #print(df)\n",
    "    \n",
    "df[\"Truth\"] = labels\n",
    "df.to_csv(\"cnn_out_18_pred_blind.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.678649237472767"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pickle\n",
    "from sklearn.metrics import roc_auc_score\n",
    "\n",
    "df = pd.read_csv('cnn_out_19_pred_blind.csv')\n",
    "xgb_model = pickle.load(open(\"xgb_model.pickle.dat\", \"rb\"))\n",
    "\n",
    "y = df[\"Truth\"]\n",
    "x = df.drop('Truth', axis=1)\n",
    "\n",
    "yhat = xgb_model.predict_proba(x)[:, 1]\n",
    "\n",
    "roc_auc_score(y, yhat)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 5 epochs = .5 ROC\n",
    "# 10 epochs = .59 ROC\n",
    "# 15 epochs = .5 ROC\n",
    "\n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import roc_auc_score\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.patches as mpatches\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "\n",
    "df = pd.read_csv('cnn_out_19_pred_blind.csv')\n",
    "y = df[\"Truth\"]\n",
    "x = df.drop('Truth', axis=1)\n",
    "yhat = xgb_model.predict_proba(x)[:, 1]\n",
    "fpr, tpr, thresholds = metrics.roc_curve(y, yhat, pos_label=1)\n",
    "plt.plot(fpr, tpr, 'm', label='10 Epochs')\n",
    "#plt.title(\"FPR vs TPR\\nAUC=.6536\")\n",
    "fig2 = plt.gcf()\n",
    "plt.draw()\n",
    "\n",
    "df = pd.read_csv('cnn_out_20_pred_blind.csv')\n",
    "y = df[\"Truth\"]\n",
    "x = df.drop('Truth', axis=1)\n",
    "yhat = xgb_model.predict_proba(x)[:, 1]\n",
    "fpr, tpr, thresholds = metrics.roc_curve(y, yhat, pos_label=1)\n",
    "plt.plot(fpr, tpr, 'b', label='11 Epochs')\n",
    "#plt.title(\"FPR vs TPR\\nAUC=.6754\")\n",
    "plt.title('TPR vs FPR')\n",
    "fig3 = plt.gcf()\n",
    "plt.draw()\n",
    "\n",
    "df = pd.read_csv('cnn_out_21_pred_blind.csv')\n",
    "y = df[\"Truth\"]\n",
    "x = df.drop('Truth', axis=1)\n",
    "yhat = xgb_model.predict_proba(x)[:, 1]\n",
    "fpr, tpr, thresholds = metrics.roc_curve(y, yhat, pos_label=1)\n",
    "plt.plot(fpr, tpr, 'g', label='15 Epochs')\n",
    "#plt.title(\"FPR vs TPR\\nAUC=.6754\")\n",
    "plt.title('TPR vs FPR')\n",
    "fig3 = plt.gcf()\n",
    "plt.draw()\n",
    "\n",
    "df = pd.read_csv('cnn_out_22_pred_blind.csv')\n",
    "y = df[\"Truth\"]\n",
    "x = df.drop('Truth', axis=1)\n",
    "yhat = xgb_model.predict_proba(x)[:, 1]\n",
    "fpr, tpr, thresholds = metrics.roc_curve(y, yhat, pos_label=1)\n",
    "plt.plot(fpr, tpr, 'r', label='15 Epochs')\n",
    "#plt.title(\"FPR vs TPR\\nAUC=.6754\")\n",
    "plt.title('TPR vs FPR')\n",
    "fig3 = plt.gcf()\n",
    "plt.draw()\n",
    "\n",
    "mag_patch = mpatches.Patch(color='magenta', label='19 Epochs - AUC score : .6786')\n",
    "blue_patch = mpatches.Patch(color='blue', label='20 Epochs - AUC score : .7777')\n",
    "green_patch = mpatches.Patch(color='green', label='21 Epochs - AUC score : .8192')\n",
    "red_patch = mpatches.Patch(color='red', label='22 Epochs - AUC score : .6601')\n",
    "\n",
    "plt.legend(handles=[mag_patch, blue_patch, green_patch, red_patch])\n",
    "\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.show()\n",
    "\n",
    "#fig1.savefig('ROC_8.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 5 epochs = .5 ROC\n",
    "# 10 epochs = .59 ROC\n",
    "# 15 epochs = .5 ROC\n",
    "\n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import roc_auc_score\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.patches as mpatches\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "\n",
    "df = pd.read_csv('cnn_out_19_pred_blind.csv')\n",
    "y = df[\"Truth\"]\n",
    "x = df.drop('Truth', axis=1)\n",
    "yhat = xgb_model.predict_proba(x)[:, 1]\n",
    "tpr, fpr, thresholds = metrics.precision_recall_curve(y, yhat, pos_label=1)\n",
    "plt.plot(fpr,tpr, 'm', label='9 Epochs')\n",
    "#plt.title(\"TPR vs FPR\\nAUC = 0.6471\")\n",
    "fig1 = plt.gcf()\n",
    "plt.draw()\n",
    "\n",
    "df = pd.read_csv('cnn_out_20_pred_blind.csv')\n",
    "y = df[\"Truth\"]\n",
    "x = df.drop('Truth', axis=1)\n",
    "yhat = xgb_model.predict_proba(x)[:, 1]\n",
    "tpr, fpr, thresholds = metrics.precision_recall_curve(y, yhat, pos_label=1)\n",
    "plt.plot(fpr, tpr, 'b', label='10 Epochs')\n",
    "#plt.title(\"FPR vs TPR\\nAUC=.6536\")\n",
    "fig2 = plt.gcf()\n",
    "plt.draw()\n",
    "\n",
    "df = pd.read_csv('cnn_out_21_pred_blind.csv')\n",
    "y = df[\"Truth\"]\n",
    "x = df.drop('Truth', axis=1)\n",
    "yhat = xgb_model.predict_proba(x)[:, 1]\n",
    "tpr, fpr, thresholds = metrics.precision_recall_curve(y, yhat, pos_label=1)\n",
    "plt.plot(fpr, tpr, 'g', label='11 Epochs')\n",
    "#plt.title(\"FPR vs TPR\\nAUC=.6754\")\n",
    "plt.title('TPR vs FPR')\n",
    "fig3 = plt.gcf()\n",
    "plt.draw()\n",
    "\n",
    "df = pd.read_csv('cnn_out_22_pred_blind.csv')\n",
    "y = df[\"Truth\"]\n",
    "x = df.drop('Truth', axis=1)\n",
    "yhat = xgb_model.predict_proba(x)[:, 1]\n",
    "tpr, fpr, thresholds = metrics.precision_recall_curve(y, yhat, pos_label=1)\n",
    "plt.plot(fpr, tpr, 'r', label='15 Epochs')\n",
    "#plt.title(\"FPR vs TPR\\nAUC=.6754\")\n",
    "plt.title('Precision vs Recall')\n",
    "fig3 = plt.gcf()\n",
    "plt.draw()\n",
    "\n",
    "mag_patch = mpatches.Patch(color='magenta', label='19 Epochs')\n",
    "blue_patch = mpatches.Patch(color='blue', label='20 Epochs')\n",
    "green_patch = mpatches.Patch(color='green', label='21 Epochs')\n",
    "red_patch = mpatches.Patch(color='red', label='22 Epochs')\n",
    "\n",
    "plt.xlim((0,1))\n",
    "plt.xlabel('Recall')\n",
    "plt.ylabel('Precision')\n",
    "plt.legend(handles=[mag_patch, blue_patch, green_patch, red_patch])\n",
    "\n",
    "plt.show()\n",
    "\n",
    "#fig1.savefig('ROC_8.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "x = [18, 19, 20, 21, 22, 23]\n",
    "y = [.4804, .6786, .7777, .8192, .6601, .6721]\n",
    "\n",
    "plt.plot(x, y, 'b')\n",
    "plt.xlabel('Number of Epochs')\n",
    "plt.ylabel('Testing AUC Score')\n",
    "plt.title('Learning Curve')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}