main_train_test.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "# -*- coding: utf-8 -*-\n",
    "import sys, os\n",
    "import cv2\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import json\n",
    "import gc\n",
    "\n",
    "def load_Img(imgDir,imgFoldName,subFold):\n",
    "    imgs = os.listdir(imgDir+\"/\"+imgFoldName)\n",
    "    labels = os.listdir(imgDir+\"/\"+subFold)\n",
    "    imgNum = len(imgs)\n",
    "    labelNum = len(labels)\n",
    "    #input size is 512 * 512 * 1\n",
    "\n",
    "    data = np.empty((imgNum * 2,512,512,1))\n",
    "    label = np.empty((imgNum * 2,512,512,1))\n",
    "    if labelNum == imgNum:\n",
    "        print (imgDir + ' imgNum: ',imgNum)\n",
    "        for i in range (imgNum):\n",
    "            img = cv2.imread(imgDir+\"/\"+imgFoldName+\"/\"+imgs[i])\n",
    "            img = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)\n",
    "            img0 = img / 255            \n",
    "            arr_0 = cv2.flip(img0,1)         \n",
    "            \n",
    "            data[i,:,:,0] = img0\n",
    "            data[i + imgNum,:,:,0] = arr_0         \n",
    "\n",
    "            img = cv2.imread(imgDir+\"/\"+ subFold +\"/\"+labels[i])\n",
    "            img0 = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)\n",
    "            arr_0 = cv2.flip(img0,1)                            \n",
    "            label[i,:,:,0] = img0\n",
    "            label[i + imgNum,:,:,0] = arr_0\n",
    "            gc.collect()\n",
    "            \n",
    "    return data,label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from sklearn.cross_validation import train_test_split\n",
    "from keras.models import *\n",
    "from keras.layers import Input, merge, Conv2D, MaxPooling2D, UpSampling2D, Dropout, Cropping2D, BatchNormalization,concatenate\n",
    "from keras.optimizers import *\n",
    "from keras.callbacks import ModelCheckpoint, LearningRateScheduler\n",
    "from keras.callbacks import ModelCheckpoint, Callback, EarlyStopping\n",
    "from keras.callbacks import TensorBoard\n",
    "from keras import backend as K\n",
    "\n",
    "def dice_coef(y_true, y_pred):\n",
    "    smooth = 1\n",
    "    y_true_f = K.flatten(y_true)\n",
    "    y_pred_f = K.flatten(y_pred)\n",
    "    intersection = K.sum(y_true_f * y_pred_f)\n",
    "    return (2. * intersection + smooth ) / (K.sum(y_true_f) + K.sum(y_pred_f) + smooth)\n",
    "def dice_coef_loss(y_true, y_pred):\n",
    "    return - dice_coef(y_true, y_pred)\n",
    "\n",
    "\n",
    "def get_unet():\n",
    "        inputs = Input((512, 512, 1))\n",
    "\n",
    "        conv1 = Conv2D(64, 3, activation='relu', padding='same', kernel_initializer='he_normal')(inputs)\n",
    "        conv1 = Conv2D(64, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv1)\n",
    "        pool1 = MaxPooling2D(pool_size=(2, 2))(conv1)\n",
    "\n",
    "        conv2 = Conv2D(128, 3, activation='relu', padding='same', kernel_initializer='he_normal')(pool1)\n",
    "        conv2 = Conv2D(128, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv2)\n",
    "        pool2 = MaxPooling2D(pool_size=(2, 2))(conv2)\n",
    "\n",
    "        conv3 = Conv2D(256, 3, activation='relu', padding='same', kernel_initializer='he_normal')(pool2)\n",
    "        conv3 = Conv2D(256, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv3)\n",
    "        pool3 = MaxPooling2D(pool_size=(2, 2))(conv3)\n",
    "\n",
    "        conv4 = Conv2D(512, 3, activation='relu', padding='same', kernel_initializer='he_normal')(pool3)\n",
    "        conv4 = Conv2D(512, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv4)\n",
    "        drop4 = Dropout(0.5)(conv4)\n",
    "        pool4 = MaxPooling2D(pool_size=(2, 2))(drop4)\n",
    "\n",
    "        conv5 = Conv2D(1024, 3, activation='relu', padding='same', kernel_initializer='he_normal')(pool4)\n",
    "        conv5 = Conv2D(1024, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv5)\n",
    "        drop5 = Dropout(0.5)(conv5)\n",
    "\n",
    "        up6 = Conv2D(512, 2, activation='relu', padding='same', kernel_initializer='he_normal')(UpSampling2D(size=(2, 2))(drop5))\n",
    "        merge6 = concatenate([drop4, up6], axis=3)\n",
    "        conv6 = Conv2D(512, 3, activation='relu', padding='same', kernel_initializer='he_normal')(merge6)\n",
    "        conv6 = Conv2D(512, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv6)\n",
    "\n",
    "\n",
    "        up7 = Conv2D(256, 2, activation='relu', padding='same', kernel_initializer='he_normal')(UpSampling2D(size=(2, 2))(conv6))\n",
    "        merge7 = concatenate([conv3, up7], axis=3)\n",
    "        conv7 = Conv2D(256, 3, activation='relu', padding='same', kernel_initializer='he_normal')(merge7)\n",
    "        conv7 = Conv2D(256, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv7)\n",
    "\n",
    "        up8 = Conv2D(128, 2, activation='relu', padding='same', kernel_initializer='he_normal')(UpSampling2D(size=(2, 2))(conv7))\n",
    "        merge8 = concatenate([conv2, up8], axis=3)\n",
    "        conv8 = Conv2D(128, 3, activation='relu', padding='same', kernel_initializer='he_normal')(merge8)\n",
    "        conv8 = Conv2D(128, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv8)\n",
    "\n",
    "        up9 = Conv2D(64, 2, activation='relu', padding='same', kernel_initializer='he_normal')(UpSampling2D(size=(2, 2))(conv8))\n",
    "        merge9 = concatenate([conv1, up9], axis=3)\n",
    "        conv9 = Conv2D(64, 3, activation='relu', padding='same', kernel_initializer='he_normal')(merge9)\n",
    "        conv9 = Conv2D(64, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv9)\n",
    "        conv9 = Conv2D(2, 3, activation='relu', padding='same', kernel_initializer='he_normal')(conv9)\n",
    "        conv10 = Conv2D(1, 1, activation='sigmoid')(conv9)\n",
    "\n",
    "        model = Model(input=inputs, output=conv10)\n",
    "\n",
    "        model.compile(optimizer=Adam(lr=1e-4), loss=dice_coef_loss, metrics=[dice_coef])\n",
    "        print('model compile')\n",
    "#         model.summary()\n",
    "        return model\n",
    "#     binary_crossentropy\n",
    "    \n",
    "def get_callbacks(filepath,part,patience=2):\n",
    "    es = EarlyStopping('val_loss', patience=patience, mode=\"min\")\n",
    "    msave = ModelCheckpoint(filepath, save_best_only=True)\n",
    "    tb_log = TensorBoard(log_dir='./log/' + part)\n",
    "    return [es, msave,tb_log]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "liver imgNum:  2064\n",
      "(4128, 512, 512, 1)\n"
     ]
    }
   ],
   "source": [
    "if __name__ == '__main__':\n",
    "    liver_data, liver_label =load_Img('liver','img','cls')\n",
    "    print (liver_data.shape)\n",
    "#     spleen_data, spleen_label =load_Img('spleen','img','cls')\n",
    "#     print (spleen_data.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "    x_train,x_test,y_train,y_test = train_test_split(liver_data, liver_label, random_state=0, train_size=0.4)\n",
    "#     del liver_data\n",
    "#     del liver_label\n",
    "#     gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model compile\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_2 (InputLayer)            (None, 512, 512, 1)  0                                            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_25 (Conv2D)              (None, 512, 512, 64) 640         input_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_26 (Conv2D)              (None, 512, 512, 64) 36928       conv2d_25[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_5 (MaxPooling2D)  (None, 256, 256, 64) 0           conv2d_26[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_27 (Conv2D)              (None, 256, 256, 128 73856       max_pooling2d_5[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_28 (Conv2D)              (None, 256, 256, 128 147584      conv2d_27[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_6 (MaxPooling2D)  (None, 128, 128, 128 0           conv2d_28[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_29 (Conv2D)              (None, 128, 128, 256 295168      max_pooling2d_6[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_30 (Conv2D)              (None, 128, 128, 256 590080      conv2d_29[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_7 (MaxPooling2D)  (None, 64, 64, 256)  0           conv2d_30[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_31 (Conv2D)              (None, 64, 64, 512)  1180160     max_pooling2d_7[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_32 (Conv2D)              (None, 64, 64, 512)  2359808     conv2d_31[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_3 (Dropout)             (None, 64, 64, 512)  0           conv2d_32[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_8 (MaxPooling2D)  (None, 32, 32, 512)  0           dropout_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_33 (Conv2D)              (None, 32, 32, 1024) 4719616     max_pooling2d_8[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_34 (Conv2D)              (None, 32, 32, 1024) 9438208     conv2d_33[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_4 (Dropout)             (None, 32, 32, 1024) 0           conv2d_34[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "up_sampling2d_5 (UpSampling2D)  (None, 64, 64, 1024) 0           dropout_4[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_35 (Conv2D)              (None, 64, 64, 512)  2097664     up_sampling2d_5[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_5 (Concatenate)     (None, 64, 64, 1024) 0           dropout_3[0][0]                  \n",
      "                                                                 conv2d_35[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_36 (Conv2D)              (None, 64, 64, 512)  4719104     concatenate_5[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_37 (Conv2D)              (None, 64, 64, 512)  2359808     conv2d_36[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "up_sampling2d_6 (UpSampling2D)  (None, 128, 128, 512 0           conv2d_37[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_38 (Conv2D)              (None, 128, 128, 256 524544      up_sampling2d_6[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_6 (Concatenate)     (None, 128, 128, 512 0           conv2d_30[0][0]                  \n",
      "                                                                 conv2d_38[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_39 (Conv2D)              (None, 128, 128, 256 1179904     concatenate_6[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_40 (Conv2D)              (None, 128, 128, 256 590080      conv2d_39[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "up_sampling2d_7 (UpSampling2D)  (None, 256, 256, 256 0           conv2d_40[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_41 (Conv2D)              (None, 256, 256, 128 131200      up_sampling2d_7[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_7 (Concatenate)     (None, 256, 256, 256 0           conv2d_28[0][0]                  \n",
      "                                                                 conv2d_41[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_42 (Conv2D)              (None, 256, 256, 128 295040      concatenate_7[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_43 (Conv2D)              (None, 256, 256, 128 147584      conv2d_42[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "up_sampling2d_8 (UpSampling2D)  (None, 512, 512, 128 0           conv2d_43[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_44 (Conv2D)              (None, 512, 512, 64) 32832       up_sampling2d_8[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_8 (Concatenate)     (None, 512, 512, 128 0           conv2d_26[0][0]                  \n",
      "                                                                 conv2d_44[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_45 (Conv2D)              (None, 512, 512, 64) 73792       concatenate_8[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_46 (Conv2D)              (None, 512, 512, 64) 36928       conv2d_45[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_47 (Conv2D)              (None, 512, 512, 2)  1154        conv2d_46[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_48 (Conv2D)              (None, 512, 512, 1)  3           conv2d_47[0][0]                  \n",
      "==================================================================================================\n",
      "Total params: 31,031,685\n",
      "Trainable params: 31,031,685\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n",
      "Epoch 1/7"
     ]
    }
   ],
   "source": [
    "    def data_generator(data, targets, batch_size):       \n",
    "        idx = np.arange(data.shape[0])\n",
    "        np.random.shuffle(idx)\n",
    "        while True:\n",
    "            for i in range(int(data.shape[0]/batch_size+1)):\n",
    "                yield data[i * batch_size:(i + 1) * batch_size], targets[i * batch_size:(i + 1) * batch_size]    \n",
    "    model = get_unet()\n",
    "    model.load_weights('liver2_8_unet.hdf5',by_name = False)\n",
    "    file_path = \"liver_dice_8_unet.hdf5\"\n",
    "    callbacks_s = get_callbacks(file_path,'liver',patience=5)\n",
    "    batch_size = 256\n",
    "    model.fit_generator(data_generator(x_train, y_train,batch_size), steps_per_epoch = 4128 * 0.6//8,epochs=7, verbose=1,\n",
    "            validation_data = (x_test, y_test),validation_steps =4128 * 0.4//8 ,callbacks = callbacks_s)\n",
    "    \n",
    "    batch_size = 128\n",
    "    with tf.device('/gpu:0'):\n",
    "        model.fit_generator(data_generator( x_train, y_train,batch_size), steps_per_epoch = (y_train.shape[0] + batch_size - 1) // batch_size,\n",
    "                            nb_epoch=2,verbose=1,callbacks = callbacks_s,validation_data = data_generator( x_test, y_test,batch_size))\n",
    "    #########################################################################################################"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model compile\n",
      "Train on 2889 samples, validate on 1239 samples\n",
      "Epoch 1/7"
     ]
    }
   ],
   "source": [
    "    model = get_unet()\n",
    "    model.load_weights('liver2_8_unet.hdf5',by_name = False)\n",
    "    file_path = \"liver_dice_8_unet.hdf5\"\n",
    "    callbacks_s = get_callbacks(file_path,'liver', patience=5)\n",
    "    model.fit(liver_data,liver_label, batch_size=8, epochs=7, verbose=1,validation_split=0.3, shuffle=True,callbacks = callbacks_s)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "    json_string = model2.to_json()\n",
    "    with open('./liver_model_architecture.json','w') as json_file:\n",
    "        json_file.write(json_string)"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "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": "3.5.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}