-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathclassification.py
426 lines (332 loc) · 17.1 KB
/
classification.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
#######################Loading the libararies ####################################
import numpy as np
import os
import cv2
import tensorflow as tf
########GPU Configuration#############
os.environ["CUDA_VISIBLE_DEVICES"]="0"
from keras import backend as K
config=tf.ConfigProto()
config.gpu_options.allow_growth=True
config.gpu_options.per_process_gpu_memory_fraction=0.98
K.tensorflow_backend.set_session(tf.Session(config=config))
######Keras Utilities #################
from keras.models import Sequential, Model
from keras.layers import Conv2D, ZeroPadding2D, MaxPooling2D, GlobalMaxPooling2D, Input
from keras.layers.core import Flatten, Dense, Dropout, Lambda
from keras import applications
from keras.utils import np_utils
from time import time
from sklearn.metrics import roc_auc_score,roc_curve
from sklearn.metrics import roc_curve
#import scikitplot as skplt
import matplotlib.pyplot as plt
import math
###Keras
import keras
from keras.utils import np_utils
from keras.preprocessing import image
from keras.applications.imagenet_utils import preprocess_input
from keras.layers import Dense, Activation, Flatten
from keras.layers import merge, Input
from keras.layers import MaxPooling2D
from keras.models import Model
from sklearn.utils import shuffle
#from sklearn.cross_validation import train_test_split
from keras.optimizers import SGD
from keras.preprocessing.image import ImageDataGenerator
from keras.callbacks import TensorBoard
###Sklearn
from sklearn.preprocessing import StandardScaler
from sklearn.metrics import roc_curve
from sklearn.metrics import auc
from sklearn.externals import joblib
from sklearn.metrics import precision_recall_fscore_support
from sklearn.metrics import precision_recall_curve
from sklearn.metrics import accuracy_score
from sklearn.metrics import roc_auc_score
from sklearn.utils import shuffle
from sklearn.metrics import roc_curve
from sklearn.metrics import roc_auc_score
from sklearn.externals import joblib
import pandas as pd
from sklearn import metrics
from scipy.stats import zscore
from sklearn.model_selection import KFold
from keras.layers import GlobalMaxPooling2D
from keras import optimizers
from keras.utils import np_utils
from keras.callbacks import Callback, EarlyStopping
from keras.callbacks import ModelCheckpoint
from sklearn.metrics import confusion_matrix, f1_score, precision_score, recall_score
from keras.metrics import categorical_accuracy
############################### User Input #################################
Number_of_folds=4
batch_size=32
epochs=50
Disease_name='Resnet50_Edena_Binary_CL200withAgumentation2'
num_classes=2
training_fold_1_name='/Local/cl2019/mayClassification/5_16_2019_BinaryEdemaclassclassification_hu200/excels/PNGLung_CV-Training-Fold-1.csv'
validation_fold_1_name='/Local/cl2019/mayClassification/5_16_2019_BinaryEdemaclassclassification_hu200/excels/PNGLung_CV-Validation-Fold-1.csv'
training_fold_2_name='/Local/cl2019/mayClassification/5_16_2019_BinaryEdemaclassclassification_hu200/excels/PNGLung_CV-Training-Fold-2.csv'
validation_fold_2_name='/Local/cl2019/mayClassification/5_16_2019_BinaryEdemaclassclassification_hu200/excels/PNGLung_CV-Validation-Fold-2.csv'
training_fold_3_name='/Local/cl2019/mayClassification/5_16_2019_BinaryEdemaclassclassification_hu200/excels/PNGLung_CV-Training-Fold-3.csv'
validation_fold_3_name='/Local/cl2019/mayClassification/5_16_2019_BinaryEdemaclassclassification_hu200/excels/PNGLung_CV-Validation-Fold-3.csv'
training_fold_4_name='/Local/cl2019/mayClassification/5_16_2019_BinaryEdemaclassclassification_hu200/excels/PNGLung_CV-Training-Fold-4.csv'
validation_fold_4_name='/Local/cl2019/mayClassification/5_16_2019_BinaryEdemaclassclassification_hu200/excels/PNGLung_CV-Validation-Fold-4.csv'
########################################################################
####################### Coder Info######################################
print('Code is written by @Fakrul Islam Tushar,4/10/2019')
print('Dept. of Radiology, RAI LABS, Duke University')
print('For any confusion Mail: [email protected]\n')
print('Data is loaded from the Folder')
########################################################################
###############Functions######################
#This function Compute Run time
def auc_roc(y_true, y_pred):
# any tensorflow metric
value, update_op = tf.contrib.metrics.streaming_auc(y_pred, y_true)
# find all variables created for this metric
metric_vars = [i for i in tf.local_variables() if 'auc_roc' in i.name.split('/')[1]]
# Add metric variables to GLOBAL_VARIABLES collection.
# They will be initialized for new session.
for v in metric_vars:
tf.add_to_collection(tf.GraphKeys.GLOBAL_VARIABLES, v)
# force to update metric values
with tf.control_dependencies([update_op]):
value = tf.identity(value)
return value
def f1_score(y_true, y_pred):
# any tensorflow metric
value, update_op = tf.contrib.metrics.f1_score(y_true,y_pred)
# find all variables created for this metric
metric_vars = [i for i in tf.local_variables() if 'f1_score' in i.name.split('/')[1]]
# Add metric variables to GLOBAL_VARIABLES collection.
# They will be initialized for new session.
for v in metric_vars:
tf.add_to_collection(tf.GraphKeys.GLOBAL_VARIABLES, v)
# force to update metric values
with tf.control_dependencies([update_op]):
value = tf.identity(value)
return value
def get_model(net,num_classes=2):
if net=='Resnet50':
image_input=Input(shape=(224,224,3))
model=keras.applications.resnet50.ResNet50(input_tensor=image_input,include_top=False,weights='imagenet')
last_layer = model.output
z= keras.layers.GlobalMaxPooling2D()(last_layer)
out = Dense(num_classes, activation='softmax', name='output_layer')(z)
custom_model = Model(inputs=image_input,outputs= out)
#Freezing the upper layers
#for layer in custom_model.layers[:-2]:
#layer.trainable=False
#sgd = optimizers.SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True)
custom_model.compile(loss='binary_crossentropy', optimizer='adam',metrics=['accuracy',auc_roc,f1_score])
return custom_model
elif net=='Vgg16':
image_input=Input(shape=(224,224,3))
model=keras.applications.vgg16.VGG16(input_tensor=image_input,include_top=False,weights='imagenet')
last_layer = model.output
z= keras.layers.GlobalMaxPooling2D()(last_layer)
out = Dense(num_classes, activation='softmax', name='output_layer')(z)
custom_model = Model(inputs=image_input,outputs= out)
custom_model.compile(loss='categorical_crossentropy', optimizer='adam',metrics=['accuracy',auc_roc])
return custom_model
elif net=='InceptionV3':
image_input=Input(shape=(224,224,3))
model=keras.applications.inception_v3.InceptionV3(input_tensor=image_input,include_top=False,weights='imagenet')
last_layer = model.output
z= keras.layers.GlobalMaxPooling2D()(last_layer)
out = Dense(num_classes, activation='softmax', name='output_layer')(z)
custom_model = Model(inputs=image_input,outputs= out)
#Freezing the upper layers
#for layer in custom_model.layers[:-1]:
#layer.trainable=False
custom_model.compile(loss='categorical_crossentropy', optimizer='adam',metrics=['categorical_accuracy',auc_roc])
return custom_model
elif net=='DenseNet':
image_input=Input(shape=(224,224,3))
model=keras.applications.densenet.DenseNet121(input_tensor=image_input,include_top=False,weights='imagenet')
last_layer = model.output
z= keras.layers.GlobalMaxPooling2D()(last_layer)
out = Dense(num_classes, activation='sigmoid', name='output_layer')(z)
custom_model = Model(inputs=image_input,outputs= out)
#Freezing the upper layers
for layer in custom_model.layers[:-2]:
layer.trainable=False
custom_model.compile(loss='categorical_crossentropy', optimizer='adam',metrics=['categorical_accuracy','acc'])
return custom_model
##########################CrossValidation############################
#This part will make the cross-validation splits and
#automatically save the Cross-validated Datas name,
#labels and Path to csv file
img_data_list=[]
Sample_Count_list=[] #How many samples in each folder
Image_name=[] # name of the images
Image_Path=[] # Path of the images
Save_Training_folds_x=[]
Save_Training_folds_y=[]
Save_Validation_folds_x=[]
Save_Validation_folds_y=[]
f1_train=pd.read_csv(training_fold_1_name,dtype=object,keep_default_na=False,na_values=[])
f2_train=pd.read_csv(training_fold_2_name,dtype=object,keep_default_na=False,na_values=[])
f3_train=pd.read_csv(training_fold_3_name,dtype=object,keep_default_na=False,na_values=[])
f4_train=pd.read_csv(training_fold_4_name,dtype=object,keep_default_na=False,na_values=[])
f1_val=pd.read_csv(validation_fold_1_name,dtype=object,keep_default_na=False,na_values=[])
f2_val=pd.read_csv(validation_fold_2_name,dtype=object,keep_default_na=False,na_values=[])
f3_val=pd.read_csv(validation_fold_3_name,dtype=object,keep_default_na=False,na_values=[])
f4_val=pd.read_csv(validation_fold_4_name,dtype=object,keep_default_na=False,na_values=[])
################Fold-1#####################
f1_train_x=f1_train['imgs']
f1_train_y=f1_train['lbls']
f1_val_x=f1_val['imgs']
f1_val_y=f1_val['lbls']
Save_Training_folds_x.append(f1_train_x)
Save_Training_folds_y.append(f1_train_y)
Save_Validation_folds_x.append(f1_val_x)
Save_Validation_folds_y.append(f1_val_y)
##################Fold-2##################
f2_train_x=f2_train['imgs']
f2_train_y=f2_train['lbls']
f2_val_x=f2_val['imgs']
f2_val_y=f2_val['lbls']
Save_Training_folds_x.append(f2_train_x)
Save_Training_folds_y.append(f2_train_y)
Save_Validation_folds_x.append(f2_val_x)
Save_Validation_folds_y.append(f2_val_y)
##################Fold-3##################
f3_train_x=f3_train['imgs']
f3_train_y=f3_train['lbls']
f3_val_x=f3_val['imgs']
f3_val_y=f3_val['lbls']
Save_Training_folds_x.append(f3_train_x)
Save_Training_folds_y.append(f3_train_y)
Save_Validation_folds_x.append(f3_val_x)
Save_Validation_folds_y.append(f3_val_y)
##################Fold-4##################
f4_train_x=f4_train['imgs']
f4_train_y=f4_train['lbls']
f4_val_x=f4_val['imgs']
f4_val_y=f4_val['lbls']
Save_Training_folds_x.append(f4_train_x)
Save_Training_folds_y.append(f4_train_y)
Save_Validation_folds_x.append(f4_val_x)
Save_Validation_folds_y.append(f4_val_y)
################################Excel is Made########################################
###########Now Reading the image and Labels From the Folds######################
for f in range(0,Number_of_folds):
cc=f+1
print('----------- Start Loading Fold-{} Data----------------'.format(cc))
loading_fold_training_images_path=Save_Training_folds_x[f]
loading_fold_training_label=Save_Training_folds_y[f]
loading_fold_Validation_images_path=Save_Validation_folds_x[f]
loading_fold_Validation_label=Save_Validation_folds_y[f]
#print(loading_fold_training_images_path,loading_fold_training_label)
TRAINING_X=[]
VALIDATION_X=[]
###fold Counter #########
#########################
for CT_TRAIN in loading_fold_training_images_path:
#print(CT_TRAIN)
img1 = cv2.imread(CT_TRAIN)
img1 = cv2.resize(img1,(224,224))
img1 = np.tile(img1,(1,1,1))
x=image.img_to_array(img1) #converting the image to array
x=np.expand_dims(x,axis=0) #putting them in the row axis
x=preprocess_input(x) #preprocessing
TRAINING_X.append(x)
traing_CT_img = np.array(TRAINING_X)
traing_CT_img =np.rollaxis(traing_CT_img,1,0)
traing_CT_img=traing_CT_img[0]
print(traing_CT_img.shape)
Number_of_Training_sample_in_this_fold=traing_CT_img.shape[0]
print('Total Number of Training Sample In Fold-{} is {}'.format(cc,Number_of_Training_sample_in_this_fold))
print('<----------Validation---------->"')
for CT_VAL in loading_fold_Validation_images_path:
#print(CT_VAL)
img1 = cv2.imread(CT_VAL)
img1 = cv2.resize(img1,(224,224))
img1 = np.tile(img1,(1,1,1))
x=image.img_to_array(img1) #converting the image to array
x=np.expand_dims(x,axis=0) #putting them in the row axis
x=preprocess_input(x) #preprocessing
VALIDATION_X.append(x)
Validation_CT_img = np.array(VALIDATION_X)
Validation_CT_img =np.rollaxis(Validation_CT_img,1,0)
Validation_CT_img=Validation_CT_img[0]
print(Validation_CT_img.shape)
Number_of_Validation_sample_in_this_fold=Validation_CT_img.shape[0]
print('Total Number of Validation Sample In Fold-{} is {}'.format(cc,Number_of_Validation_sample_in_this_fold))
print('#-----------Done Loading Fold-{} Data----------------#'.format(cc))
#########################Data Pre-Procesing ############################
#############################CNN Part Starts here ######################
#Image Generator
datagen = ImageDataGenerator(
rescale=1./255.0,
featurewise_center=True,
featurewise_std_normalization=True,
rotation_range=20,
width_shift_range=0.2,
height_shift_range=0.2,
horizontal_flip=True)
test_datagen = ImageDataGenerator(rescale=1./255.0)
##This is required when featurewise_center/featurewise_std_normalization is True
datagen.fit(traing_CT_img)
test_datagen.fit(Validation_CT_img)
#### Batch Normalization
traing_CT_img= datagen.standardize(traing_CT_img)
Validation_CT_img= test_datagen.standardize(Validation_CT_img)
loading_fold_training_label_categorical= np_utils.to_categorical(loading_fold_training_label,num_classes)
loading_fold_Validation_label_categorical= np_utils.to_categorical(loading_fold_Validation_label,num_classes)
###Generating batches
train_datagen=datagen.flow(traing_CT_img,loading_fold_training_label_categorical,batch_size=batch_size)
validation_datagen=test_datagen.flow(Validation_CT_img,loading_fold_Validation_label_categorical,batch_size=batch_size)
###Check-point
tensorboard = TensorBoard(log_dir="logs{}_5_16_2019_HU200/Fold-{}_{}".format(Disease_name,cc,time()))
filepath= '{}_weights-best_model-Fold-{}'.format(Disease_name,cc)+'.hdf5'
checkpoint = ModelCheckpoint(filepath, monitor='val_acc', verbose=1, save_best_only=True, mode='max')
callbacks_list = [checkpoint,EarlyStopping(monitor='val_acc', patience=300, verbose=1, mode='max'),tensorboard]
custom_resnet_model=get_model('Resnet50')
print('----Starting training.......>.>.>..')
his=custom_resnet_model.fit_generator(train_datagen,
validation_data=validation_datagen,
epochs=epochs,
steps_per_epoch=math.ceil(traing_CT_img.shape[0]/batch_size),
verbose=1,
validation_steps=math.ceil(Validation_CT_img.shape[0]/batch_size),
callbacks=callbacks_list)
##### Saving The History
History_name = '{}-History-Fold-{}'.format(Disease_name,cc)+'.csv'
pd.DataFrame(his.history).to_csv(History_name)
####Prediction
pred = custom_resnet_model.predict(Validation_CT_img)
score= custom_resnet_model.evaluate(Validation_CT_img, loading_fold_Validation_label_categorical)
probability_for_normal_class=pred[:,0]
probability_for_edema_class=pred[:,1]
prediction=np.argmax(pred, axis=-1)
roc_name = '{}_roc-Fold-{}'.format(Disease_name,cc)+'.csv'
#fpr_keras, tpr_keras, _ = roc_curve(loading_fold_Validation_label,probability_for_atlectasis_class)
Inf0_data=pd.DataFrame(list(zip(pred,probability_for_normal_class,probability_for_edema_class,prediction,
loading_fold_Validation_label)),
columns=['pred','pred0','pred1','prediction','y_true'])
Inf0_data.to_csv(roc_name, encoding='utf-8', index=False)
###calculating AUC
#auc_keras = auc(fpr_keras, tpr_keras)
####Saving Validation Data
print('Test score:', score[0])
print('Test accuracy:', score[1])
PredictionAndTrueLabel_name = '{}_PredictionAndTruelabel-Fold-{}'.format(Disease_name,cc)+'.csv'
Inf0_data=pd.DataFrame(list(zip(prediction,loading_fold_Validation_label)),
columns=['pre','lbls'])
Inf0_data.to_csv(PredictionAndTrueLabel_name, encoding='utf-8', index=False)
####Model and Normalizer
model_name = '{}_Model-Fold-{}'.format(Disease_name,cc)+'.json'
normalizer_name = '{}_Normalizer-Fold-{}'.format(Disease_name,cc)+'.pkl'
#####Saving t he model##########
model_json = custom_resnet_model.to_json()
with open(model_name, 'w') as json_file:
json_file.write(model_json)
joblib.dump(datagen,normalizer_name)
print("model and weights have been saved\n")
print('#-----------Done Training Fold-{} Data----------------#\n'.format(cc))
print('----------------------Training Ends------------------------------ ')