[56a46c]: / train.py

Download this file

260 lines (206 with data), 11.9 kB

  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
import numpy as np
import tf_models
from sklearn.preprocessing import scale
import tensorflow as tf
from tensorflow.contrib.keras.python.keras.backend import learning_phase
from tensorflow.contrib.keras.python.keras.layers import concatenate, Conv3D
from nibabel import load as load_nii
import os
import argparse
import keras
def parse_inputs():
parser = argparse.ArgumentParser(description='train the model')
parser.add_argument('-r', '--root-path', dest='root_path', default='/media/lele/Data/spie/Brats17TrainingData/HGG')
parser.add_argument('-sp', '--save-path', dest='save_path', default='dense24_correction')
parser.add_argument('-lp', '--load-path', dest='load_path', default='dense24_correction')
parser.add_argument('-ow', '--offset-width', dest='offset_w', type=int, default=12)
parser.add_argument('-oh', '--offset-height', dest='offset_h', type=int, default=12)
parser.add_argument('-oc', '--offset-channel', dest='offset_c', nargs='+', type=int, default=12)
parser.add_argument('-ws', '--width-size', dest='wsize', type=int, default=38)
parser.add_argument('-hs', '--height-size', dest='hsize', type=int, default=38)
parser.add_argument('-cs', '--channel-size', dest='csize', type=int, default=38)
parser.add_argument('-ps', '--pred-size', dest='psize', type=int, default=12)
parser.add_argument('-bs', '--batch-size', dest='batch_size', type=int, default=2)
parser.add_argument('-e', '--num-epochs', dest='num_epochs', type=int, default=5)
parser.add_argument('-c', '--continue-training', dest='continue_training', type=bool, default=False)
parser.add_argument('-mn', '--model_name', dest='model_name', type=str, default='dense24')
parser.add_argument('-nc', '--n4correction', dest='correction', type=bool, default=False)
parser.add_argument('-gpu', '--gpu_id', dest='gpu_id', type=str, default='0')
return vars(parser.parse_args())
options = parse_inputs()
os.environ["CUDA_VISIBLE_DEVICES"] = options['gpu_id']
def acc_tf(y_pred, y_true):
correct_prediction = tf.equal(tf.cast(tf.argmax(y_pred, -1), tf.int32), tf.cast(tf.argmax(y_true, -1), tf.int32))
return 100 * tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
def get_patches_3d(data, labels, centers, hsize, wsize, csize, psize, preprocess=True):
"""
:param data: 4D nparray (h, w, c, ?)
:param centers:
:param hsize:
:param wsize:
:param csize:
:return:
"""
patches_x, patches_y = [], []
offset_p = (hsize - psize) / 2
for i in range(len(centers[0])):
h, w, c = centers[0, i], centers[1, i], centers[2, i]
h_beg = min(max(0, h - hsize / 2), 240 - hsize)
w_beg = min(max(0, w - wsize / 2), 240 - wsize)
c_beg = min(max(0, c - csize / 2), 155 - csize)
ph_beg = h_beg + offset_p
pw_beg = w_beg + offset_p
pc_beg = c_beg + offset_p
vox = data[h_beg:h_beg + hsize, w_beg:w_beg + wsize, c_beg:c_beg + csize, :]
vox_labels = labels[ph_beg:ph_beg + psize, pw_beg:pw_beg + psize, pc_beg:pc_beg + psize]
patches_x.append(vox)
patches_y.append(vox_labels)
return np.array(patches_x), np.array(patches_y)
def positive_ratio(x):
return float(np.sum(np.greater(x, 0))) / np.prod(x.shape)
def norm(image):
image = np.squeeze(image)
image_nonzero = image[np.nonzero(image)]
return (image - image_nonzero.mean()) / image_nonzero.std()
def segmentation_loss(y_true, y_pred, n_classes):
y_true = tf.reshape(y_true, (-1, n_classes))
y_pred = tf.reshape(y_pred, (-1, n_classes))
return tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=y_true,
logits=y_pred))
def vox_preprocess(vox):
vox_shape = vox.shape
vox = np.reshape(vox, (-1, vox_shape[-1]))
vox = scale(vox, axis=0)
return np.reshape(vox, vox_shape)
def one_hot(y, num_classes):
y_ = np.zeros([len(y), num_classes])
y_[np.arange(len(y)), y] = 1
return y_
def dice_coef_np(y_true, y_pred, num_classes):
"""
:param y_true: sparse labels
:param y_pred: sparse labels
:param num_classes: number of classes
:return:
"""
y_true = y_true.astype(int)
y_pred = y_pred.astype(int)
y_true = y_true.flatten()
y_true = one_hot(y_true, num_classes)
y_pred = y_pred.flatten()
y_pred = one_hot(y_pred, num_classes)
intersection = np.sum(y_true * y_pred, axis=0)
return (2. * intersection) / (np.sum(y_true, axis=0) + np.sum(y_pred, axis=0))
def vox_generator(all_files, n_pos, n_neg,correction= False):
path = options['root_path']
while 1:
for file in all_files:
if correction:
flair = load_nii(os.path.join(path, file, file + '_flair_corrected.nii.gz')).get_data()
t2 = load_nii(os.path.join(path, file, file + '_t2_corrected.nii.gz')).get_data()
t1 = load_nii(os.path.join(path, file, file + '_t1_corrected.nii.gz')).get_data()
t1ce = load_nii(os.path.join(path, file, file + '_t1ce_corrected.nii.gz')).get_data()
else:
flair = load_nii(os.path.join(path, file, file + '_flair.nii.gz')).get_data()
t2 = load_nii(os.path.join(path, file, file + '_t2.nii.gz')).get_data()
t1 = load_nii(os.path.join(path, file, file + '_t1.nii.gz')).get_data()
t1ce = load_nii(os.path.join(path, file, file + '_t1ce.nii.gz')).get_data()
data_norm = np.array([norm(flair), norm(t2), norm(t1), norm(t1ce)])
data_norm = np.transpose(data_norm, axes=[1, 2, 3, 0])
labels = load_nii(os.path.join(path, file, file+'_seg.nii.gz')).get_data()
foreground = np.array(np.where(labels > 0))
background = np.array(np.where((labels == 0) & (flair > 0)))
# n_pos = int(foreground.shape[1] * discount)
foreground = foreground[:, np.random.permutation(foreground.shape[1])[:n_pos]]
background = background[:, np.random.permutation(background.shape[1])[:n_neg]]
centers = np.concatenate((foreground, background), axis=1)
centers = centers[:, np.random.permutation(n_neg+n_pos)]
yield data_norm, labels, centers
def label_transform(y, nlabels):
return [
keras.utils.to_categorical(np.copy(y).astype(dtype=np.bool),
num_classes=2).reshape([y.shape[0], y.shape[1], y.shape[2], y.shape[3], 2]),
keras.utils.to_categorical(y,
num_classes=nlabels).reshape([y.shape[0], y.shape[1], y.shape[2], y.shape[3], nlabels])
]
def train():
NUM_EPOCHS = options['num_epochs']
LOAD_PATH = options['load_path']
SAVE_PATH = options['save_path']
PSIZE = options['psize']
HSIZE = options['hsize']
WSIZE = options['wsize']
CSIZE = options['csize']
model_name= options['model_name']
BATCH_SIZE = options['batch_size']
continue_training = options['continue_training']
files = []
num_labels = 5
with open('train.txt') as f:
for line in f:
files.append(line[:-1])
print '%d training samples' % len(files)
flair_t2_node = tf.placeholder(dtype=tf.float32, shape=(None, HSIZE, WSIZE, CSIZE, 2))
t1_t1ce_node = tf.placeholder(dtype=tf.float32, shape=(None, HSIZE, WSIZE, CSIZE, 2))
flair_t2_gt_node = tf.placeholder(dtype=tf.int32, shape=(None, PSIZE, PSIZE, PSIZE, 2))
t1_t1ce_gt_node = tf.placeholder(dtype=tf.int32, shape=(None, PSIZE, PSIZE, PSIZE, 5))
if model_name == 'dense48':
flair_t2_15, flair_t2_27 = tf_models.BraTS2ScaleDenseNetConcat_large(input=flair_t2_node, name='flair')
t1_t1ce_15, t1_t1ce_27 = tf_models.BraTS2ScaleDenseNetConcat_large(input=t1_t1ce_node, name='t1')
elif model_name == 'no_dense':
flair_t2_15, flair_t2_27 = tf_models.PlainCounterpart(input=flair_t2_node, name='flair')
t1_t1ce_15, t1_t1ce_27 = tf_models.PlainCounterpart(input=t1_t1ce_node, name='t1')
elif model_name == 'dense24':
flair_t2_15, flair_t2_27 = tf_models.BraTS2ScaleDenseNetConcat(input=flair_t2_node, name='flair')
t1_t1ce_15, t1_t1ce_27 = tf_models.BraTS2ScaleDenseNetConcat(input=t1_t1ce_node, name='t1')
else:
print' No such model name '
t1_t1ce_15 = concatenate([t1_t1ce_15, flair_t2_15])
t1_t1ce_27 = concatenate([t1_t1ce_27, flair_t2_27])
flair_t2_15 = Conv3D(2, kernel_size=1, strides=1, padding='same', name='flair_t2_15_cls')(flair_t2_15)
flair_t2_27 = Conv3D(2, kernel_size=1, strides=1, padding='same', name='flair_t2_27_cls')(flair_t2_27)
t1_t1ce_15 = Conv3D(num_labels, kernel_size=1, strides=1, padding='same', name='t1_t1ce_15_cls')(t1_t1ce_15)
t1_t1ce_27 = Conv3D(num_labels, kernel_size=1, strides=1, padding='same', name='t1_t1ce_27_cls')(t1_t1ce_27)
flair_t2_score = flair_t2_15[:, 13:25, 13:25, 13:25, :] + \
flair_t2_27[:, 13:25, 13:25, 13:25, :]
t1_t1ce_score = t1_t1ce_15[:, 13:25, 13:25, 13:25, :] + \
t1_t1ce_27[:, 13:25, 13:25, 13:25, :]
loss = segmentation_loss(flair_t2_gt_node, flair_t2_score, 2) + \
segmentation_loss(t1_t1ce_gt_node, t1_t1ce_score, 5)
acc_flair_t2 = acc_tf(y_pred=flair_t2_score, y_true=flair_t2_gt_node)
acc_t1_t1ce = acc_tf(y_pred=t1_t1ce_score, y_true=t1_t1ce_gt_node)
update_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
with tf.control_dependencies(update_ops):
optimizer = tf.train.AdamOptimizer(learning_rate=5e-4).minimize(loss)
saver = tf.train.Saver(max_to_keep=15)
data_gen_train = vox_generator(all_files=files, n_pos=200, n_neg=200,correction = options['correction'])
with tf.Session() as sess:
if continue_training:
saver.restore(sess, LOAD_PATH)
else:
sess.run(tf.global_variables_initializer())
for ei in range(NUM_EPOCHS):
for pi in range(len(files)):
acc_pi, loss_pi = [], []
data, labels, centers = data_gen_train.next()
n_batches = int(np.ceil(float(centers.shape[1]) / BATCH_SIZE))
for nb in range(n_batches):
offset_batch = min(nb * BATCH_SIZE, centers.shape[1] - BATCH_SIZE)
data_batch, label_batch = get_patches_3d(data, labels, centers[:, offset_batch:offset_batch + BATCH_SIZE], HSIZE, WSIZE, CSIZE, PSIZE, False)
label_batch = label_transform(label_batch, 5)
_, l, acc_ft, acc_t1c = sess.run(fetches=[optimizer, loss, acc_flair_t2, acc_t1_t1ce],
feed_dict={flair_t2_node: data_batch[:, :, :, :, :2],
t1_t1ce_node: data_batch[:, :, :, :, 2:],
flair_t2_gt_node: label_batch[0],
t1_t1ce_gt_node: label_batch[1],
learning_phase(): 1})
acc_pi.append([acc_ft, acc_t1c])
loss_pi.append(l)
n_pos_sum = np.sum(np.reshape(label_batch[0], (-1, 2)), axis=0)
print 'epoch-patient: %d, %d, iter: %d-%d, p%%: %.4f, loss: %.4f, acc_flair_t2: %.2f%%, acc_t1_t1ce: %.2f%%' % \
(ei + 1, pi + 1, nb + 1, n_batches, n_pos_sum[1]/float(np.sum(n_pos_sum)), l, acc_ft, acc_t1c)
print 'patient loss: %.4f, patient acc: %.4f' % (np.mean(loss_pi), np.mean(acc_pi))
saver.save(sess, SAVE_PATH, global_step=ei)
print 'model saved'
if __name__ == '__main__':
train()