-
Notifications
You must be signed in to change notification settings - Fork 1
/
Data_preprocess.py
215 lines (195 loc) · 8.18 KB
/
Data_preprocess.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
import glob
import nibabel
import cv2
from keras.preprocessing.image import array_to_img, img_to_array, load_img
import numpy as np
from keras.utils import to_categorical
import tensorflow as tf
import keras.backend.tensorflow_backend as KTF
sess = tf.Session(config=tf.ConfigProto(log_device_placement=True))
KTF.set_session(tf.Session(config=tf.ConfigProto(device_count={'gpu': 0})))
#读取原始nii数据的格式：每个病例中元素的最大值、最小值以及病例的形状,并裁剪中心区域160*160保存为npy
def read_data_format(file_dir,save_dir):
imgname = glob.glob(file_dir + '\\*' + '.nii.gz')
for file_name in imgname:
midname = file_name[file_name.rindex("\\") + 1:]
img = nibabel.load(file_dir + midname).get_data()
img_max = np.amax(img)
img_min = np.amin(img)
print('%s: max is %d, min is %d' %(midname, img_max, img_min))
print(img.shape)
img = img[int((img.shape[0]-256)/2):int((img.shape[0]+256)/2), int((img.shape[1]-256)/2):int((img.shape[1]+256)/2)]
print(img.shape)
np.save(save_dir+midname[0:-7], img)
#read_data_format('F:\\7 医学图像数据集\\2020MICCAI心肌病分割挑战赛数据\\train25_myops_gd\\')
#将有标签的切片筛选出来(初步筛选)2020心肌分割挑战赛给的数据都是有效标签，其实不用筛选
def select_slice(file_dir, label_dir, save_file_dir, save_label_dir):
imglist = []
labellist = []
imgname = glob.glob(label_dir+'\\*'+'.npy')
for file_name in imgname:
j = 0
midname = file_name[file_name.rindex("\\") + 1:]
print(midname)
img = np.load(file_dir + midname[0:-6]+'DE.npy')
label = np.load(label_dir + midname)
print(img.shape)
newimg = np.zeros((img.shape[0], img.shape[1], img.shape[2]),dtype='float32')
newlabel = np.zeros((img.shape[0], img.shape[1], label.shape[2]),dtype='float32')
for i in range(label.shape[2]):#筛选当前切片总和大于0的
if np.sum(label[:, :, i:i+1]) > 0:
newimg[:, :, j:j+1] = img[:, :, i:i+1]
newlabel[:, :, j:j+1] = label[:, :, i:i+1]
j = j + 1
finalimg = newimg[:, :, 0: j]
finallabel = newlabel[:, :, 0: j]
print(finalimg.shape)
imglist.append(finalimg)
labellist.append(finallabel)
img_save = np.concatenate(imglist, axis=-1)
print(img_save.shape)
label_save = np.concatenate(labellist, axis=-1)
print(label_save.shape)
np.save(save_file_dir + 'DE_img.npy', img_save)
np.save(save_label_dir + 'label.npy', label_save)
'''
select_slice("F:\\2020MICCAI_Cardiac_Segmentation\\myops2020\\train25\\DE\\",
'F:\\2020MICCAI_Cardiac_Segmentation\\myops2020\\train25_myops_gd\\',
'F:\\2020MICCAI_Cardiac_Segmentation\\myops2020\\',
'F:\\2020MICCAI_Cardiac_Segmentation\\myops2020\\')
'''
#重置标签值并one-hot
def reset_value(label):
for i in range(label.shape[0]):
for j in range(label.shape[1]):
for k in range(label.shape[2]):
if label[i][j][k] == 200:
label[i][j][k] = 1
elif label[i][j][k] == 500:
label[i][j][k] = 2
elif label[i][j][k] == 600:
label[i][j][k] = 3
elif label[i][j][k] == 1220:
label[i][j][k] = 4
elif label[i][j][k] == 2221:
label[i][j][k] = 5
else:
label[i][j][k] = 0
newlabel = to_categorical(label, num_classes=6)
return newlabel
#np.save(label_dir+'newlabel.npy', newlabel)
#reset_value('F:\\2020MICCAI_Cardiac_Segmentation\\myops2020\\label.npy')
#将生成的numpy转换成jpg并保存下来
def test_npy(file_dir,save_dir):
npy = np.load(file_dir)
#npy = np.argmax(npy, axis=-1)
for i in range(npy.shape[2]):
img = npy[:,:,i:i+1]
img = array_to_img(img)
img.save(save_dir+'patient%d.jpg'%i)
'''
file_dir='F:\\2020MICCAI_Cardiac_Segmentation\\myops2020\\T2_img.npy'
save_dir='F:\\2020MICCAI_Cardiac_Segmentation\\myops2020\\T2\\'
test_npy(file_dir, save_dir)
'''
def gamma_aug(data):
value = [0.5, 0.6, 0.7, 0.8, 0.9, 1]
gamma_list = []
for i in range(len(value)):
data_gamma = tf.image.adjust_gamma(data, value[i])
session = tf.Session()
data_gamma = session.run(data_gamma)
gamma_list.append(data_gamma)
gamma_list_save = np.concatenate(gamma_list, axis=0)
return gamma_list_save
def flip_rot_aug(data):
data_rot90 = tf.image.rot90(data)
data_rot180 = tf.image.rot90(data_rot90)
data_rot270 = tf.image.rot90(data_rot180)
data_u2d = tf.image.flip_up_down(data)
data_l2r = tf.image.flip_left_right(data)
data_trans = tf.image.transpose_image(data)
session = tf.Session()
data_rot90 = session.run(data_rot90)
data_rot180 = session.run(data_rot180)
data_rot270 = session.run(data_rot270)
data_u2d = session.run(data_u2d)
data_l2r = session.run(data_l2r)
data_trans = session.run(data_trans)
data_flip_rot_save = np.concatenate([data, data_rot90, data_rot180, data_rot270, data_u2d, data_l2r, data_trans], axis=0)
print(data_flip_rot_save.shape)
return data_flip_rot_save
def data_aug(data):
data1 = gamma_aug(data)
data2 = flip_rot_aug(data1)
print(data2.shape)
return data2
'''
data0 = np.load('F:\\2020MICCAI_Cardiac_Segmentation\\myops2020\\T2_img.npy')
data = data0[:, :, :, np.newaxis].transpose(2, 0, 1, 3)#(102, 256, 256, 1) img
#data = data0.transpose(2, 0, 1, 3)#label
test_list = [0, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100]
train_data = []
test_data = []
for i in range(data.shape[0]):
if i not in test_list:
train_data.append(data[i:i+1, :, :, :])
train = np.concatenate(train_data, axis=0)
for i in test_list:
test_data.append(data[i:i+1, :, :, :])
test = np.concatenate(test_data, axis=0)
train_aug = data_aug(train)
print(train_aug.shape)
np.save('F:\\2020MICCAI_Cardiac_Segmentation\\myops2020\\T2_train.npy', train_aug)
np.save('F:\\2020MICCAI_Cardiac_Segmentation\\myops2020\\T2_test.npy', test)
print(train_aug.shape)
print(test.shape)
'''
def append_slice(file_dir):
imglist = []
imgname = glob.glob(file_dir+'\\*'+'.npy')
for file_name in imgname:
midname = file_name[file_name.rindex("\\") + 1:]
img = np.load(file_dir + midname)
img = img[int((img.shape[0] - 256) / 2):int((img.shape[0] + 256) / 2),
int((img.shape[1] - 256) / 2):int((img.shape[1] + 256) / 2)]
img = np.expand_dims(img, 0)
label = img
for i in range(label.shape[0]):
for j in range(label.shape[1]):
for k in range(label.shape[2]):
if label[i][j][k] == 200:
label[i][j][k] = 1
elif label[i][j][k] == 500:
label[i][j][k] = 2
elif label[i][j][k] == 600:
label[i][j][k] = 3
elif label[i][j][k] == 1220:
label[i][j][k] = 4
elif label[i][j][k] == 2221:
label[i][j][k] = 5
else:
label[i][j][k] = 0
img = to_categorical(label, num_classes=6)
print(img.shape)
imglist.append(img)
img_save = np.concatenate(imglist, axis=0)
print(img_save.shape)
np.save(file_dir + 'train_label.npy', img_save)
#append_slice('F:\media\media\LIBRARY\Datasets\MyoPS2020\Augdata\\train1\\npy\\Labels\\')
#标签像素值统计
def read_data_format(file_dir):
imgname = glob.glob(file_dir + '\\*' + '.nii.gz')
for file_name in imgname:
r=0
midname = file_name[file_name.rindex("\\") + 1:]
img = nibabel.load(file_dir + midname).get_data()
#print(img.shape[2])
for i in range(img.shape[2]):
#print(img.shape[0] * img.shape[1])
for j in range(img.shape[0]):
for k in range(img.shape[1]):
if img[j][k].any()>0:
r=r+1
print(r)
#read_data_format('F:\\7 医学图像数据集\\2020MICCAI心肌病分割挑战赛数据\\train25_myops_gd\\')