-
Notifications
You must be signed in to change notification settings - Fork 13
/
Copy pathCTpreprocess.py
189 lines (176 loc) · 7.78 KB
/
CTpreprocess.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
import os
import time
import shutil
import numpy as np
import SimpleITK as sitk
import scipy.ndimage as ndimage
from utils import *
import warnings
warnings.simplefilter("ignore")
def window_transform_truncted(ct_array, windowWidth, windowCenter, normal=False):
"""
return: trucated image according to window center and window width
and normalized to [0,1]
"""
minWindow = float(windowCenter) - 0.5 * float(windowWidth)
maxWindow = float(windowCenter) + 0.5 * float(windowWidth)
ct_array[ct_array < minWindow] = minWindow;
ct_array[ct_array > maxWindow] = maxWindow
if not normal:
ct_array = (ct_array * 255).astype('uint8')
return ct_array
def window_transform(ct_array, windowWidth, windowCenter, normal=False):
"""
return: trucated image according to window center and window width
and normalized to [0,1]
"""
minWindow = float(windowCenter) - 0.5*float(windowWidth)
newimg = (ct_array - minWindow) / float(windowWidth)
newimg[newimg < 0] = 0
newimg[newimg > 1] = 1
if not normal:
newimg = (newimg * 255).astype('uint8')
return newimg
def generate_subimage(ct_array,seg_array,stridez, stridex, stridey, blockz, blockx, blocky,
saved_idx,origin,direction,xyz_thickness,savedct_path,savedseg_path,saved_prefix):
num_z = (ct_array.shape[0]-blockz)//stridez + 1#math.floor()
num_x = (ct_array.shape[1]-blockx)//stridex + 1
num_y = (ct_array.shape[2]-blocky)//stridey + 1
idx = 0
savedct_path = os.path.join(savedct_path,saved_prefix)
savedseg_path = os.path.join(savedseg_path,saved_prefix.replace('volume','segmentation'))
if os.path.exists(savedct_path)|os.path.exists(savedseg_path):
shutil.rmtree(savedct_path)
shutil.rmtree(savedseg_path)
os.mkdir(savedct_path)
os.mkdir(savedseg_path)
for z in range(num_z):
for x in range(num_x):
for y in range(num_y):
seg_block = seg_array[z*stridez:z*stridez+blockz,x*stridex:x*stridex+blockx,y*stridey:y*stridey+blocky]
if seg_block.any():
ct_block = ct_array[z * stridez:z * stridez + blockz, x * stridex:x * stridex + blockx,
y * stridey:y * stridey + blocky]
saved_ctname = os.path.join(savedct_path,str(idx) +'.nii')
saved_segname = os.path.join(savedseg_path,str(idx)+'.nii')
saved_preprocessed(ct_block,origin,direction,xyz_thickness,saved_ctname)
saved_preprocessed(seg_block,origin,direction,xyz_thickness,saved_segname)
idx = idx + 1
return saved_idx + idx
def saved_preprocessed(savedImg,origin,direction,xyz_thickness,saved_name):
newImg = sitk.GetImageFromArray(savedImg)
newImg.SetOrigin(origin)
newImg.SetDirection(direction)
newImg.SetSpacing((xyz_thickness[0], xyz_thickness[1], xyz_thickness[2]))
sitk.WriteImage(newImg, saved_name)
def preprocess(blockzxy,config):
# Clear saved dir
if os.path.exists(config['savedct_path']) is True:
shutil.rmtree(config['savedct_path'])
os.makedirs(config['savedct_path'])
if os.path.exists(config['savedseg_path']) is True:
shutil.rmtree(config['savedseg_path'])
os.makedirs(config['savedseg_path'])
for i in range(config['num_file']):#num_file
saved_prefix = 'volume-'+str(i)
ct = sitk.ReadImage(os.path.join(config['file_path'],'volume-'+str(i)+'.nii'), sitk.sitkFloat32)# sitk.sitkInt16 Read one image using SimpleITK
origin = ct.GetOrigin()
direction = ct.GetDirection()
ct_array = sitk.GetArrayFromImage(ct)
seg = sitk.ReadImage(os.path.join(config['file_path'],'segmentation-'+str(i)+'.nii'), sitk.sitkFloat32)
seg_array = sitk.GetArrayFromImage(seg)
print('-------','volume-'+str(i)+'.nii','-------')
print('original space', np.array(ct.GetSpacing()))
print('original shape',ct_array.shape)
# step1: spacing interpolation
# order=0:nearest interpolation;order=1:bilinear interpolation;order=3:cubic interpolation
ct_array = ndimage.zoom(ct_array, (ct.GetSpacing()[-1] / config['xyz_thickness'][-1],
ct.GetSpacing()[0] / config['xyz_thickness'][0],
ct.GetSpacing()[1] / config['xyz_thickness'][1]), order=3)
# 对金标准插值不应该使用高级插值方式，这样会破坏边界部分,检查数据输出很重要！！！
# 使用order=1可确保zoomed seg unique = [0,1,2]
seg_array = ndimage.zoom(seg_array, (ct.GetSpacing()[-1] / config['xyz_thickness'][-1],
ct.GetSpacing()[0] / config['xyz_thickness'][0],
ct.GetSpacing()[1] / config['xyz_thickness'][1]), order=0)
print('new space', config['xyz_thickness'])
print('zoomed shape:', ct_array.shape, ',', seg_array.shape)
# step2:window transform
print('zoomed seg unique:',np.unique(seg_array))
seg_tumor = seg_array == 2
ct_tumor = ct_array * seg_tumor
tumor_min = ct_tumor.min()
tumor_max = ct_tumor.max()
tumor_wide = tumor_max - tumor_min
if config['window_wc']:
# by customosed uniform window wide and center
ct_array = window_transform(ct_array, config['window_wc'][0], config['window_wc'][1], normal=True)
print('window transform(', config['window_wc'][0], ',', config['window_wc'][1], '):', ct_array.min(), ct_array.max())
elif tumor_wide == 0:# handle the case which does not contain tumor
# by liver
seg_liver = seg_array >= 1
ct_liver = ct_array * seg_liver
liver_min = ct_liver.min()
liver_max = ct_liver.max()
liver_wide = liver_max - liver_min
liver_center = (liver_max + liver_min) / 2
ct_array = window_transform(ct_array, liver_wide, liver_center, normal=True)
print('window transform(', liver_wide, ',', liver_center, '):', ct_array.min(), ct_array.max())
else:
# by tumor
tumor_center = (tumor_max + tumor_min) / 2
ct_array = window_transform(ct_array, tumor_wide, tumor_center, normal=True)
print('window transform(', tumor_wide, ',', tumor_center, '):', ct_array.min(), ct_array.max())
# step3:get mask effective range(startpostion:endpostion)
z = np.any(seg_array, axis=(1, 2)) # seg_array.shape(125, 256, 256)
start_slice, end_slice = np.where(z)[0][[0, -1]]
if start_slice - config['expand_slice'] < 0:
start_slice = 0
else:
start_slice -= config['expand_slice']
if end_slice + config['expand_slice'] >= seg_array.shape[0]:
end_slice = seg_array.shape[0] - 1
else:
end_slice += config['expand_slice']
ct_array = ct_array[start_slice:end_slice + 1, :, :]
seg_array = seg_array[start_slice:end_slice + 1, :, :]
print('effective shape:', ct_array.shape,',',seg_array.shape)
if ct_array.shape[0] < blockzxy[0]:
print('generate no subimage !')
else:
# step4:generate subimage
# step5 save the preprocessed data
config['saved_idx'] = generate_subimage(ct_array, seg_array,
config['stridezxy'][0], config['stridezxy'][1], config['stridezxy'][2],
blockzxy[0], blockzxy[1], blockzxy[2],
config['saved_idx'], origin, direction,config['xyz_thickness'],
config['savedct_path'],config['savedseg_path'],saved_prefix)
print(config['saved_idx'])
if __name__ == '__main__':
start_time = time.time()
logfile = './printLog'
if os.path.isfile(logfile):
os.remove(logfile)
sys.stdout = Logger(logfile)#see utils.py
##########hyperparameters##########
blockzxy = [64, 256, 256]
config = {
'file_path' : "/data/lihuiyu/LiTS/Training_dataset",
'savedct_path' : "/data/lihuiyu/LiTS/Preprocessed_S5_W20040/ct",
'savedseg_path' : "/data/lihuiyu/LiTS/Preprocessed_S5_W20040/seg",
'num_file' : 131,
'window_wc':[200,40],#[]means by automatic liver and tumor center
'stridezxy' : [blockzxy[0] // 5, blockzxy[1] // 5, blockzxy[2] // 5],
'expand_slice' : 10,
'xyz_thickness' : [1.0, 1.0, 1.0],
'saved_idx' : 0
}
##########end hyperparameters#######
# Normal preprocess
print(config['savedct_path'].split('/')[-2])
print(config['window_wc'])
print(config['stridezxy'])
preprocess(blockzxy,config)
# Decide preprocess of different stride and window
# Decide_preprocess(blockzxy,config)
print('Time {:.3f} min'.format((time.time() - start_time) / 60))
print(time.strftime('%Y/%m/%d-%H:%M:%S', time.localtime()))