-
Notifications
You must be signed in to change notification settings - Fork 3
Expand file tree
/
Copy pathpreprocessing.py
More file actions
133 lines (110 loc) · 6.1 KB
/
preprocessing.py
File metadata and controls
133 lines (110 loc) · 6.1 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
import numpy as np
import os
import medpy.io as medio
import tensorflow as tf
import shutil
import SimpleITK as sitk
def DeepMedicFormat_crop():
raw_data_pth = ''
save_data_pth = ''
data_fds = os.listdir(raw_data_pth)
data_fds.sort()
for data_fd in data_fds:
img_fds = os.listdir(raw_data_pth+'/'+data_fd)
img_fds.sort()
for img_fd in img_fds:
print img_fd
if not os.path.exists(save_data_pth+'/'+data_fd+'_'+img_fd):
os.makedirs(save_data_pth+'/'+data_fd+'_'+img_fd)
modality_fds = os.listdir(raw_data_pth+'/'+data_fd+'/'+img_fd)
modality_fds.sort()
modality_fds[1], modality_fds[0] = modality_fds[0], modality_fds[1]
brainmask_arr_list = []
for modality_fd in modality_fds:
if 'MR' in modality_fd:
modality_name = modality_fd.split('.')[4].split('_')[-1]+'_subtrMeanDivStd'
else:
modality_name = 'OTMultiClass'
if os.path.exists(raw_data_pth + '/' + data_fd + '/' + img_fd + '/' + modality_fd + '/' + modality_fd + '.mha'):
sitk.WriteImage(sitk.ReadImage(raw_data_pth + '/' + data_fd + '/' + img_fd + '/' + modality_fd + '/' + modality_fd + '.mha'),
save_data_pth + '/'+data_fd+'_'+img_fd + '/' + modality_name + '.nii.gz')
if 'MR' in modality_fd:
image_arr, image_header = medio.load(save_data_pth + '/' +data_fd+'_'+img_fd + '/' + modality_name + '.nii.gz')
brainmask_arr = image_arr.copy()
brainmask_arr[brainmask_arr > 0] = 1
brainmask_arr_list.append(brainmask_arr)
brainmask_arr = brainmask_arr_list[0]
for m in xrange(1,len(brainmask_arr_list)):
brainmask_arr = brainmask_arr+brainmask_arr_list[m]
brainmask_arr[brainmask_arr>0]=1
# brainmask_arr[brainmask_arr <4] = 0
medio.save(brainmask_arr,
save_data_pth + '/'+data_fd+'_'+img_fd + '/brainmask.nii.gz',
image_header)
brainmask_arr, brainmask_header = medio.load(save_data_pth + '/' + data_fd + '_' + img_fd + '/brainmask.nii.gz')
roi_ind = np.where(brainmask_arr > 0)
roi_bbx = [roi_ind[0].min(), roi_ind[0].max(), roi_ind[1].min(), roi_ind[1].max(), roi_ind[2].min(), roi_ind[2].max()]
for modality_fd in modality_fds:
if 'MR' in modality_fd:
modality_name = modality_fd.split('.')[4].split('_')[-1]+'_subtrMeanDivStd'
image_arr, image_header = medio.load(save_data_pth + '/'+data_fd+'_'+img_fd + '/' + modality_name + '.nii.gz')
roi_arr = image_arr[brainmask_arr>0]
lower_limit = np.percentile(roi_arr, 1)
upper_limit = np.percentile(roi_arr, 99)
roi_arr = roi_arr[roi_arr>lower_limit]
roi_arr = roi_arr[roi_arr<upper_limit]
roi_mean = roi_arr.mean()
roi_std = roi_arr.std()
image_arr = (image_arr-roi_mean)/roi_std
image_arr_crop = image_arr[roi_bbx[0]:roi_bbx[1]+1, roi_bbx[2]:roi_bbx[3]+1, roi_bbx[4]:roi_bbx[5]+1]
medio.save(image_arr_crop,
save_data_pth + '/'+data_fd+'_'+img_fd + '/' + modality_name + '.nii.gz',
image_header)
else:
modality_name = 'OTMultiClass'
image_arr, image_header = medio.load(save_data_pth + '/' + data_fd + '_' + img_fd + '/' + modality_name + '.nii.gz')
image_arr_crop = image_arr[roi_bbx[0]:roi_bbx[1] + 1, roi_bbx[2]:roi_bbx[3] + 1, roi_bbx[4]:roi_bbx[5] + 1]
medio.save(image_arr_crop,
save_data_pth + '/' + data_fd + '_' + img_fd + '/' + modality_name + '.nii.gz',
image_header)
brainmask_arr_crop = brainmask_arr[roi_bbx[0]:roi_bbx[1]+1, roi_bbx[2]:roi_bbx[3]+1, roi_bbx[4]:roi_bbx[5]+1]
medio.save(brainmask_arr_crop,
save_data_pth + '/' + data_fd + '_' + img_fd + '/brainmask.nii.gz',
brainmask_header)
def nii2tfrecord():
os.environ['CUDA_VISIBLE_DEVICES'] = '0'
raw_data_pth = ''
save_data_pth = ''
pid_all = os.listdir(raw_data_pth)
pid_all.sort()
with open('../Data/train.txt', 'r') as fp:
rows = fp.readlines()
pid_all = [row[:-1] for row in rows]
pid_all.sort()
cnt = 0
for pid_indx, pid in enumerate(pid_all):
cnt +=1
modality_all = os.listdir(raw_data_pth+'/'+pid)
modality_all.sort()
for modality in modality_all:
data_arr, data_header = medio.load(raw_data_pth + '/' + pid + '/' + modality)
data_arr = np.float32(data_arr)
dsize_dim0_val = data_arr.shape[0]
dsize_dim1_val = data_arr.shape[1]
dsize_dim2_val = data_arr.shape[2]
if not os.path.exists(save_data_pth+'/'+pid):
os.makedirs(save_data_pth+'/'+pid)
writer = tf.python_io.TFRecordWriter(save_data_pth+'/'+pid+'/'+modality.split('.')[0]+'.tfrecords')
feature = {'data_vol': tf.train.Feature(
bytes_list=tf.train.BytesList(value=[tf.compat.as_bytes(data_arr.tostring())])),
'dsize_dim0': tf.train.Feature(int64_list=tf.train.Int64List(value=[dsize_dim0_val])),
'dsize_dim1': tf.train.Feature(int64_list=tf.train.Int64List(value=[dsize_dim1_val])),
'dsize_dim2': tf.train.Feature(int64_list=tf.train.Int64List(value=[dsize_dim2_val])),
'data_indx': tf.train.Feature(int64_list=tf.train.Int64List(value=[pid_indx])),
}
example = tf.train.Example(features=tf.train.Features(feature=feature))
writer.write(example.SerializeToString())
writer.close()
print (pid, modality, cnt)
if __name__=='__main__':
nii2tfrecord()