-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathGet_patch.py
205 lines (159 loc) · 6.95 KB
/
Get_patch.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
import numpy as np
import nibabel as nib
import time
import os
import pandas as pd
import multiprocessing
from skimage.morphology import skeletonize
def get_patch(image, label, enhance, patch_size):
'''
根据真实标签中心线来选取训练集，训练集有无标签比例为1：1
:param image: 真实图像
:param label: 真实标签
:param flag_label: 先验区域
:param patch_size: patch的大小
:return: data_patch_list:图像patch合集 ;label_patch_list,标签patch合集; loc_record patch位置合集
'''
flag_label = label
shape = flag_label.shape
shape = np.array(shape)
label_patch_list = []
data_patch_list = []
enhance_patch_list = []
flag_record = skeletonize(flag_label.astype(np.uint8))
i, j, k = np.where(flag_record == 1)
loc_record = []
# 沿着中心线裁剪
pos = 0
neg = 0
for index in range(i.shape[0]):
if flag_record[i[index], j[index], k[index]] == 1:
c = np.array([i[index], j[index], k[index]])
s = c - patch_size // 2
e = c + patch_size // 2
z_s = np.zeros_like(s)
e[s < z_s] = patch_size
s[s < z_s] = 0
s[e > shape] = shape[e > shape] - patch_size
e[e > shape] = shape[e > shape]
label_patch = label[s[0]:e[0], s[1]:e[1], s[2]:e[2]]
label_patch_list.append(label_patch)
data_patch = image[s[0]:e[0], s[1]:e[1], s[2]:e[2]]
data_patch_list.append(data_patch)
enhance_patch = enhance[s[0]:e[0], s[1]:e[1], s[2]:e[2]]
enhance_patch_list.append(enhance_patch)
loc_record.append(s)
flag_record[s[0]:e[0], s[1]:e[1], s[2]:e[2]] = 0
pos = pos + 1
x = np.arange(patch_size // 2, shape[0], patch_size)
y = np.arange(patch_size // 2, shape[1], patch_size)
z = np.arange(patch_size // 2, shape[2], patch_size)
i, j, k = np.meshgrid(x, y, z)
i, j, k = i.flatten(), j.flatten(), k.flatten()
arr = np.arange(i.shape[0])
np.random.shuffle(arr)
arr = arr.tolist()
for index in arr:
c = np.array([i[index], j[index], k[index]])
s = c - patch_size // 2
e = c + patch_size // 2
z_s = np.zeros_like(s)
e[s < z_s] = patch_size
s[s < z_s] = 0
s[e > shape] = shape[e > shape] - patch_size
e[e > shape] = shape[e > shape]
label_patch = label[s[0]:e[0], s[1]:e[1], s[2]:e[2]]
if label_patch.sum() == 0:
label_patch_list.append(label_patch)
data_patch = image[s[0]:e[0], s[1]:e[1], s[2]:e[2]]
data_patch_list.append(data_patch)
enhance_patch = enhance[s[0]:e[0], s[1]:e[1], s[2]:e[2]]
enhance_patch_list.append(enhance_patch)
loc_record.append(s)
neg = neg + 1
if neg >= pos:
break
return data_patch_list, label_patch_list, enhance_patch_list, loc_record
def get_patch_valid(image, label, enhance, patch_size):
'''
规则裁剪，制作重复区域
:param image:
:param label:
:param enhance:
:param patch_size:
:return:
'''
shape = label.shape
shape = np.array(shape)
label_patch_list = []
data_patch_list = []
enhance_patch_list = []
img_record = label.copy()
# i,j,k=np.where(label==1)
x = np.arange(patch_size // 2, shape[0], patch_size // 2)
y = np.arange(patch_size // 2, shape[1], patch_size //2)
z = np.arange(patch_size // 2, shape[2], patch_size //2)
i, j, k = np.meshgrid(x, y, z)
i, j, k = i.flatten(), j.flatten(), k.flatten()
loc_record = []
for index in range(i.shape[0]):
c = np.array([i[index], j[index], k[index]])
s = c - patch_size // 2
e = c + patch_size // 2
z_s = np.zeros_like(s)
e[s < z_s] = patch_size
s[s < z_s] = 0
s[e > shape] = shape[e > shape] - patch_size
e[e > shape] = shape[e > shape]
label_patch = label[s[0]:e[0], s[1]:e[1], s[2]:e[2]]
label_patch_list.append(label_patch)
data_patch = image[s[0]:e[0], s[1]:e[1], s[2]:e[2]]
data_patch_list.append(data_patch)
enhance_patch_list.append(enhance[s[0]:e[0], s[1]:e[1], s[2]:e[2]])
loc_record.append(s)
img_record[s[0]:e[0], s[1]:e[1], s[2]:e[2]] = 0
return data_patch_list, label_patch_list, enhance_patch_list, loc_record
class Get_patch:
def __init__(self, data_path, label_path, frangi_path, save_path, patch_size, data_type):
self.data_path = data_path
self.label_path = label_path
self.frangi_path = frangi_path
self.save_img_path = os.path.join(save_path, 'patch_%d' % patch_size, '%s_img_patch' % data_type)
self.save_label_path = os.path.join(save_path, 'patch_%d' % patch_size, '%s_label_patch' % data_type)
self.save_frangi_path = os.path.join(save_path, 'patch_%d' % patch_size, '%s_enhance_patch' % data_type)
self.save_record_path = os.path.join(save_path, 'patch_%d' % patch_size, 'csv_patch_record')
self.patch_size = patch_size
self.data_type = data_type
os.makedirs(self.save_img_path,exist_ok=True)
os.makedirs(self.save_label_path,exist_ok=True)
os.makedirs(self.save_frangi_path,exist_ok=True)
os.makedirs(self.save_record_path,exist_ok=True)
def run_main(self, i):
print(i)
df = pd.DataFrame(columns=['x', 'y', 'z'])
d_path = os.path.join(self.data_path, i, 'img.nii.gz')
l_path = os.path.join(self.label_path, i, 'label.nii.gz')
e_path = os.path.join(self.frangi_path, i, 'frangi.nii.gz')
d_nii = nib.load(d_path)
l_nii = nib.load(l_path)
e_nii = nib.load(e_path)
p_label = l_nii.get_fdata()
p_label[p_label > 0] = 1
p_label[p_label <= 0] = 0
img = d_nii.get_fdata()
enhance = e_nii.get_fdata()
if self.data_type == 'train':
g_p = get_patch
else:
g_p = get_patch_valid
img_list, label_list, enhance_list, loc_record = g_p(img, p_label, enhance, self.patch_size)
ID = i
affine = l_nii.affine
for index, p in enumerate(img_list):
nib.save(nib.Nifti1Image(p, affine), os.path.join(self.save_img_path, ID + '_%d.nii.gz' % index))
nib.save(nib.Nifti1Image(label_list[index], affine),
os.path.join(self.save_label_path, ID + '_%d.nii.gz' % index))
df.loc[ID + '_%d.nii.gz' % index] = [loc_record[index][0], loc_record[index][1], loc_record[index][2]]
nib.save(nib.Nifti1Image(enhance_list[index], affine),
os.path.join(self.save_frangi_path, ID + '_%d.nii.gz' % index))
df.to_csv(os.path.join(self.save_record_path, ID + '.csv'))