forked from Nightmare4214/CTPelvic1K
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmove_data_suitable.py
212 lines (189 loc) · 7.99 KB
/
move_data_suitable.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
#!/usr/bin/env python
# _*_ coding:utf-8 _*_
import glob
import os
import re
import shutil
from batchgenerators.utilities.file_and_folder_operations import subfiles
from sklearn.model_selection import train_test_split
from tqdm import tqdm
"""
dataset format:
image: *_data.nii.gz
label: *_mask_4label.nii.gz
test:
image: *_data.nii.gz
Data will be converted into a unified format used in nnunet
"""
TASK = 'Task11_CERVIX'
home_dir = '/data/datasets/CTPelvic1K'
train_dir = os.path.join(home_dir, f'all_data/nnUNet/rawdata/{TASK}') # 训练集
test_dir = os.path.join(home_dir, f'all_data/nnUNet/rawdata/{TASK}_test') # 测试集
def split_nii_gz(filename):
"""
切割nii.gz的文件名与扩展名
:param filename: nii.gz文件
:return: (文件名,'.nii.gz')
"""
temp, ext1 = os.path.splitext(filename)
temp, ext2 = os.path.splitext(temp)
return temp, ext2 + ext1
def get_filename(path):
"""
获取nii.gz的文件名，并转换
:param path: nii.gz文件
:return: 转换后文件名
"""
path_dir, filename = os.path.split(path)
filename, ext = split_nii_gz(filename)
# -替换为_,去掉dataset{index}_,去掉_data,
return re.sub('_data', '', re.sub(r'^dataset\d_', '', filename.replace('-', '_')))
def get_last_dir(path):
"""
获取最深的目录名
:param path: 路径
:return: 最深的目录名
"""
return os.path.split(os.path.dirname(path))[-1]
# 先验知识： 标签的名字=dataset{index}_{训练集对应的标签的名字}_mask_4label.nii.gz
data_inform = {
'ABDOMEN': {
'index': 1,
'skip': False, # 是否跳过（不使用这个数据集）
'img_path': '/data/datasets/Abdomen/Abdomen', # 修改
'label_path': '/data/datasets/CTPelvic1K/CTPelvic1K_dataset1_mask_mappingback', # 修改
'patten': ['RawData/Training/img/*.nii.gz', 'RawData/Testing/img/*.nii.gz'],
'convert_function': get_filename, # 训练集对应的标签的名字
'test_size': 7,
'all_test': True # 全部作为测试集
},
'COLONOG': { # FIXME 修改convert_function,patten
'index': 2,
'skip': True,
'img_path': '/data/datasets/COLONOGRAPHY', # 修改
'label_path': '/data/datasets/CTPelvic1K/CTPelvic1K_dataset2_mask_mappingback', # 修改
'patten': [],
'convert_function': get_filename,
'test_size': 145,
'all_test': True # 全部作为测试集
},
'MSD_T10': {
'index': 3,
'skip': False,
'img_path': '/data/datasets/MSD_T10/Task10_Colon/Task10_Colon', # 修改
'label_path': '/data/datasets/CTPelvic1K/CTPelvic1K_dataset3_mask_mappingback', # 修改
'patten': ['imagesTr/*.nii.gz', 'imagesTs/*.nii.gz'],
'convert_function': get_filename,
'test_size': 31,
'all_test': True # 全部作为测试集
},
'KITS19': {
'index': 4,
'skip': False,
'img_path': '/data/datasets/kits19', # 修改
'label_path': '/data/datasets/CTPelvic1K/CTPelvic1K_dataset4_mask_mappingback', # 修改
'patten': ['*/imaging.nii.gz'],
'convert_function': get_last_dir,
'test_size': 9,
'all_test': True # 全部作为测试集
},
'CERVIX': {
'index': 5,
'skip': False,
'img_path': '/data/datasets/CERVIX/Cervix', # 修改
'label_path': '/data/datasets/CTPelvic1K/CTPelvic1K_dataset5_mask_mappingback', # 修改
'patten': ['RawData/Training/img/*.nii.gz', 'RawData/Testing/img/*.nii.gz'],
'convert_function': get_filename,
'test_size': 9,
'all_test': False # 全部作为测试集
},
'CLINIC': {
'index': 6,
'skip': False,
'img_path': '/data/datasets/CTPelvic1K/CTPelvic1K_dataset6_data', # 修改
'label_path': '/data/datasets/CTPelvic1K/CTPelvic1K_dataset6_Anonymized_mask/ipcai2021_dataset6_Anonymized',
# 修改
'patten': ['*.nii.gz'],
'convert_function': get_filename,
'test_size': 21,
'all_test': True # 全部作为测试集
},
'CLINIC-metal': {
'index': 7,
'skip': False,
'img_path': '/data/datasets/CTPelvic1K/CTPelvic1K_dataset7_data', # 修改
'label_path': '/data/datasets/CTPelvic1K/CTPelvic1K_dataset7_mask', # 修改
'patten': ['*.nii.gz'],
'convert_function': get_filename,
'test_size': 14,
'all_test': True # 全部作为测试集
}
}
if __name__ == '__main__':
print(train_dir)
os.makedirs(train_dir, exist_ok=True)
assert 0 == len(os.listdir(train_dir))
print(test_dir)
os.makedirs(test_dir, exist_ok=True)
for dataset_name, v in data_inform.items():
print(dataset_name)
if v['skip']:
continue
index = v['index']
img_path = v['img_path']
label_path = v['label_path']
test_size = v['test_size']
images_list = [] # 所有的图片（包括训练集、测试集）
label_list = glob.glob(f'{label_path}/*_mask_4label.nii.gz') # 标签
for patten in v['patten']:
images_list += glob.glob(os.path.join(img_path, patten))
target_img_list = [] # 有标签的图片
flag = True
for path in images_list:
file_name = v['convert_function'](path)
target_label_path = f'{label_path}/dataset{index}_{file_name}_mask_4label.nii.gz'
target_label_path2 = f'{label_path}/{file_name}_mask_4label.nii.gz'
if not flag:
if target_label_path2 in label_list:
target_img_list.append(path)
elif target_label_path in label_list:
target_img_list.append(path)
else:
if target_label_path2 in label_list:
flag = False
target_img_list.append(path)
# 按转换后的文件名排序，保证对应
target_img_list.sort(key=lambda x: v['convert_function'](x))
label_list.sort(key=lambda x: os.path.basename(x))
if v['all_test']:
test_size = len(label_list)
if len(label_list) > test_size:
image_train_list, image_test_list, label_train_list, label_test_list = train_test_split(target_img_list,
label_list,
test_size=test_size)
else:
image_train_list = []
label_train_list = []
image_test_list = target_img_list
label_test_list = label_list
prefix = '' if flag else f'dataset{index}_'
for image_filename, label_filename in tqdm(zip(image_train_list, label_train_list)):
file_name = v['convert_function'](image_filename)
shutil.copyfile(image_filename, os.path.join(train_dir, f'dataset{index}_{file_name}_data.nii.gz'))
shutil.copyfile(label_filename, os.path.join(train_dir, prefix + os.path.basename(label_filename)))
for image_filename, label_filename in tqdm(zip(image_test_list, label_test_list)):
file_name = v['convert_function'](image_filename)
shutil.copyfile(image_filename, os.path.join(test_dir, f'dataset{index}_{file_name}_data.nii.gz'))
shutil.copyfile(label_filename, os.path.join(test_dir, prefix + os.path.basename(label_filename)))
# check
nii_files_tr_data = subfiles(train_dir, True, None, "_data.nii.gz", True)
nii_files_tr_seg = subfiles(train_dir, True, None, "_mask_4label.nii.gz", True)
nii_files_ts = subfiles(test_dir, True, None, "_data.nii.gz", True)
print('checking')
for img, seg in zip(nii_files_tr_data, nii_files_tr_seg):
img = os.path.basename(img)
seg = os.path.basename(seg)
if img.replace('_data', '_mask_4label') != seg:
print('标签和数据集对不上？？？')
print(img, seg)
break