-
Notifications
You must be signed in to change notification settings - Fork 6
/
Copy pathnathip.py
458 lines (385 loc) · 19.1 KB
/
nathip.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
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
'''
Handles our HCUnicamp dataset.
'''
import os
import multiprocessing as mp
import queue
import time
from sys import argv
import pickle
import torch
import nibabel as nib
import numpy as np
import glob
import h5py
from scipy.ndimage import zoom
import pandas as pd
from torch.utils import data
from utils import int_to_onehot, viewnii, one_hot
from shutil import copy2
from multiview import MultiViewer
from tqdm import tqdm
external_hd_path = os.path.join("/media", "diedre", "CSF Things", "bruna_manual", "Bruna_imagens_manuais", "Bruna_imagens")
nathip_path = os.path.join("/home", "diedre", "bigdata", "nathip")
nathip_drop_path = os.path.join("/home", "diedre", "Dropbox", "bigdata", "Hippocampus", "Bruna_imagens")
nathip_path = os.path.join("/home", "diedre", "Dropbox", "bigdata", "nathip")
destiny_path = os.path.join("/home", "diedre", "Dropbox", "bigdata", "Hippocampus", "volbrain")
class NatHIP(data.Dataset):
'''
Manual annotations of MNIHIP
'''
KEY_LIST = {'PACIENTES': ['200909251345', '200909301556', '201001060838', '201110100732', '201110211623', '201111180812',
'201201031305', '201302041513', '201305081437', '201305201200', '201307121700', '201307291741',
'201310091054', '201005261625', '201006160953', '201007160717', '201008270834', '201009291601',
'201210260720', '201211070851', '201212210948', '201301251102', '201302200709', '201303081647',
'201407251021', '201411211420', '201504271127', '201506190741', '201507250814', '201508141025',
'201011050840', '201102101331', '201102161644', '201103031408', '201103041056', '201304231333',
'201304260806', '201311190943', '201312060726', '201312161050', '201403210834', '201404020902',
'201404041001', '201404171503', '201004060908', '201004231438', '201105131519', '201108090803',
'201203061157', '201203231514', '201205021303', '201205171644', '201205180955', '201206121402',
'201207300912', '201208081146', '201407111104', '201407180933', '201407250725', '201407250816',
'201405301559', '201402281121', '201409230939', '201305201710', '201005121236', '201505231634',
'200911161134', '201601061306', '201410251404', '201005281730', '201409291101', '201504231331',
'201412181053', '201511051400', '201512020723', '201205160756', '201503041654', '201305240943',
'201607181310', '201005181125', '201504141828', '201006171334', '201505141137', '201210021625',
'201409171639', '201411270832', '201401171639', '201604121545', '201404171340', '201604291324',
'201003291022', '201407141217', '201507081811', '201101070709', '201601061521', '201107211451',
'201602050914', '201407261010', '201111301202', '201110041833', '201501301245', '201409190745',
'201511271334', '201410011758', '201504171317', '201507011003', '201405281449', '201509251203',
'201209100725', '201507080946', '201507170735', '201607201501', '201503110744', '201401241022',
'201408281419', '201110211657', '201401151601', '201103040953', '201507291610', '201410171025',
'201404081103', '201201300731', '201604251403', '201212050815', '201111210932', '201203141314',
'201311181040', '201003021154', '201412050729', '201508191604', '201601061447', '201602171247',
'201607141304', '201507221749', '201606141728'],
'CONTROLES': ['200909011201', '200910211051', '200912020907', '201112211057', '201209051122', '201301301214',
'201307121053', '201005291021', '201008111720', '201009031037', '201210311022', '201211141058',
'201212120845', '201301090923', '201302061146', '201505161428', '201505201642', '201009301418',
'201011241222', '201012011302', '201312041335', '201402011405', '201402121517', '201406261439',
'201003291549', '201110071545', '201201031538', '201003101413', '201106031312', '201602170833',
'201102261238', '201011011335', '201103301351', '201204210836', '201308301233', '201305290834',
'201011241125', '201301301151', '201008061748', '201110190950', '201402251516', '201011011231',
'201011171149', '201102261153', '200910221801', '201207210819', '201008041451', '201005291312',
'201209191114', '201105191533', '201403291615', '201207201405', '201206231613', '201307231611',
'201207201346', '201302191714', '201611211411', '201508111758', '201111251016', '201003131148',
'201605181000', '201104011343', '201003131531', '201104151339']}
# Software corrections to sheet labeling mistakes, last resort
# INCLUDES NOISY SCANS
# REMOVED_LIST = ['201409171639', '201005261625', '201112211057', '201011241125', '201411270832',
# '201107211451', '201511051400', '201111301202', '201504231331', '201305201710',
# '201501301245', '201404041001', '201105131519', '201404171340', '201005291021']
# Only big mistakes
REMOVED_LIST = ["201409171639", "201112211057", "201011241125", "201105131519", "201404171340", "201005291021",
"201404171340", "201602170833", # no tag file
"200912020907", "200910211051"] # wrong segs
def __init__(self, **kwargs):
super(NatHIP, self).__init__()
# Sanitize mandatory args
assert kwargs["group"] in ['all', 'CONTROLES', 'PACIENTES']
assert kwargs["orientation"] in [None, "sagital", "coronal", "axial"]
assert kwargs["mode"] in ["all", "train", "validation", "test"]
assert kwargs["orientation"] is None, "slices not supported yet"
# Extract kwargs, define properties
self.group = kwargs["group"]
self.orientation = kwargs["orientation"]
self.mode = kwargs["mode"]
self.e2d = kwargs["e2d"]
fold = kwargs["fold"]
self.return_onehot = kwargs["return_onehot"]
self.folder = kwargs["folder"] if "folder" in kwargs else nathip_path
self.return_fname = kwargs["return_fname"] if "return_fname" in kwargs else False
self.verbose = kwargs["verbose"] if "verbose" in kwargs else False
self.transform = kwargs["transform"] if "transform" in kwargs else None
self.reconstruction_orientations = ["sagital", "coronal", "axial"] # compatibility sake
# Build file list
if self.group == "all":
self.subjects = NatHIP.KEY_LIST["PACIENTES"] + NatHIP.KEY_LIST["CONTROLES"]
else:
self.subjects = NatHIP.KEY_LIST[self.group]
subject_len = len(self.subjects)
train_len, test_len = int(subject_len*0.7), int(subject_len*0.2)
# Fold Selection
if fold in range(1, 6):
self.train_val_set = self.subjects[:(fold-1)*test_len] + self.subjects[fold*test_len:]
self.train_set = self.train_val_set[:train_len]
self.val_set = self.train_val_set[train_len:]
self.test_set = self.subjects[(fold-1)*test_len:fold*test_len]
else:
print("WARNING: Fold not selected! >.<")
if self.mode == "train":
self.subjects = self.train_set
elif self.mode == "validation":
self.subjects = self.val_set
elif self.mode == "test":
self.subjects = self.test_set
elif self.mode == "all":
# No changes to subjects
pass
# Software workaround wrong dataset
for toremove in NatHIP.REMOVED_LIST:
if toremove in self.subjects:
self.subjects.remove(toremove)
# Get exam paths from selected subjects
self.keys = self.subjects
# Add slices to keys if in slice mode
if self.orientation is not None:
raise NotImplementedError("Slice mode not implemented yet")
print("NatHIP initialized with {}".format(kwargs))
print(self.keys)
print("{} volumes".format(len(self.keys)))
def __len__(self):
'''
Returns length of the dataset
'''
return len(self.keys)
def __getitem__(self, x):
'''
Decompress saved item and returns it
'''
subject = self.keys[x]
if self.verbose:
print(subject)
if self.orientation is None:
h5file = h5py.File(os.path.join(self.folder, "3d_h5dataset", "h5nathip.hdf5"), 'r')
data_source = h5file[subject]
img = data_source["img"][:]
else:
# TODO update to h5py slices
raise NotImplementedError("Slices not implemented")
if self.e2d:
center_data = data_source["img"]
slice_number = int(os.path.basename(subject).split('.')[0][1:])
post_slice = os.path.join(os.path.dirname(subject), self.orientation[0] + str(slice_number + 1) + ".npz")
pre_slice = os.path.join(os.path.dirname(subject), self.orientation[0] + str(slice_number - 1) + ".npz")
if os.path.exists(post_slice):
post_data = np.load(post_slice)["img"]
else:
post_data = center_data
if os.path.exists(pre_slice):
pre_data = np.load(pre_slice)["img"]
else:
pre_data = center_data
img = np.zeros((3, center_data.shape[0], center_data.shape[1]), dtype=center_data.dtype)
img[0] = pre_data
img[1] = center_data
img[2] = post_data
else:
img = data_source["img"][:]
tgt = data_source["tgt"][:]
h5file.close()
if self.return_onehot:
tgt = int_to_onehot(tgt)
assert one_hot(torch.from_numpy(tgt).unsqueeze(0))
if self.transform is not None:
img, tgt = self.transform(img, tgt)
if self.return_fname:
return img, tgt, os.path.basename(subject)
else:
return img, tgt
def get_volids(self):
return self.subjects
def get_by_name(self, name):
for i in range(len(self)):
if name in self.keys[i]:
return self[i]
def get_group():
if '-p' in argv:
group = "PACIENTES"
elif '-c' in argv:
group = "CONTROLES"
else:
group = "all"
print("Selected NatHIP group: {} group".format(group))
return group
def display_dataset(dataset, exit_sign):
'''
GUI Process
'''
while True:
data = dataset.get()
if data is None:
return
else:
volume, fname = data
retcode = MultiViewer(volume, window_name=fname).display()
if retcode == 'ESC':
print("ESC captured.")
exit_sign.value = 1
return
def try_put(data, q, exit_sign, wait=0.1):
success = False
while not success:
if exit_sign.value > 0:
q.close()
q.join_thread()
return 'QUIT'
try:
q.put_nowait(data)
success = True
except queue.Full:
time.sleep(wait)
return 'OK'
def checkfiles(save_npz=False):
dataset = mp.Queue(maxsize=20)
exit_sign = mp.Value('i', 0)
displayer = mp.Process(target=display_dataset, args=(dataset, exit_sign))
displayer.start()
for folder in glob.glob(os.path.join(nathip_drop_path, "IC2")):
print("Walking thorugh folder {}".format(folder))
for fpath in glob.glob(os.path.join(folder, "*.mnc")):
print("\nReading {}...".format(fpath))
tag_path = fpath.split('.')[0] + ".tag"
# Read mnc file
img = nib.load(fpath)
img_data = img.get_fdata()
img_data = (img_data - img_data.min()) / (img_data.max() - img_data.min())
img_affine = img.get_affine()
tgt_data = np.zeros_like(img_data)
# Read line by line points of tag file
pts = []
try:
with open(tag_path) as tagfile:
line = tagfile.readline()
while line:
line_tokens = line.split(' ')
try:
point = np.array([int(line_tokens[1]), int(line_tokens[2]), int(line_tokens[3])], dtype=np.int)
# tgt_data[point[0], point[1], point[2]] = 1
pts.append(point)
except Exception:
pass
line = tagfile.readline()
except FileNotFoundError as fnfe:
print(fnfe)
print("Skipping {}".format(fpath))
continue
npts = len(pts)
print("Npoints: {}".format(npts))
if npts < 1000:
print("Skipping {} because segmentation is empty/weird".format(fpath))
continue
pts = nib.affines.apply_affine(img_affine, pts).astype(np.int)
for pt in pts:
tgt_data[pt[0], pt[1], pt[2]] = 1.0
if save_npz:
np.savez_compressed(fpath.split('.')[0] + ".npz", img=img_data, tgt=tgt_data, onehot=int_to_onehot(tgt_data))
else:
# Display results
cube_side = 200
volume_shape = img_data.shape
zoom_factors = (cube_side/volume_shape[0], cube_side/volume_shape[1], cube_side/volume_shape[2])
display = zoom(img_data, zoom_factors, order=2) + zoom(tgt_data, zoom_factors, order=0)
display[display > 1] = 1
data = (display, os.path.basename(fpath).split('.')[0])
retcode = try_put(data, dataset, exit_sign)
if retcode == 'QUIT':
print("Quitting early.")
displayer.join()
exit()
dataset.put(None)
dataset.close()
dataset.join_thread()
displayer.join()
def process_csvs(operation=None):
'''
Move manual segmentations to correct folder
Intended structure
Controls - Patients
Foldernames with patient abbrv
npz with full vol (date.npz), tag mask, one hot mask
'''
sjs = [pd.read_excel(os.path.join(nathip_path, "IC1_Editado.ods"), header=None, engine="odf").values,
pd.read_excel(os.path.join(nathip_path, "IC2_Editado.ods"), header=None, engine="odf").values,
pd.read_excel(os.path.join(nathip_path, "IC3_Editado.ods"), header=None, engine="odf").values]
foundpair = 0
os.makedirs("data/CONTROLES", exist_ok=True)
os.makedirs("data/PACIENTES", exist_ok=True)
# Sheet 1
key_list = {'PACIENTES': [], 'CONTROLES': []}
bruna_hash = []
for i, sj in enumerate(sjs):
for j, row in enumerate(sj):
if i == 0:
name = row[0]
status = row[1]
elif i == 1 or i == 2:
name = row[1]
if name[0] == "'":
name = name[1:-1]
status = row[2]
tokens = name.split('-')
date = ''.join(tokens[:5])
print(name, date, status)
if status == 'P':
dst_path = "data/PACIENTES/{}".format(date)
key_list["PACIENTES"].append(date)
elif status == 'C':
dst_path = "data/CONTROLES/{}".format(date)
key_list["CONTROLES"].append(date)
else:
continue
if operation != "key_list":
fmt = ".npz" if operation == "make_copy" else ".mnc"
glob_string = os.path.join(nathip_drop_path, "IC{}".format(i + 1), "*_{}{}".format(j+1, fmt))
print("Source: {}".format(glob_string), "Dest: {}".format(dst_path))
if operation is None:
print("Not copied, just testing. Press anything to continue.")
input("")
else:
try:
src = glob.glob(glob_string)[0]
foundpair += 1
except IndexError:
print("IC{} Bruna {} didnt have an npz".format(i + 1, j+1))
continue
if operation == "make_copy":
os.makedirs(dst_path, exist_ok=True)
dst = os.path.join(dst_path, date + ".npz")
copy2(src, dst)
print("Copied {} -> {}".format(src, dst))
elif operation == "make_hash":
original_name = os.path.splitext(os.path.basename(src))[0]
for items in bruna_hash:
if items[0] == original_name or items[1] == date:
raise ValueError("Duplicated entry in hash, something is wrong")
bruna_hash.append((original_name, date))
print("{}: {}".format(original_name, date))
print("Saveds {} hashes".format(len(bruna_hash)))
print("Key list: {}".format(key_list))
print("Found pair for {} brunas!".format(foundpair))
if operation == "make_hash":
save_hash = os.path.join("data", "bruna_hash.pkl")
print("Saving hash in {}".format(save_hash))
with open(save_hash, "wb") as save_pkl:
pickle.dump(bruna_hash, save_pkl)
if __name__ == "__main__":
if "check" in argv:
checkfiles(save_npz=True)
elif "process" in argv:
operation = None
if "-op" in argv:
try:
operation = argv[argv.index("-op") + 1]
except IndexError:
print("When using -op give something to do.")
process_csvs(operation=operation)
elif "display" in argv:
print("Testing all modes and folds...")
for mode in ["train", "validation", "test"]:
for fold in range(1, 6):
print(mode, fold)
db = NatHIP(group="all", return_fname=True, mode=mode, orientation=None, fold=fold, verbose=True, e2d=False,
return_onehot=False)
print(db.subjects)
print('*'*10)
for group in ["CONTROLES", "PACIENTES"]:
db = NatHIP(group=group, return_fname=True, mode="all", orientation=None, fold=None, verbose=True, e2d=False,
return_onehot=False)
print(len(db))
itc = iter(db)
for img, tgt, fname in tqdm(itc):
if "load_test" in argv:
continue
viewnii(img, mask=tgt, wait=1, id=group, label=fname,
border_only=False, rotate=0)
else:
print("No args given.")