-
Notifications
You must be signed in to change notification settings - Fork 26
/
build_index.py
355 lines (324 loc) · 14.4 KB
/
build_index.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
import argparse
import os
import time
import numpy as np
import h5py
import glob
import torch
import multiprocessing as mp
import openslide
import copy
import pickle
from collections import OrderedDict
from veb import VEB
from models.vqvae import LargeVectorQuantizedVAE_Encode
from dataset import Mosaic_Bag_FP
from torchvision.models import densenet121
from torchvision import transforms
from torch.utils.data import DataLoader
from tqdm import tqdm
def set_everything(seed):
"""
Function used to set all random seeds
"""
torch.manual_seed(seed)
torch.cuda.manual_seed(seed)
torch.cuda.manual_seed_all(seed) # if you are using multi-GPU.
# np.random.seed(seed) # Numpy module.
# random.seed(seed) # Python random module.
torch.manual_seed(seed)
torch.backends.cudnn.benchmark = False
torch.backends.cudnn.deterministic = True
def save_hdf5(output_path, asset_dict, mode='a'):
"""
Function that used to store hdf5 file chunk by chunk
"""
file = h5py.File(output_path, mode)
for key, val in asset_dict.items():
data_shape = val.shape
if key not in file:
data_type = val.dtype
chunk_shape = (1, ) + data_shape[1:]
maxshape = (None, ) + data_shape[1:]
dset = file.create_dataset(key, shape=data_shape,
maxshape=maxshape, chunks=chunk_shape,
dtype=data_type)
dset[:] = val
else:
dset = file[key]
dset.resize(len(dset) + data_shape[0], axis=0)
dset[-data_shape[0]:] = val
file.close()
return output_path
def to_latent_semantic(latent, codebook_semantic):
"""
Convert the original VQ-VAE latent code by using re-ordered codebook
Input:
latent (64 x 64 np.array): The original latent code from VQ-VAE encoder
codebook_semantic (dict): The dictionary that map old codewords to
the new ones
Output:
latent_semantic: The latent code with new codewords
"""
latent_semantic = np.zeros_like(latent)
for i in range(latent_semantic.shape[0]):
for j in range(latent_semantic.shape[1]):
latent_semantic[i][j] = codebook_semantic[latent[i][j]]
return latent_semantic
def slide_to_index(latent, codebook_semantic, pool_layers, pool=None):
"""
Convert VQ-VAE latent code into an integer
Input:
latent (N x 64 x 64 np array): The latent code from VQ-VAE enecoder
codebook_semantic (128 x 256): The codebook from VQ-VAE encoder
pool_layers (torch.nn.Sequential): A series of pool layers that convert the latent code into an integer
Output:
index (int): An integer index that represents the latent code
"""
if pool is None:
result = to_latent_semantic(latent, codebook_semantic)
feat = torch.from_numpy(np.expand_dims(result, 0))
else:
iterable = [(lt, codebook_semantic) for lt in latent]
result = pool.starmap(to_latent_semantic, iterable)
feat = torch.from_numpy(np.array(result))
num_level = list(range(len(pool_layers) + 1))
level_sum_dict = {level: None for level in num_level}
for level in num_level:
if level == 0:
level_sum_dict[level] = torch.sum(feat, (1, 2)).numpy().astype(float)
else:
feat = pool_layers[level - 1](feat)
level_sum_dict[level] = torch.sum(feat, (1, 2)).numpy().astype(float)
level_power = [0, 0, 1e6, 1e11]
index = 0
for level, power in enumerate(level_power):
if level == 1:
index = copy.deepcopy(level_sum_dict[level])
elif level > 1:
index += level_sum_dict[level] * power
return index
def min_max_binarized(feat):
"""
Min-max algorithm proposed in paper: Yottixel-An Image Search Engine for Large Archives of
Histopathology Whole Slide Images.
Input:
feat (1 x 1024 np.arrya): Features from the last layer of DenseNet121.
Output:
output_binarized (str): A binary code of length 1024
"""
prev = float('inf')
output_binarized = []
for ele in feat:
if ele < prev:
code = 0
output_binarized.append(code)
elif ele >= prev:
code = 1
output_binarized.append(code)
prev = ele
output_binarized = "".join([str(e) for e in output_binarized])
return output_binarized
def compute_latent_features(wsi, mosaic_path, save_path, resolution, transform,
vqvae, batch_size=8, num_workers=4):
"""
Copmute the latent code of input by VQ-VAE encoder
Input:
wsi (openslide objet): The slide to compute
mosaic_path (str): The path that store wsi mosaic
save_path (str): Path to store latent code
resolution (str): The resolution of wsi (e.g., 20x or 40x)
transoform (torch.transforms): The transform applied to image before
feeding into VQ-VAE
vqvae (torch.models): VQ-VAE encoder along with codebook with weight
from the checkpoints
batch_size (int): The number of data processed by VQ-VAE per loop
num_workers (int): Number of cpu used by Dataloader to load the data
Output:
feature list (list): list of vq-vqae latent code of length = #mosaics
in the wsi
"""
dataset = Mosaic_Bag_FP(mosaic_path, wsi, int(resolution[:-1]),
custom_transforms=transform)
dataloader = DataLoader(dataset=dataset, batch_size=batch_size,
num_workers=num_workers, shuffle=False,
pin_memory=True)
count = 0
total = len(dataset)
if total == 0:
return None
features_list = []
mode = 'w'
save_vqvae_path = os.path.join(save_path, 'vqvae', os.path.basename(mosaic_path))
with torch.no_grad():
for mosaic, coord in dataloader:
mosaic = torch.squeeze(mosaic, 1)
mosaic = mosaic.to(device, non_blocking=True)
features = vqvae(mosaic)
features = features.cpu().numpy()
features_list.append(features)
count += features.shape[0]
print("Numebr of latent code processed {}/{}".format(count, total))
asset_dict = {'features': features, 'coords': coord.numpy()}
save_hdf5(save_vqvae_path, asset_dict, mode=mode)
mode = 'a'
return np.concatenate(features_list, 0)
def compute_densenet_features(wsi, mosaic_path, save_path, resolution,
transform, densenet,
batch_size=8, num_workers=4):
"""
Copmute the texture feature (h_{i}) of input by DenseNet121.
Input:
wsi (openslide objet): The slide to compute
mosaic_path (str): The path that store wsi mosaic
save_path (str): Path to store latent code
resolution (str): The resolution of wsi (e.g., 20x or 40x)
transoform (torch.transforms): The transform applied to image before
feeding into DenseNet.
densenet (torch.models): A pretrained Densenet121 model loaded from pytorch
batch_size (int): The number of data processed by VQ-VAE per loop
num_workers (int): Number of cpu used by Dataloader to load the data
Output:
feature list (list): list of binarized texture feature of length = #mosaics
in the wsi
"""
dataset = Mosaic_Bag_FP(mosaic_path, wsi, int(resolution[:-1]),
custom_transforms=transform)
dataloader = DataLoader(dataset=dataset, batch_size=batch_size,
num_workers=num_workers, shuffle=False,
pin_memory=True)
features_list = []
count = 0
total = len(dataset)
save_dense_path = os.path.join(save_path, 'densenet',
os.path.basename(mosaic_path).replace(".h5",".pkl"))
with torch.no_grad():
for mosaic, coord in dataloader:
mosaic = torch.squeeze(mosaic, 1)
mosaic = mosaic.to(device, non_blocking=True)
features = densenet(mosaic)
features = features.cpu().numpy()
features_list.append(features)
count += features.shape[0]
print("Number of dense feature rocessed {}/{}".format(count, total))
features_list = np.squeeze(np.concatenate(features_list, 0))
features_binarized = [min_max_binarized(feat) for feat in features_list]
with open(save_dense_path, 'wb') as handle:
pickle.dump(features_binarized, handle)
return features_binarized
if __name__ == "__main__":
parser = argparse.ArgumentParser(description='Build database of FISH')
parser.add_argument("--mosaic_path", type=str, default="./DATA/MOSAICS/",
help="Path to mosaics")
parser.add_argument("--slide_path", type=str, default="./DATA/WSI",
help="Path to WSIs")
parser.add_argument("--site", type=str, required=True,
help="The anatomic site where the database is built on")
parser.add_argument("--slide_ext", default='.svs',
help="Slide file format")
parser.add_argument("--checkpoint", type=str, default="./checkpoints/model_9.pt",
help="Path to VQ-VAE checkpoint")
parser.add_argument("--codebook_semantic", type=str, default="./checkpoints/codebook_semantic.pt",
help="Path to semantic codebook")
args = parser.parse_args()
# Create the save path of database
if not os.path.exists("DATABASES"):
os.makedirs("DATABASES")
if not os.path.exists(os.path.join("DATABASES")):
os.makedirs(os.path.join("DATABASES"))
save_path_indextree = os.path.join("DATABASES", args.site, "index_tree")
save_path_indexmeta = os.path.join("DATABASES", args.site, "index_meta")
if not os.path.exists(save_path_indextree):
os.makedirs(save_path_indextree)
if not os.path.exists(save_path_indexmeta):
os.makedirs(save_path_indexmeta)
# Set up cpu and device
device = torch.device("cuda:0,1") if torch.cuda.is_available() else torch.device('cpu')
num_workers = mp.cpu_count()
pool = mp.Pool(num_workers)
t_total_start = time.time()
# initialize the database related object
database = {}
key_list = []
# initialize transforms for densenet and vq-vae
transform_densenet = transforms.Compose([transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406],
[0.229, 0.224, 0.225])])
transform_vqvqe = transforms.Compose([transforms.Lambda(lambda x: 2 * transforms.ToTensor()(x) - 1)])
# Load the Densenet
densenet = densenet121(pretrained=True)
densenet = torch.nn.Sequential(*list(densenet.children())[:-1],
torch.nn.AvgPool2d(kernel_size=(32, 32)))
densenet.to(device)
densenet.eval()
# Load the codebook and vq-vae encoder
codebook_semantic = torch.load(args.codebook_semantic)
vqvae = LargeVectorQuantizedVAE_Encode(256, 128)
if torch.cuda.device_count() > 1:
vqvae = torch.nn.DataParallel(vqvae, device_ids=[0, 1])
vqvae_weight_value = torch.load(args.checkpoint)['model']
vqvae_weight_enc = OrderedDict({k: v for k, v in vqvae_weight_value.items()
if 'encoder' in k or 'codebook' in k})
vqvae.load_state_dict(vqvae_weight_enc)
vqvae = vqvae.to(device)
vqvae.eval()
# Initialize pooling layer
pool_layers = [torch.nn.AvgPool2d(kernel_size=(2, 2)),
torch.nn.AvgPool2d(kernel_size=(2, 2)),
torch.nn.AvgPool2d(kernel_size=(2, 2))]
t_enc_start = time.time()
mosaic_all = os.path.join(args.mosaic_path, args.site,
"*", "*", "coord_clean", "*")
total = len(glob.glob(mosaic_all))
count = 0
number_of_mosaic = 0
for mosaic_path in tqdm(glob.glob(mosaic_all)):
print(mosaic_path)
t_start = time.time()
resolution = mosaic_path.split("/")[-3]
diagnosis = mosaic_path.split("/")[-4]
slide_id = os.path.basename(mosaic_path).replace(".h5", "")
slide_path = os.path.join(args.slide_path, args.site, diagnosis,
resolution, slide_id + ".svs")
save_path_latent = os.path.join("./DATA/LATENT/",
args.site, diagnosis, resolution)
if not os.path.exists(save_path_latent):
os.makedirs(os.path.join(save_path_latent, 'vqvae'))
os.makedirs(os.path.join(save_path_latent, 'densenet'))
with h5py.File(mosaic_path, 'r') as hf:
mosaic_coord = hf['coords'][:]
wsi = openslide.open_slide(slide_path)
latent = compute_latent_features(wsi, mosaic_path, save_path_latent,
resolution, transform_vqvqe, vqvae)
dense_feat = compute_densenet_features(wsi, mosaic_path,
save_path_latent, resolution,
transform_densenet, densenet)
slide_index = slide_to_index(latent, codebook_semantic,
pool_layers=pool_layers, pool=pool)
for idx, key in enumerate(slide_index):
tmp = {'slide_name': slide_id, 'dense_binarized': dense_feat[idx],
'x': mosaic_coord[idx][0], 'y': mosaic_coord[idx][1],
'slide_ext': args.slide_ext, 'diagnosis': diagnosis,
'site': args.site}
if key not in database:
database[key] = [tmp]
else:
database[key].append(tmp)
key_list.append(int(key))
count += 1
print("{}/{} Processing slide {} with diagnosis {} takes {}".
format(count, total, slide_id, diagnosis, time.time() - t_start))
number_of_mosaic += len(mosaic_coord)
print("")
print("Encoding takes {}".format(time.time() - t_enc_start))
database_keys = key_list
universe = max(database_keys)
number_of_index = len(database_keys)
print("Universe size of veb tree:", universe)
veb = VEB(universe)
for k in database_keys:
veb.insert(int(k))
with open(os.path.join(save_path_indextree, "veb.pkl"), 'wb') as handle:
pickle.dump(veb, handle)
with open(os.path.join(save_path_indexmeta, "meta.pkl"), 'wb') as handle:
pickle.dump(database, handle)