-
Notifications
You must be signed in to change notification settings - Fork 1
/
baseline.py
128 lines (107 loc) · 4.51 KB
/
baseline.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
import argparse
from multiprocessing import Pool
import os
from time import time
import matplotlib.pyplot as plt
import numpy as np
import torch
from torch.utils.data import Dataset
from utils.data_utils import DatasetHandler, load_mr_scan
from utils import evaluation, utils
class DataPreloader(Dataset):
def __init__(self, paths, img_size, slices_lower_upper):
super().__init__()
self.samples, self.segmentations, self.brain_masks = [], [], []
self.load_to_ram(paths, img_size, slices_lower_upper)
def __len__(self):
return len(self.samples)
@staticmethod
def load_batch(paths, img_size, slices_lower_upper):
samples = []
segmentations = []
for p in paths:
# Samples are shape [1, slices, height, width]
sample, segmentation = load_mr_scan(
p, img_size, equalize=True,
slices_lower_upper=slices_lower_upper
)
samples.append(sample)
segmentations.append(segmentation)
return {
'samples': samples,
'segmentations': segmentations,
}
def load_to_ram(self, paths, img_size, slices_lower_upper):
# Set number of cpus used
num_cpus = os.cpu_count() - 4
# Split list into batches
batches = [list(p) for p in np.array_split(
paths, num_cpus) if len(p) > 0]
# Start multiprocessing
with Pool(processes=num_cpus) as pool:
temp = pool.starmap(
self.load_batch,
zip(batches, [img_size for _ in batches], [slices_lower_upper for _ in batches])
)
# temp = self.load_batch(paths, img_size, slices_lower_upper)
# Collect results
self.samples = [s for t in temp for s in t['samples']]
self.segmentations = [s for t in temp for s in t['segmentations']]
def __getitem__(self, idx):
return self.samples[idx], self.segmentations[idx]
if __name__ == '__main__':
parser = argparse.ArgumentParser()
# Data params
parser.add_argument("--img_size", type=int, default=128,
help="Image resolution")
parser.add_argument("--test_ds", type=str, required=True,
choices=["BraTS", "MSLUB", "WMH", "MSSEG2015"])
parser.add_argument("--weighting", type=str, default="FLAIR",
choices=["T1", "t1", "T2", "t2", "FLAIR", "flair"])
parser.add_argument("--test_prop", type=float, default=1.0,
help="Fraction of data to evaluate on")
parser.add_argument("--slices_lower_upper", nargs='+', type=int,
default=[15, 125],
help="Upper and lower bound for MRI slices. Use "
"[15, 125] for experiment 1 and [84, 88] for "
"experiment 2")
# Logging params
parser.add_argument("--n_images_log", type=int, default=30)
parser.add_argument("--save_dir", type=str, default="./logs/baseline/")
args = parser.parse_args()
args.save_dir = f"{args.save_dir}{args.img_size}_" \
f"{args.slices_lower_upper[0]}-" \
f"{args.slices_lower_upper[1]}/"
# Get train and test paths
ds_handler = DatasetHandler()
paths, _ = ds_handler(args.test_ds, args.weighting)
paths = paths[-int(len(paths) * args.test_prop):]
# Load data to RAM
print("Loading data")
t_data_start = time()
ds = DataPreloader(paths, args.img_size, args.slices_lower_upper)
print(f"Finished loading data in {time() - t_data_start:.2f}s, found {len(ds)} samples.")
anomaly_maps = torch.cat(ds.samples, 0)
segmentations = torch.cat(ds.segmentations, 0)
auroc, aupr, dice, th = evaluation.evaluate(
predictions=anomaly_maps,
targets=segmentations,
# auroc=False,
# auprc=False,
proauc=False,
)
# Binarize anomaly_maps
bin_map = torch.where(anomaly_maps > th, 1., 0.)
# Connected component filtering
bin_map = utils.connected_components_3d(bin_map)
print("Saving some images")
c = (args.slices_lower_upper[1] - args.slices_lower_upper[0]) // 2
images = [
anomaly_maps[:, c][:, None],
bin_map[:, c][:, None],
segmentations[:, c][:, None]
]
titles = ['Anomaly map', 'Binarized map', 'Ground truth']
fig = evaluation.plot_results(images, titles, n_images=args.n_images_log)
os.makedirs(args.save_dir, exist_ok=True)
plt.savefig(f"{args.save_dir}{args.test_ds}_{args.weighting}_samples.png")