-
Notifications
You must be signed in to change notification settings - Fork 0
/
test_seg.py
179 lines (147 loc) · 7.98 KB
/
test_seg.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
import argparse
import os
import time
import numpy as np
import torch
import torch.nn.parallel
from torch.utils.data import DataLoader
from torch.autograd import Variable
import torch.nn.functional as F
import datetime
import logging
import yaml
from pathlib import Path
import nibabel as nib
from models.resunet import ResUNet, DAResNet3d
from models.deepmedic import DeepMedic
from losses.loss import FCCELoss, FocalLoss, DiceLoss
from dataprocess.dataloader import TrainDataset, TestDataset
#from dataprocess.dualdata import TestDualDataset
from metrics.DiceEval import diceEval, AverageMeter
import pdb
NII_FOLDER = '/home/jhebu/dataset/CoronaryArtery/challengedata/'
task_path = '/home/jhebu/PycharmProject/ArterySeg_AAAI/experiment/da_resnet34_3d-ce_loss+dice_loss+mse_loss-2019-09-22_18-11/'
PATCH_SIZE = (32,32,32)
def parse_args():
parser = argparse.ArgumentParser('ArterySeg')
parser.add_argument('--pretrain', type=str, default=None,help='whether use pretrain model')
parser.add_argument('--gpu', nargs='+', type=int, default=[0], help='which gpu to select')
return parser.parse_args()
def main(args):
'''CONFIG LOAD'''
config_path = './config.yaml'
config_reader = open(config_path)
cfg = yaml.load(config_reader, Loader=yaml.FullLoader)
model_path = os.path.join(task_path,'models/seg_model_epoch_2672.pth')
pred_path = os.path.join(task_path,'prediction')
if not os.path.isdir(pred_path):
os.mkdir(pred_path)
os.environ["CUDA_VISIBLE_DEVICES"]=','.join(str(i) for i in args.gpu)
np.random.seed(7)
torch.manual_seed(7)
torch.cuda.manual_seed_all(7)
'''model'''
if cfg['MODEL']['NAME'] == 'da_resnet34_3d':
model = DAResNet3d(cfg['MODEL']['NCLASS'], k = 32)
elif cfg['MODEL']['NAME'] == 'da_resunet':
model = DAResUNet(cfg['MODEL']['NCLASS'], k=16)
elif cfg['MODEL']['NAME'] == 'resunet':
model = ResUNet(cfg['MODEL']['NCLASS'], k = 32)
elif cfg['MODEL']['NAME'] == 'deepmedic':
model = DeepMedic(cfg['MODEL']['NCLASS'])
else:
pass
device_ids = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.cuda(device_ids)
model = torch.nn.DataParallel(model)
model.load_state_dict(torch.load(model_path))
model.eval()
model.cuda()
if cfg['MODEL']['NAME'] == 'deepmedic':
TEST_DATASET = TestDualDataset(train_lst=cfg['DATASET']['TRAIN_LIST'],
flip=True, patch_size = PATCH_SIZE)
else:
TEST_DATASET = TestDataset(test_lst=NII_FOLDER+'test.lst', flip=True, patch_size = PATCH_SIZE)
testdataloader = DataLoader(TEST_DATASET, batch_size=1,
shuffle=False,num_workers=int(cfg['TEST']['WORKER']),drop_last = False)
with torch.no_grad():
for subject in TEST_DATASET.subjects:
image_shape = TEST_DATASET.sample(subject)
print("The subject ID is: " + subject)
print("The number of test data is:", len(TEST_DATASET))
full_pred_max = np.zeros(image_shape)
full_pred_min = np.ones(image_shape)
full_heatmap_max = np.zeros(image_shape)
full_heatmap_min = np.ones(image_shape)*50
for batch_id, (data, coord) in enumerate(testdataloader):
if cfg['MODEL']['NAME'] == 'deepmedic':
image, sub_image = data
image, sub_image, mask = Variable(image.float()), Variable(sub_image.float())
image, sub_image, mask = image.cuda(), sub_image.cuda()
seg_pred = model((image,subimage))
else:
image = Variable(data.float())
image = image.cuda()
image_pred = model(image)
seg_pred = image_pred[:,0:2,:,:,:]
heatmap_pred = image_pred[:,2,:,:,:]
seg_pred = F.softmax(seg_pred, dim=1)
seg_pred = seg_pred.cpu().numpy()
heatmap_pred = heatmap_pred.cpu().numpy()
patch_pred_max = full_pred_max[coord[0]:coord[0]+PATCH_SIZE[0],
coord[1]:coord[1]+PATCH_SIZE[1],
coord[2]:coord[2]+PATCH_SIZE[2]]
patch_pred_max = np.maximum(patch_pred_max, seg_pred[0][1])
full_pred_max[coord[0]:coord[0]+PATCH_SIZE[0],
coord[1]:coord[1]+PATCH_SIZE[1],
coord[2]:coord[2]+PATCH_SIZE[2]] = patch_pred_max
patch_pred_min = full_pred_min[coord[0]:coord[0] + PATCH_SIZE[0],
coord[1]:coord[1] + PATCH_SIZE[1],
coord[2]:coord[2] + PATCH_SIZE[2]]
patch_pred_min = np.minimum(patch_pred_min, seg_pred[0][1])
full_pred_min[coord[0]:coord[0] + PATCH_SIZE[0],
coord[1]:coord[1] + PATCH_SIZE[1],
coord[2]:coord[2] + PATCH_SIZE[2]] = patch_pred_min
patch_heatmap_max = full_heatmap_max[coord[0]:coord[0]+PATCH_SIZE[0],
coord[1]:coord[1]+PATCH_SIZE[1],
coord[2]:coord[2]+PATCH_SIZE[2]]
patch_heatmap_max = np.maximum(patch_heatmap_max, heatmap_pred[0])
full_heatmap_max[coord[0]:coord[0]+PATCH_SIZE[0],
coord[1]:coord[1]+PATCH_SIZE[1],
coord[2]:coord[2]+PATCH_SIZE[2]] = patch_heatmap_max
patch_heatmap_min = full_heatmap_min[coord[0]:coord[0] + PATCH_SIZE[0],
coord[1]:coord[1] + PATCH_SIZE[1],
coord[2]:coord[2] + PATCH_SIZE[2]]
patch_heatmap_min = np.minimum(patch_heatmap_min, heatmap_pred[0])
full_heatmap_min[coord[0]:coord[0] + PATCH_SIZE[0],
coord[1]:coord[1] + PATCH_SIZE[1],
coord[2]:coord[2] + PATCH_SIZE[2]] = patch_heatmap_min
full_pred_avg = (full_pred_min + full_pred_max)/2
full_heatmap_avg = (full_heatmap_min + full_heatmap_max)/2
#print(batch_id)
object_path = os.path.join(NII_FOLDER, subject)
img = nib.load(os.path.join(object_path,'image_resample.nii.gz'))
#img = nib.load(os.path.join(NII_FOLDER, '%s.nii.gz' % subject))
full_pred_max = np.transpose(full_pred_max, (1, 2, 0))
full_pred_max = full_pred_max.astype('float32')
affine = img.affine
full_pred_img = nib.Nifti1Image(full_pred_max, affine)
nib.save(full_pred_img, os.path.join(pred_path,'pred_max_'+subject+'.nii.gz'))
full_pred_min = np.transpose(full_pred_min, (1, 2, 0))
full_pred_min = full_pred_min.astype('float32')
affine = img.affine
full_pred_img = nib.Nifti1Image(full_pred_min, affine)
nib.save(full_pred_img, os.path.join(pred_path,'pred_min_'+subject+'.nii.gz'))
full_pred_avg = np.transpose(full_pred_avg, (1, 2, 0))
full_pred_avg = full_pred_avg.astype('float32')
affine = img.affine
full_pred_img = nib.Nifti1Image(full_pred_avg, affine)
nib.save(full_pred_img, os.path.join(pred_path,'pred_avg_'+subject+'.nii.gz'))
full_heatmap_avg = np.transpose(full_heatmap_avg, (1, 2, 0))
full_heatmap_avg = full_heatmap_avg.astype('float32')
affine = img.affine
full_heatmap_img = nib.Nifti1Image(full_heatmap_avg, affine)
nib.save(full_heatmap_img, os.path.join(pred_path,'heatmap_avg_'+subject+'.nii.gz'))
if __name__ == '__main__':
args = parse_args()
main(args)