-
Notifications
You must be signed in to change notification settings - Fork 15
/
Copy pathmain.py
181 lines (164 loc) · 8.13 KB
/
main.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
import os
import time
import numpy as np
import matplotlib.pyplot as plt
import SimpleITK as sitk
import utils
import paths
import deephistreg as dhr
import torch
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
dataset_path = paths.parsed_data_path
### Identity - to create results without any registration (useful for creating the baseline)
# output_path = None # TO DEFINE
# dhr_params = dict()
# dhr_params['segmentation_mode'] = "deep_segmentation"
# dhr_params['initial_rotation'] = False
# dhr_params['affine_registration'] = False
# dhr_params['nonrigid_registration'] = False
# segmentation_params = dict()
# segmentation_params['model_path'] = None # Path to segmentation model
# dhr_params['segmentation_params'] = segmentation_params
# load_masks = False
###
### Seg + Rotation Params - to create results from the initial rotation (useful for creating the affine training dataset)
# output_path = None # TO DEFINE
# dhr_params = dict()
# dhr_params['segmentation_mode'] = "deep_segmentation"
# dhr_params['initial_rotation'] = True
# dhr_params['affine_registration'] = False
# dhr_params['nonrigid_registration'] = False
# initial_rotation_params = dict()
# initial_rotation_params['angle_step'] = 1
# dhr_params['initial_rotation_params'] = initial_rotation_params
# segmentation_params = dict()
# segmentation_params['model_path'] = None # Path to segmentation model
# dhr_params['segmentation_params'] = segmentation_params
# load_masks = False
###
### Seg + Rotation Params + Affine - to create results from the initial rotation + affine registration (useful for creating the nonrigid training dataset)
# output_path = None # TO DEFINE
# dhr_params = dict()
# dhr_params['segmentation_mode'] = "deep_segmentation"
# dhr_params['initial_rotation'] = True
# dhr_params['affine_registration'] = True
# dhr_params['nonrigid_registration'] = False
# initial_rotation_params = dict()
# initial_rotation_params['angle_step'] = 1
# dhr_params['initial_rotation_params'] = initial_rotation_params
# affine_registration_params = dict()
# models_path = paths.models_path
# affine_registration_params['model_path'] = None # TO DEFINE
# affine_registration_params['affine_type'] = "simple"
# dhr_params['affine_registration_params'] = affine_registration_params
# segmentation_params = dict()
# segmentation_params['model_path'] = None # Path to segmentation model
# dhr_params['segmentation_params'] = segmentation_params
# load_masks = False
###
### Seg + Rotation Params + Affine + Nonrigid - to create final nonrigid results
output_path = None # TO DEFINE
dhr_params = dict()
dhr_params['segmentation_mode'] = "deep_segmentation"
dhr_params['initial_rotation'] = True
dhr_params['affine_registration'] = True
dhr_params['nonrigid_registration'] = True
initial_rotation_params = dict()
initial_rotation_params['angle_step'] = 1
dhr_params['initial_rotation_params'] = initial_rotation_params
affine_registration_params = dict()
models_path = None # TO DEFINE
affine_registration_params['model_path'] = None # TO DEFINE
affine_registration_params['affine_type'] = "simple"
dhr_params['affine_registration_params'] = affine_registration_params
nonrigid_registration_params = dict() # Params used during training
nonrigid_registration_params['stride'] = 128
nonrigid_registration_params['patch_size'] = (256, 256)
nonrigid_registration_params['number_of_patches'] = 32
nonrigid_registration_params['num_levels'] = 3
nonrigid_registration_params['inner_iterations_per_level'] = [3, 3, 3]
nonrigid_registration_params['model_path'] = None # TO DEFINE
dhr_params['nonrigid_registration_params'] = nonrigid_registration_params
segmentation_params = dict()
segmentation_params['model_path'] = None # TO DEFINE
dhr_params['segmentation_params'] = segmentation_params
load_masks = False
###
def run():
show = False
ids = range(0, 481)
for current_id in ids:
b_loading = time.time()
current_pair = str(current_id)
if load_masks:
source, target, source_landmarks, target_landmarks, status, source_mask, target_mask = utils.load_pair(current_pair, dataset_path, load_masks=load_masks)
else:
source, target, source_landmarks, target_landmarks, status = utils.load_pair(current_pair, dataset_path, load_masks=load_masks)
print("Current pair: ", current_pair)
print("Status: ", status)
source = torch.from_numpy(source).to(device)
target = torch.from_numpy(target).to(device)
if load_masks:
source_mask = torch.from_numpy(source_mask).to(device)
target_mask = torch.from_numpy(target_mask).to(device)
dhr_params['segmentation_params']['source_mask'] = target_mask
dhr_params['segmentation_params']['target_mask'] = source_mask
e_loading = time.time()
loading_time = e_loading - b_loading
print("Time for loading and memory transfer: ", loading_time)
b_registration = time.time()
target, source, transformed_target, displacement_field, _ = dhr.deephistreg(target, source, device, dhr_params)
e_registration = time.time()
registration_time = e_registration - b_registration
print("Time for registration: ", registration_time)
if show:
plt.figure()
plt.subplot(2, 3, 1)
plt.imshow(source.cpu().numpy(), cmap='gray')
plt.axis('off')
plt.subplot(2, 3, 2)
plt.imshow(transformed_target.cpu().numpy(), cmap='gray')
plt.axis('off')
plt.subplot(2, 3, 3)
plt.imshow(target.cpu().numpy(), cmap='gray')
plt.axis('off')
plt.subplot(2, 3, 4)
plt.imshow(displacement_field[0, :, :].cpu().numpy(), cmap='gray')
plt.axis('off')
plt.subplot(2, 3, 5)
plt.imshow(displacement_field[1, :, :].cpu().numpy(), cmap='gray')
plt.axis('off')
plt.show()
transformed_source_landmarks = utils.transform_landmarks(source_landmarks, displacement_field)
source_save_path = os.path.join(output_path, current_pair, "source.mha")
target_save_path = os.path.join(output_path, current_pair, "target.mha")
transformed_target_save_path = os.path.join(output_path, current_pair, "transformed_target.mha")
source_landmarks_path = os.path.join(output_path, current_pair, "source_landmarks.csv")
transformed_source_landmarks_path = os.path.join(output_path, current_pair, "transformed_source_landmarks.csv")
if status == "training":
target_landmarks_path = os.path.join(output_path, current_pair, "target_landmarks.csv")
if not os.path.isdir(os.path.dirname(source_save_path)):
os.makedirs(os.path.dirname(source_save_path))
sitk.WriteImage(sitk.GetImageFromArray(source.cpu().numpy()), source_save_path)
sitk.WriteImage(sitk.GetImageFromArray(target.cpu().numpy()), target_save_path)
sitk.WriteImage(sitk.GetImageFromArray(transformed_target.cpu().numpy()), transformed_target_save_path)
utils.save_landmarks(source_landmarks, source_landmarks_path)
utils.save_landmarks(transformed_source_landmarks, transformed_source_landmarks_path)
if status == "training":
utils.save_landmarks(target_landmarks, target_landmarks_path)
try:
image_diagonal = np.sqrt(source.shape[0]**2 + source.shape[1]**2)
rtre_initial = utils.calculate_rtre(source_landmarks, target_landmarks, image_diagonal)
rtre_final = utils.calculate_rtre(transformed_source_landmarks, target_landmarks, image_diagonal)
string_to_save = "Initial TRE: " + str(np.median(rtre_initial)) + "\n" + "Resulting TRE: " + str(np.median(rtre_final))
txt_path = os.path.join(output_path, current_pair, "tre.txt")
with open(txt_path, "w") as file:
file.write(string_to_save)
except:
pass
time_to_save = str(registration_time + loading_time)
time_txt_path = os.path.join(output_path, current_pair, "time.txt")
with open(time_txt_path, "w") as file:
file.write(time_to_save)
if __name__ == "__main__":
run()