-
Notifications
You must be signed in to change notification settings - Fork 2
/
visualize.py
207 lines (161 loc) · 10.2 KB
/
visualize.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
import imp
import os
import PIL
from numpy.core.fromnumeric import repeat
from scipy.io import loadmat
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.animation as animation
import seaborn as sns
from PIL import Image
from scipy.ndimage import zoom
from torch import prelu
from torchvision.transforms import ToTensor
import io
def func_findParentPath(org_path,num_itr=1):
parentPath = org_path
for i in range(num_itr):
parentPath = os.path.abspath(os.path.join(parentPath,os.pardir))
return parentPath
def func_plotEndoEpiArea(endo_area, epi_area):
plt.figure()
plt.xlabel('Frame number')
plt.ylabel('Area')
plt.xticks(np.arange(1, endo_area.shape[0]+1, 1.0))
plt.plot(np.arange(1,endo_area.shape[0]+1), endo_area, label='Endocardium')
plt.plot(np.arange(1,epi_area.shape[0]+1), epi_area, label='Epicardium')
plt.legend()
plt.show()
def func_plotEndoEpiMaskOverlay(image, epi_mask, endo_mask, numOfFrame):
plt.figure()
for frame_index in range(numOfFrame):
plt.imshow(image[:,:,frame_index], cmap='gray')
plt.imshow(epi_mask[:,:,frame_index], alpha=0.3)
plt.imshow(endo_mask[:,:,frame_index], alpha=0.3)
plt.title('Frame {}'.format(str(frame_index+1)))
plt.show()
def func_saveEachFrame(dest_path, image, endo_mask, epi_mask, numOfFrame):
for frame_index in range(numOfFrame):
plt.axis('off')
plt.tight_layout()
plt.imsave(os.path.join(dest_path, 'image_frame{:d}.png'.format(frame_index+1)), image[:,:,frame_index], cmap='gray')
plt.imsave(os.path.join(dest_path, 'epi_mask_frame{:d}.png'.format(frame_index+1)), epi_mask[:,:,frame_index], cmap='gray')
plt.imsave(os.path.join(dest_path, 'endo_mask_frame{:d}.png'.format(frame_index+1)), endo_mask[:,:,frame_index], cmap='gray')
def func_transformDVF(DVF):
# The DVF produced by the network for F.grid_sample() seems to need some transform to match the image location for plot
# horizontal flip
DVF = np.flip(DVF, axis=1)
# DVF = np.flip(DVF)
return DVF
def func_plotDVF(mov, fix, moved, DVF, n_step=4, req_reg = 'False'):
assert req_reg in ['False', 'optflow', 'bspline']
if req_reg == 'False':
grid_x, grid_y = np.meshgrid(np.arange(0, DVF.shape[1], n_step), np.arange(0, DVF.shape[2], n_step))
if req_reg == 'optflow' or req_reg == 'bspline':
grid_x, grid_y = np.meshgrid(np.arange(0, DVF.shape[0], n_step), np.arange(0, DVF.shape[1], n_step))
# print(grid_x.shape)
if moved.ndim == 3:
moved = moved[:,:,0]
# normalize for plot
thres_mov, thres_fix, thres_moved = np.float(np.max(mov)), np.float(np.max(fix)), np.float(np.max(moved))
fig, axes = plt.subplots(1, 6, figsize=(30,4))
# transform the DVF for plot purpose
if req_reg == 'False':
DVF = func_transformDVF(DVF)
# fig.set_size_inches(24, 4)
axes[0].imshow(mov, cmap='gray')
# axes[0].set_yticklabels([])
# axes[0].set_xticklabels([])
axes[0].set_xlabel('moving')
axes[1].imshow(fix, cmap='gray')
axes[1].set_xlabel('fixed')
axes[2].imshow(moved, cmap='gray')
axes[2].set_xlabel('moved')
axes[3].imshow(mov/thres_mov - fix/thres_fix, cmap='gray')
axes[3].set_xlabel('diff before')
axes[4].imshow(moved/thres_moved - fix/thres_fix, cmap='gray')
axes[4].set_xlabel('diff after')
if req_reg == 'False':
axes[5].quiver(grid_x, grid_y, DVF[0,::n_step,::n_step], DVF[1, ::n_step,::n_step]) # make the vector field more sparse
if req_reg == 'optflow':
# print(grid_x.shape)
# print(grid_y.shape)
# print(DVF.shape)
axes[5].quiver(grid_x, grid_y, DVF[::n_step,::n_step,0], DVF[::n_step,::n_step,1])
if req_reg == 'bspline':
axes[5].quiver(grid_x, grid_y, DVF[::n_step,::n_step,1], DVF[::n_step,::n_step,0])
axes[5].set_xlabel('DVF')
def func_ACDC2D_training_visual_check(visual_check_path, batch_idx, mov, fix, manual_moved, mov_LV_mask, mov_RV_mask, mov_epi_mask, fix_LV_mask, fix_RV_mask, fix_epi_mask, moved_LV_mask, moved_RV_mask, moved_epi_mask):
mov_slice = mov.cpu().numpy()[0,0,:,:]
fix_slice = fix.cpu().numpy()[0,0,:,:]
moved_slice = manual_moved.detach().cpu().numpy()[0,0,:,:]
mov_LV_slice = mov_LV_mask.cpu().numpy()[0,0,:,:]
mov_RV_slice = mov_RV_mask.cpu().numpy()[0,0,:,:]
mov_epi_slice = mov_epi_mask.cpu().numpy()[0,0,:,:]
fix_LV_slice = fix_LV_mask.cpu().numpy()[0,0,:,:]
fix_RV_slice = fix_RV_mask.cpu().numpy()[0,0,:,:]
fix_epi_slice = fix_epi_mask.cpu().numpy()[0,0,:,:]
moved_LV_slice = moved_LV_mask.detach().cpu().numpy()[0,0,:,:]
moved_RV_slice = moved_RV_mask.detach().cpu().numpy()[0,0,:,:]
moved_epi_slice = moved_epi_mask.detach().cpu().numpy()[0,0,:,:]
plt.imsave(os.path.join(visual_check_path, 'idx_{}_mov_slice.png'.format(str(batch_idx))), mov_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_fix_slice.png'.format(str(batch_idx))), fix_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_moved_slice.png'.format(str(batch_idx))), moved_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_mov_LV_slice.png'.format(str(batch_idx))), mov_LV_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_fix_LV_slice.png'.format(str(batch_idx))), fix_LV_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_moved_LV_slice.png'.format(str(batch_idx))), moved_LV_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_mov_RV_slice.png'.format(str(batch_idx))), mov_RV_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_fix_RV_slice.png'.format(str(batch_idx))), fix_RV_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_moved_RV_slice.png'.format(str(batch_idx))), moved_RV_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_mov_epi_slice.png'.format(str(batch_idx))), mov_epi_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_fix_epi_slice.png'.format(str(batch_idx))), fix_epi_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_moved_epi_slice.png'.format(str(batch_idx))), moved_epi_slice)
def func_ACDC2D_image_check_dict(mov, fix, manual_moved, mov_LV_mask, mov_RV_mask, mov_epi_mask, fix_LV_mask, fix_RV_mask, fix_epi_mask, moved_LV_mask, moved_RV_mask, moved_epi_mask):
mov_slice = mov.cpu().numpy()[0,:,:,:]
fix_slice = fix.cpu().numpy()[0,:,:,:]
moved_slice = manual_moved.detach().cpu().numpy()[0,:,:,:]
# mov_LV_slice = mov_LV_mask.cpu().numpy()[0,:,:,:]
# mov_RV_slice = mov_RV_mask.cpu().numpy()[0,:,:,:]
# mov_epi_slice = mov_epi_mask.cpu().numpy()[0,:,:,:]
fix_LV_slice = fix_LV_mask.cpu().numpy()[0,:,:,:]
fix_RV_slice = fix_RV_mask.cpu().numpy()[0,:,:,:]
fix_epi_slice = fix_epi_mask.cpu().numpy()[0,:,:,:]
moved_LV_slice = moved_LV_mask.detach().cpu().numpy()[0,:,:,:]
moved_RV_slice = moved_RV_mask.detach().cpu().numpy()[0,:,:,:]
moved_epi_slice = moved_epi_mask.detach().cpu().numpy()[0,:,:,:]
return {'mov_slice': mov_slice, 'fix_slice': fix_slice, 'moved_slice': moved_slice,
'diff_LV': moved_LV_slice/np.float(np.max(moved_LV_slice)) - fix_LV_slice/np.float(np.max(fix_LV_slice)),
'diff_RV': moved_RV_slice/np.float(np.max(moved_RV_slice)) - fix_RV_slice/np.float(np.max(fix_RV_slice)),
'diff_epi': moved_epi_slice/np.float(np.max(moved_epi_slice)) - fix_epi_slice/np.float(np.max(fix_epi_slice))}
def func_LVQuant_training_visual_check(visual_check_path, batch_idx, mov, fix, manual_moved, mov_endo_mask, mov_epi_mask, fix_endo_mask, fix_epi_mask, moved_endo_mask, moved_epi_mask):
mov_slice = mov.cpu().numpy()[0,0,:,:]
fix_slice = fix.cpu().numpy()[0,0,:,:]
moved_slice = manual_moved.detach().cpu().numpy()[0,0,:,:]
mov_endo_slice = mov_endo_mask.cpu().numpy()[0,0,:,:]
mov_epi_slice = mov_epi_mask.cpu().numpy()[0,0,:,:]
fix_endo_slice = fix_endo_mask.cpu().numpy()[0,0,:,:]
fix_epi_slice = fix_epi_mask.cpu().numpy()[0,0,:,:]
moved_endo_slice = moved_endo_mask.detach().cpu().numpy()[0,0,:,:]
moved_epi_slice = moved_epi_mask.detach().cpu().numpy()[0,0,:,:]
plt.imsave(os.path.join(visual_check_path, 'idx_{}_mov_slice.png'.format(str(batch_idx))), mov_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_fix_slice.png'.format(str(batch_idx))), fix_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_moved_slice.png'.format(str(batch_idx))), moved_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_mov_endo_slice.png'.format(str(batch_idx))), mov_endo_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_fix_endo_slice.png'.format(str(batch_idx))), fix_endo_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_moved_endo_slice.png'.format(str(batch_idx))), moved_endo_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_mov_epi_slice.png'.format(str(batch_idx))), mov_epi_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_fix_epi_slice.png'.format(str(batch_idx))), fix_epi_slice)
plt.imsave(os.path.join(visual_check_path, 'idx_{}_moved_epi_slice.png'.format(str(batch_idx))), moved_epi_slice)
def func_LVQuant_image_check_dict(mov, fix, manual_moved, mov_endo_mask, mov_epi_mask, fix_endo_mask, fix_epi_mask, moved_endo_mask, moved_epi_mask):
mov_slice = mov.cpu().numpy()[0,:,:,:]
fix_slice = fix.cpu().numpy()[0,:,:,:]
moved_slice = manual_moved.detach().cpu().numpy()[0,:,:,:]
# mov_endo_slice = mov_endo_mask.cpu().numpy()[0,:,:,:]
# mov_epi_slice = mov_epi_mask.cpu().numpy()[0,:,:,:]
fix_endo_slice = fix_endo_mask.cpu().numpy()[0,:,:,:]
fix_epi_slice = fix_epi_mask.cpu().numpy()[0,:,:,:]
moved_endo_slice = moved_endo_mask.detach().cpu().numpy()[0,:,:,:]
moved_epi_slice = moved_epi_mask.detach().cpu().numpy()[0,:,:,:]
return {'mov_slice': mov_slice, 'fix_slice': fix_slice, 'moved_slice': moved_slice,
'diff_endo': moved_endo_slice/np.float(np.max(moved_endo_slice)) - fix_endo_slice/np.float(np.max(fix_endo_slice)),
'diff_epi': moved_epi_slice/np.float(np.max(moved_epi_slice)) - fix_epi_slice/np.float(np.max(fix_epi_slice))}