/
l2r_2021_convexAdam_task3_docker.py
233 lines (174 loc) · 10.1 KB
/
l2r_2021_convexAdam_task3_docker.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
import numpy as np
import nibabel as nib
import struct
import scipy.ndimage
from scipy.ndimage import zoom, map_coordinates
import torch
import torch.nn as nn
import torch.nn.functional as F
print(torch.__version__)
import sys
import time
sys.path.append('voxelmorph/pytorch/')
import losses
print(losses.mind_loss)
def gpu_usage():
print('gpu usage (current/max): {:.2f} / {:.2f} GB'.format(torch.cuda.memory_allocated()*1e-9, torch.cuda.max_memory_allocated()*1e-9))
H = 160
W = 192
D = 224
def dice_coeff(outputs, labels, max_label):
dice = torch.FloatTensor(max_label-1).fill_(0)
for label_num in range(1, max_label):
iflat = (outputs==label_num).view(-1).float()
tflat = (labels==label_num).view(-1).float()
intersection = torch.mean(iflat * tflat)
dice[label_num-1] = (2. * intersection) / (1e-8 + torch.mean(iflat) + torch.mean(tflat))
return dice
identity = np.stack(np.meshgrid(np.arange(H), np.arange(W), np.arange(D), indexing='ij'))
#print(fixed.shape)
#correlation layer: dense discretised displacements to compute SSD cost volume with box-filter
def correlate(mind_fix,mind_mov,disp_hw,grid_sp):
torch.cuda.synchronize()
C_mind = mind_fix.shape[1]
t0 = time.time()
with torch.no_grad():
mind_unfold = F.unfold(F.pad(mind_mov,(disp_hw,disp_hw,disp_hw,disp_hw,disp_hw,disp_hw)).squeeze(0),disp_hw*2+1)
mind_unfold = mind_unfold.view(C_mind,-1,(disp_hw*2+1)**2,W//grid_sp,D//grid_sp)
ssd = torch.zeros((disp_hw*2+1)**3,H//grid_sp,W//grid_sp,D//grid_sp,dtype=mind_fix.dtype, device=mind_fix.device)#.cuda().half()
ssd_argmin = torch.zeros(H//grid_sp,W//grid_sp,D//grid_sp).long()
with torch.no_grad():
for i in range(disp_hw*2+1):
mind_sum = (mind_fix.permute(1,2,0,3,4)-mind_unfold[:,i:i+H//grid_sp]).abs().sum(0,keepdim=True)
ssd[i::(disp_hw*2+1)] = F.avg_pool3d(mind_sum.transpose(2,1),3,stride=1,padding=1).squeeze(1)
ssd = ssd.view(disp_hw*2+1,disp_hw*2+1,disp_hw*2+1,H//grid_sp,W//grid_sp,D//grid_sp).transpose(1,0).reshape((disp_hw*2+1)**3,H//grid_sp,W//grid_sp,D//grid_sp)
ssd_argmin = torch.argmin(ssd,0)#
#ssd = F.softmax(-ssd*1000,0)
torch.cuda.synchronize()
t1 = time.time()
print(t1-t0,'sec (ssd)')
gpu_usage()
return ssd,ssd_argmin
#solve two coupled convex optimisation problems for efficient global regularisation
def coupled_convex(ssd,ssd_argmin,disp_mesh_t,grid_sp):
disp_soft = F.avg_pool3d(disp_mesh_t.view(3,-1)[:,ssd_argmin.view(-1)].reshape(1,3,H//grid_sp,W//grid_sp,D//grid_sp),3,padding=1,stride=1)
coeffs = torch.tensor([0.003,0.01,0.03,0.1,0.3,1])
for j in range(6):
ssd_coupled_argmin = torch.zeros_like(ssd_argmin)
with torch.no_grad():
for i in range(H//grid_sp):
coupled = ssd[:,i,:,:]+coeffs[j]*(disp_mesh_t-disp_soft[:,:,i].view(3,1,-1)).pow(2).sum(0).view(-1,W//grid_sp,D//grid_sp)
ssd_coupled_argmin[i] = torch.argmin(coupled,0)
#print(coupled.shape)
disp_soft = F.avg_pool3d(disp_mesh_t.view(3,-1)[:,ssd_coupled_argmin.view(-1)].reshape(1,3,H//grid_sp,W//grid_sp,D//grid_sp),3,padding=1,stride=1)
return disp_soft
#enforce inverse consistency of forward and backward transform
def inverse_consistency(disp_field1s,disp_field2s,iter=20):
#factor = 1
B,C,H,W,D = disp_field1s.size()
#make inverse consistent
with torch.no_grad():
disp_field1i = disp_field1s.clone()
disp_field2i = disp_field2s.clone()
identity = F.affine_grid(torch.eye(3,4).unsqueeze(0),(1,1,H,W,D)).permute(0,4,1,2,3).to(disp_field1s.device).to(disp_field1s.dtype)
for i in range(iter):
disp_field1s = disp_field1i.clone()
disp_field2s = disp_field2i.clone()
disp_field1i = 0.5*(disp_field1s-F.grid_sample(disp_field2s,(identity+disp_field1s).permute(0,2,3,4,1)))
disp_field2i = 0.5*(disp_field2s-F.grid_sample(disp_field1s,(identity+disp_field2s).permute(0,2,3,4,1)))
return disp_field1i,disp_field2i
def combineDeformation3d(disp_1st,disp_2nd,identity):
disp_composition = disp_2nd + F.grid_sample(disp_1st,disp_2nd.permute(0,2,3,4,1)+identity)
return disp_composition
grid_sp = 2
disp_hw = 3
nu = 438
fixed = torch.from_numpy(nib.load('L2R2021/Task3/skull_stripped/nnunet/img0'+str(nu)+'.nii.gz').get_fdata()).float()
moving = torch.from_numpy(nib.load('L2R2021/Task3/skull_stripped/nnunet/img0'+str(nu+1)+'.nii.gz').get_fdata()).float()
weight = 1/(torch.bincount(fixed.long().reshape(-1))+torch.bincount(moving.long().reshape(-1))).float().pow(.3)
weight /= weight.mean()
print(weight)
case_time = torch.zeros(38)
torch.cuda.synchronize()
t0_ = time.time()
for nu in range(1,39):
torch.cuda.synchronize()
t0 = time.time()
fixed = torch.from_numpy(nib.load('/data_supergrover2/heinrich/nnUNet_predict/L2R_2021_Task3_test/img'+str(nu).zfill(4)+'_0000.nii.gz').get_fdata()).float()
# mindssc_fix_ = 10*(F.one_hot(fixed.cuda().view(1,H,W,D).long()).float().permute(0,4,1,2,3).contiguous()*weight.view(1,-1,1,1,1).cuda()).half()
#RuntimeError: The size of tensor a (2) must match the size of tensor b (36) at non-singleton dimension 1
moving = torch.from_numpy(nib.load('/data_supergrover2/heinrich/nnUNet_predict/L2R_2021_Task3_test/img'+str(nu+1).zfill(4)+'_0000.nii.gz').get_fdata()).float()
#/share/data_supergrover1/hansen/temp/nnUNet/nnUNet_results/nnUNet/3d_fullres/Task509_OASIS
#OASIS001_img.nii.gz
#/data_supergrover2/heinrich/nnUNet_predict/L2R_Task3_Test/nnunet/img
fixed_seg = torch.from_numpy(nib.load('L2R2021/Task3/Lasse_nnUNet/img'+str(nu).zfill(4)+'.nii.gz').get_fdata()).float()
moving_seg = torch.from_numpy(nib.load('L2R2021/Task3/Lasse_nnUNet/img'+str(nu+1).zfill(4)+'.nii.gz').get_fdata()).float()
#moving_seg = torch.from_numpy(nib.load('/data_supergrover2/heinrich/nnUNet_predict/L2R_Task3_Test/nnunet/img'+str(nu+1).zfill(4)+'.nii.gz').get_fdata()).float()
with torch.no_grad():
mindssc_fix_ = 10*(F.one_hot(fixed_seg.cuda().view(1,H,W,D).long()).float().permute(0,4,1,2,3).contiguous()*weight.view(1,-1,1,1,1).cuda()).half()
mindssc_mov_ = 10*(F.one_hot(moving_seg.cuda().view(1,H,W,D).long()).float().permute(0,4,1,2,3).contiguous()*weight.view(1,-1,1,1,1).cuda()).half()
mind_fix_ = F.avg_pool3d(mindssc_fix_,grid_sp,stride=grid_sp)
mind_mov_ = F.avg_pool3d(mindssc_mov_,grid_sp,stride=grid_sp)
ssd,ssd_argmin = correlate(mind_fix_,mind_mov_,disp_hw,grid_sp)
disp_mesh_t = F.affine_grid(disp_hw*torch.eye(3,4).cuda().half().unsqueeze(0),(1,1,disp_hw*2+1,disp_hw*2+1,disp_hw*2+1),align_corners=True).permute(0,4,1,2,3).reshape(3,-1,1)
disp_soft = coupled_convex(ssd,ssd_argmin,disp_mesh_t,grid_sp)
del ssd,mind_fix_,mind_mov_
print(disp_soft.shape)
#del ssd
#del disp_mesh_t
torch.cuda.empty_cache()
gpu_usage()
disp_lr = F.interpolate(disp_soft*grid_sp,size=(H//2,W//2,D//2),mode='trilinear',align_corners=False)
#disp_soft*grid_sp/2#
grid_sp = 2
#extract one-hot patches
torch.cuda.synchronize()
t0 = time.time()
with torch.no_grad():
mind_fix_ = F.avg_pool3d(mindssc_fix_,grid_sp,stride=grid_sp)
mind_mov_ = F.avg_pool3d(mindssc_mov_,grid_sp,stride=grid_sp)
del mindssc_fix_,mindssc_mov_
#extract one-hot patches
#create optimisable displacement grid
net = nn.Sequential(nn.Conv3d(3,1,(H//grid_sp,W//grid_sp,D//grid_sp),bias=False))
net[0].weight.data[:] = disp_lr/grid_sp
net.cuda()
optimizer = torch.optim.Adam(net.parameters(), lr=1)
grid0 = F.affine_grid(torch.eye(3,4).unsqueeze(0).cuda(),(1,1,H//grid_sp,W//grid_sp,D//grid_sp),align_corners=False)
#run Adam optimisation with diffusion regularisation and B-spline smoothing
lambda_weight = 1.25# sad: 10, ssd:0.75
for iter in range(100):
optimizer.zero_grad()
disp_sample = F.avg_pool3d(F.avg_pool3d(net[0].weight,3,stride=1,padding=1),3,stride=1,padding=1).permute(0,2,3,4,1)
reg_loss = lambda_weight*((disp_sample[0,:,1:,:]-disp_sample[0,:,:-1,:])**2).mean()+\
lambda_weight*((disp_sample[0,1:,:,:]-disp_sample[0,:-1,:,:])**2).mean()+\
lambda_weight*((disp_sample[0,:,:,1:]-disp_sample[0,:,:,:-1])**2).mean()
scale = torch.tensor([(H//grid_sp-1)/2,(W//grid_sp-1)/2,(D//grid_sp-1)/2]).cuda().unsqueeze(0)
grid_disp = grid0.view(-1,3).cuda().float()+((disp_sample.view(-1,3))/scale).flip(1).float()
patch_mov_sampled = F.grid_sample(mind_mov_.float(),grid_disp.view(1,H//grid_sp,W//grid_sp,D//grid_sp,3).cuda(),align_corners=False,mode='bilinear')#,padding_mode='border')
#patch_mov_sampled_sq = F.grid_sample(mind_mov_.pow(2).float(),grid_disp.view(1,H//grid_sp,W//grid_sp,D//grid_sp,3).cuda(),align_corners=True,mode='bilinear')
#sampled_cost = (patch_mov_sampled_sq-2*patch_mov_sampled*mind_fix_+mind_fix_.pow(2)).mean(1)*12
sampled_cost = (patch_mov_sampled-mind_fix_).pow(2).mean(1)*12
loss = sampled_cost.mean()
(loss+reg_loss).backward()
optimizer.step()
torch.cuda.synchronize()
t1 = time.time()
print(t1-t0,'sec (optim)')
fitted_grid = disp_sample.permute(0,4,1,2,3).detach()
disp_hr = F.interpolate(fitted_grid*grid_sp,size=(H,W,D),mode='trilinear',align_corners=False)
disp_field = F.interpolate(disp_hr,scale_factor = 0.5,mode='trilinear',align_corners=False)
x1 = disp_field[0,0,:,:,:].cpu().float().data.numpy()
y1 = disp_field[0,1,:,:,:].cpu().float().data.numpy()
z1 = disp_field[0,2,:,:,:].cpu().float().data.numpy()
#x1 = zoom(x,1/2,order=2).astype('float16')
#y1 = zoom(y,1/2,order=2).astype('float16')
#z1 = zoom(z,1/2,order=2).astype('float16')
np.savez_compressed('/data_supergrover2/heinrich/L2R2021/convexAdam/submission/task_03/disp_'+str(nu).zfill(4)+'_'+str(nu+1).zfill(4)+'.npz',np.stack((x1,y1,z1),0))
torch.cuda.synchronize()
t1 = time.time()
case_time[nu-1] = t1-t0
torch.cuda.synchronize()
t1_ = time.time()
print('total time',t1_-t0_)
torch.save(case_time,'/data_supergrover2/heinrich/L2R2021/convexAdam/task3_times.pth')