forked from omarocegueda/registration
/
registrationCommon.py
442 lines (403 loc) · 15.9 KB
/
registrationCommon.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
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
# -*- coding: utf-8 -*-
"""
Created on Mon Sep 30 15:40:28 2013
@author: khayyam
"""
import numpy as np
import scipy as sp
from scipy import ndimage
import matplotlib.pyplot as plt
import nibabel as nib
import os
import tensorFieldUtils as tf
###############################################################
########################## Common ###########################
###############################################################
const_prefilter_map_coordinates=False
def getBaseFileName(fname):
base=os.path.basename(fname)
noExt=os.path.splitext(base)[0]
while(noExt!=base):
base=noExt
noExt=os.path.splitext(base)[0]
return noExt
def getDistribution(img1, img2):
sh=img1.shape
dist=np.zeros((256,256))
for i in range(sh[0]):
for j in range(sh[1]):
a=int(img1[i,j])
b=int(img2[i,j])
dist[a,b]+=1
return dist
def drawLattice2D(nrows, ncols, delta):
lattice=np.ndarray((1+(delta+1)*nrows, 1+(delta+1)*ncols), dtype=np.float64)
lattice[...]=127
for i in range(nrows+1):
lattice[i*(delta+1), :]=0
for j in range(ncols+1):
lattice[:, j*(delta+1)]=0
return lattice
def createDeformationField2D_type1(nrows, ncols, maxDistp):
deff=np.ndarray((nrows, ncols, 2), dtype=np.float64)
midCol=ncols//2
midRow=nrows//2
for i in range(nrows):
deff[i,:,0]=maxDistp*np.sin(2*np.pi*(np.array(range(ncols), dtype=np.float64)-midCol)/ncols)
for j in range(ncols):
deff[:,j,1]=maxDistp*np.sin(2*np.pi*(np.array(range(nrows), dtype=np.float64)-midRow)/nrows)
v=np.array(range(nrows), dtype=np.float64)-midRow
h=np.array(range(ncols), dtype=np.float64)-midCol
nrm=midRow**2+midCol**2
p=np.exp(-(v[:,None]**2+h[None,:]**2)/(0.1*nrm))
p=(p-p.min())/(p.max()-p.min())
deff[:,:,0]*=p
deff[:,:,1]*=p
return deff
def createDeformationField2D_type2(nrows, ncols, maxDistp):
deff=np.ndarray((nrows, ncols, 2), dtype=np.float64)
midCol=ncols//2
midRow=nrows//2
for i in range(nrows):
deff[i,:,0]=maxDistp*np.sin(2*np.pi*(np.array(range(ncols), dtype=np.float64)-midCol)/ncols)
deff[i,:,1]=maxDistp*np.sin(2*np.pi*(np.array(range(ncols), dtype=np.float64)-midCol)/ncols)
for j in range(ncols):
deff[:,j,0]*=np.sin(2*np.pi*(np.array(range(nrows), dtype=np.float64)-midRow)/nrows)
deff[:,j,1]*=np.sin(2*np.pi*(np.array(range(nrows), dtype=np.float64)-midRow)/nrows)
return deff
def createDeformationField2D_type3(nrows, ncols, maxDistp):
deff=np.ndarray((nrows, ncols, 2), dtype=np.float64)
X0,X1=np.mgrid[0:nrows, 0:ncols]
midCol=ncols//2
midRow=nrows//2
nn=np.sqrt(midCol*midCol+midRow*midRow)
factor=maxDistp/nn
deff[...,0]=(X1-midCol)*(-factor)
deff[...,1]=(X0-midRow)*(factor)
return deff
def drawLattice3D(dims, delta):
dims=np.array(dims)
nsquares=(dims-1)/(delta+1)
lattice=np.zeros(shape=dims, dtype=np.float64)
lattice[...]=127
for i in range(nsquares[0]+1):
lattice[i*(delta+1), :, :]=0
for j in range(nsquares[1]+1):
lattice[:, j*(delta+1), :]=0
for k in range(nsquares[2]+1):
lattice[:, :, k*(delta+1)]=0
return lattice
def createDeformationField3D_type2(dims, maxDistp):
deff=np.ndarray(dims+(3,), dtype=np.float64)
dims=np.array(dims, dtype=np.int32)
mid=dims//2
factor=maxDistp**(1.0/3.0)
for i in range(dims[0]):
deff[i,:,:,0]=factor*np.sin(2*np.pi*(np.array(range(dims[0]), dtype=np.float64)-mid[0])/dims[0])
deff[i,:,:,1]=factor*np.sin(2*np.pi*(np.array(range(dims[0]), dtype=np.float64)-mid[0])/dims[0])
deff[i,:,:,2]=factor*np.sin(2*np.pi*(np.array(range(dims[0]), dtype=np.float64)-mid[0])/dims[0])
for j in range(dims[1]):
deff[:,j,:,0]*=factor*np.sin(2*np.pi*(np.array(range(dims[1]), dtype=np.float64)-mid[1])/dims[1])
deff[:,j,:,1]*=factor*np.sin(2*np.pi*(np.array(range(dims[1]), dtype=np.float64)-mid[1])/dims[1])
deff[:,j,:,2]*=factor*np.sin(2*np.pi*(np.array(range(dims[1]), dtype=np.float64)-mid[1])/dims[1])
for k in range(dims[2]):
deff[:,:,k,0]*=factor*np.sin(2*np.pi*(np.array(range(dims[2]), dtype=np.float64)-mid[2])/dims[2])
deff[:,:,k,1]*=factor*np.sin(2*np.pi*(np.array(range(dims[2]), dtype=np.float64)-mid[2])/dims[2])
deff[:,:,k,2]*=factor*np.sin(2*np.pi*(np.array(range(dims[2]), dtype=np.float64)-mid[2])/dims[2])
return deff
def warpImage(image, displacement):
sh=image.shape
X0,X1=np.mgrid[0:sh[0], 0:sh[1]]
warped=ndimage.map_coordinates(image, [X0+displacement[...,0], X1+displacement[...,1]], prefilter=const_prefilter_map_coordinates)
return warped
def warpVolume(volume, displacement, volAffine=None, dispAffine=None):
sh=volume.shape
X0,X1,X2=np.mgrid[0:sh[0], 0:sh[1], 0:sh[2]]
if(dispAffine!=None):
X0, X1, X2=(dispAffine[0,0]*X0 + dispAffine[0,1]*X1 + dispAffine[0,2]*X2 + dispAffine[0,3]+displacement[...,0],
dispAffine[1,0]*X0 + dispAffine[1,1]*X1 + dispAffine[1,2]*X2 + dispAffine[1,3]+displacement[...,1],
dispAffine[2,0]*X0 + dispAffine[2,1]*X1 + dispAffine[2,2]*X2 + dispAffine[2,3]+displacement[...,2])
else:
X0+=displacement[...,0]
X1+=displacement[...,1]
X2+=displacement[...,2]
if(volAffine!=None):
volInverse=np.linalg.inv(volAffine)
X0, X1, X2=(volInverse[0,0]*X0 + volInverse[0,1]*X1 + volInverse[0,2]*X2 + volInverse[0,3],
volInverse[1,0]*X0 + volInverse[1,1]*X1 + volInverse[1,2]*X2 + volInverse[1,3],
volInverse[2,0]*X0 + volInverse[2,1]*X1 + volInverse[2,2]*X2 + volInverse[2,3])
warped=ndimage.map_coordinates(volume, [X0, X1, X2], prefilter=const_prefilter_map_coordinates)
return warped
def plotDeformationField(d):
plt.figure()
plt.quiver(d[...,1], d[...,0])
def plotDeformedLattice(d, delta=10):
nrows=d.shape[0]
ncols=d.shape[1]
lattice=drawLattice2D((nrows+delta)/(delta+1), (ncols+delta)/(delta+1), delta)
lattice=lattice[0:nrows,0:ncols]
gtLattice=warpImage(lattice, d)
plt.figure()
plt.imshow(gtLattice, cmap=plt.cm.gray)
plt.title('Deformed lattice')
return gtLattice
def plotDeformationFields(dList):
n=len(dList)
plt.figure()
for i in range(n):
plt.subplot(1,n,i+1)
plt.quiver(dList[i][...,1], dList[i][...,0])
def plotOrthogonalField(sh, b):
center=(np.array(sh)-1)/2.0
C,R=sp.meshgrid(np.array(range(sh[1]), dtype=np.float64), np.array(range(sh[0]), dtype=np.float64))
R=R-center[0]+b[0]
C=C-center[1]+b[1]
plt.figure()
plt.quiver(R, -C)
def plotPyramids(L, R):
n=len(L)
plt.figure()
for i in range(n):
plt.subplot(2,n,i+1)
plt.imshow(L[i], cmap = plt.cm.gray)
plt.title('Level: '+str(i))
plt.subplot(2,n,n+i+1)
plt.imshow(R[i], cmap = plt.cm.gray)
plt.title('Level: '+str(i))
def renormalizeImage(image):
m=np.min(image)
M=np.max(image)
if(M-m<1e-8):
return image
return 127.0*(image-m)/(M-m)
def plotOverlaidPyramids(L, R):
n=len(L)
plt.figure()
for i in range(n):
plt.subplot(1,n,i+1)
colorImage=np.zeros(shape=(L[i].shape)+(3,), dtype=np.int8)
ll=renormalizeImage(L[i]).astype(np.int8)
rr=renormalizeImage(R[i]).astype(np.int8)
colorImage[...,0]=ll*(ll>ll[0,0])
colorImage[...,1]=rr*(rr>rr[0,0])
plt.imshow(colorImage)
plt.title('Level: '+str(i))
def plotOverlaidPyramids3DCoronal(L, R):
n=len(L)
plt.figure()
for i in range(n):
sh=L[i].shape
plt.subplot(1,n,i+1)
colorImage=np.zeros(shape=(sh[0], sh[2], 3), dtype=np.int8)
ll=renormalizeImage(L[i][:,sh[1]//2,:]).astype(np.int8)
rr=renormalizeImage(R[i][:,sh[1]//2,:]).astype(np.int8)
colorImage[...,0]=ll*(ll>ll[0,0])
colorImage[...,1]=rr*(rr>rr[0,0])
plt.imshow(colorImage)
plt.title('Level: '+str(i))
def getRotationMatrix(angles):
ca=np.cos(angles[0])
cb=np.cos(angles[1])
cg=np.cos(angles[2])
sa=np.sin(angles[0])
sb=np.sin(angles[1])
sg=np.sin(angles[2])
return np.array([[cb*cg,-ca*sg+sa*sb*cg,sa*sg+ca*sb*cg],[cb*sg,ca*cg+sa*sb*sg,-sa*cg+ca*sb*sg],[-sb,sa*cb,ca*cb]])
def getRotationMatrix2D(angle):
c=np.cos(angle)
s=np.sin(angle)
return np.array([[c, -s],[s, c]])
def applyRigidTransformation3D(image, beta):
sh=image.shape
center=(np.array(sh)-1)/2.0
X0,X1,X2=np.mgrid[0:sh[0], 0:sh[1], 0:sh[2]]
X0=X0-center[0]
X1=X1-center[1]
X2=X2-center[2]
R=getRotationMatrix(beta[0:3])
X0new,X1new,X2new=(R[0,0]*X0 + R[0,1]*X1 + R[0,2]*X2 + center[0] + beta[3],
R[1,0]*X0 + R[1,1]*X1 + R[1,2]*X2 + center[1] + beta[4],
R[2,0]*X0 + R[2,1]*X1 + R[2,2]*X2 + center[2] + beta[5])
return ndimage.map_coordinates(image, [X0new, X1new, X2new], prefilter=const_prefilter_map_coordinates)
def pyramid_gaussian_3D(image, max_layer, mask=None):
yield image.copy().astype(np.float64)
for i in range(max_layer):
newImage=sp.ndimage.filters.gaussian_filter(image, 2.0/3.0)[::2,::2,::2].copy()
if(mask!=None):
mask=mask[::2,::2,::2]
newImage*=mask
image=newImage.copy()
yield newImage
#def pyramid_gaussian_3D(image, max_layer, mask=None):
# yield image.copy().astype(np.float64)
# for i in range(max_layer):
# newImage=np.array(tf.downsample_scalar_field3D(image))
# image=newImage
# yield newImage
def pyramid_gaussian_2D(image, max_layer, mask=None):
yield image.copy().astype(np.float64)
for i in range(max_layer):
newImage=np.empty(shape=((image.shape[0]+1)//2, (image.shape[1]+1)//2), dtype=np.float64)
newImage[...]=sp.ndimage.filters.gaussian_filter(image, 2.0/3.0)[::2,::2]
if(mask!=None):
mask=mask[::2,::2]
newImage*=mask
image=newImage
yield newImage
#def pyramid_gaussian_2D(image, max_layer, mask=None):
# yield image.copy().astype(np.float64)
# for i in range(max_layer):
# newImage=np.array(tf.downsample_scalar_field(image))
# image=newImage
# yield newImage
def overlayImages(img0, img1, createFig=True):
colorImage=np.zeros(shape=(img0.shape)+(3,), dtype=np.int8)
colorImage[...,0]=renormalizeImage(img0)
colorImage[...,1]=renormalizeImage(img1)
fig=None
if(createFig):
fig=plt.figure()
plt.subplot(1,3,1)
plt.imshow(img0, cmap=plt.cm.gray)
plt.title('Img0 (red)')
plt.subplot(1,3,2)
plt.imshow(colorImage)
plt.title('Overlay')
plt.subplot(1,3,3)
plt.imshow(img1, cmap=plt.cm.gray)
plt.title('Img1 (green)')
return fig
def testOverlayImages_raw():
leftName='data/t2/t2_icbm_normal_1mm_pn0_rf0.rawb'
rightName='data/t1/t1_icbm_normal_1mm_pn0_rf0.rawb'
ns=181
nr=217
nc=181
left=np.fromfile(leftName, dtype=np.ubyte).reshape(ns,nr,nc)
left=left.astype(np.float64)
right=np.fromfile(rightName, dtype=np.ubyte).reshape(ns,nr,nc)
right=right.astype(np.float64)
overlayImages(left[90], right[90])
def getColor(label):
r=label%2
g=(label/2)%2
b=(label/4)%2
return np.array([r,g,b])
def testOverlayImages_nii():
leftName='data/t1/IBSR18/IBSR_01/IBSR_01_ana_strip.nii.gz'
rightName='data/t1/IBSR18/IBSR_02/IBSR_02_ana_strip.nii.gz'
nib_left = nib.load(leftName)
nib_right = nib.load(rightName)
left=nib_left.get_data().astype(np.double)
right=nib_right.get_data().astype(np.double)
left=np.copy(left, order='C')
right=np.copy(right, order='C')
sl=np.array(left.shape)//2
sr=np.array(right.shape)//2
overlayImages(left[sl[0],:,:,0], right[sr[0],:,:,0])
overlayImages(left[:,sl[1],:,0], right[:,sr[1],:,0])
overlayImages(left[:,:,sl[2],0], right[:,:,sr[2],0])
def plotDiffeomorphism(GT, GTinv, GTres, titlePrefix, delta=10):
nrows=GT.shape[0]
ncols=GT.shape[1]
X1,X0=np.mgrid[0:GT.shape[0], 0:GT.shape[1]]
lattice=drawLattice2D((nrows+delta)/(delta+1), (ncols+delta)/(delta+1), delta)
lattice=lattice[0:nrows,0:ncols]
gtLattice=warpImage(lattice, np.array(GT))
gtInvLattice=warpImage(lattice, np.array(GTinv))
gtResidual=warpImage(lattice, np.array(GTres))
plt.figure()
plt.subplot(2, 3, 1)
plt.imshow(gtLattice, cmap=plt.cm.gray)
plt.title(titlePrefix+'[Deformation]')
plt.subplot(2, 3, 2)
plt.imshow(gtInvLattice, cmap=plt.cm.gray)
plt.title(titlePrefix+'[Inverse]')
plt.subplot(2, 3, 3)
plt.imshow(gtResidual, cmap=plt.cm.gray)
plt.title(titlePrefix+'[residual]')
#plot jacobians and residual norm
detJacobian=computeJacobianField(GT)
plt.subplot(2, 3, 4)
plt.imshow(detJacobian, cmap=plt.cm.gray)
CS=plt.contour(X0,X1,detJacobian,levels=[0.0], colors='r')
plt.clabel(CS, inline=1, fontsize=10)
plt.title('det(J(d))')
detJacobianInverse=computeJacobianField(GTinv)
plt.subplot(2, 3, 5)
plt.imshow(detJacobianInverse, cmap=plt.cm.gray)
CS=plt.contour(X0,X1,detJacobianInverse,levels=[0.0], colors='r')
plt.clabel(CS, inline=1, fontsize=10)
plt.title('det(J(d^-1))')
nrm=np.sqrt(np.sum(np.array(GTres)**2,2))
plt.subplot(2, 3, 6)
plt.imshow(nrm, cmap=plt.cm.gray)
plt.title('||residual||_2')
g00, g01=sp.gradient(GTinv[...,0])
g10, g11=sp.gradient(GTinv[...,1])
#priorEnergy=g00**2+g01**2+g10**2+g11**2
return [gtLattice, gtInvLattice, gtResidual, detJacobian]
def computeJacobianField(displacement):
g00,g01=sp.gradient(displacement[...,0])
g10,g11=sp.gradient(displacement[...,1])
return (1+g00)*(1+g11)-g10*g01
def saveDeformedLattice3D(dname, oname='deformed_lattice.nii.gz'):
'''
saveDeformedLattice3D('displacement_templateT1ToIBSR01T1_diff.npy')
saveDeformedLattice3D('displacement_templateT1ToIBSR01T1_diffMulti.npy')
'''
print 'Loading displacement...'
displacement=np.load(dname)
minVal, maxVal=tf.get_displacement_range(displacement, None)
sh=np.array([np.ceil(maxVal[0]),np.ceil(maxVal[1]),np.ceil(maxVal[2])], dtype=np.int32)
print sh.dtype
print sh
L=np.array(drawLattice3D(sh, 10))
print 'Warping lattice...'
warped=np.array(tf.warp_volume(L, displacement, np.eye(4))).astype(np.int16)
print 'Transforming to Nifti...'
img=nib.Nifti1Image(warped, np.eye(4))
print 'Saving warped lattice as:',oname
img.to_filename(oname)
print 'done.'
def readAntsAffine(fname):
'''
readAntsAffine('IBSR_01_ana_strip_t1_icbm_normal_1mm_pn0_rf0_peeledAffine.txt')
'''
try:
with open(fname) as f:
lines=[line.strip() for line in f.readlines()]
except IOError:
print 'Can not open file: ', fname
return
if not (lines[0]=="#Insight Transform File V1.0"):
print 'Unknown file format'
return
if lines[1]!="#Transform 0":
print 'Unknown transformation type'
return
A=np.zeros((3,3))
b=np.zeros((3,))
c=np.zeros((3,))
for line in lines[2:]:
data=line.split()
if data[0]=='Transform:':
if data[1]!='MatrixOffsetTransformBase_double_3_3':
print 'Unknown transformation type'
return
elif data[0]=='Parameters:':
parameters=np.array([float(s) for s in data[1:]], dtype=np.float64)
A=parameters[:9].reshape((3,3))
b=parameters[9:]
elif data[0]=='FixedParameters:':
c=np.array([float(s) for s in data[1:]], dtype=np.float64)
T=np.ndarray(shape=(4,4), dtype=np.float64)
T[:3,:3]=A[...]
T[3,:]=0
T[3,3]=1
T[:3,3]=b+c-A.dot(c)
############This conversion is necessary for compatibility between itk and nibabel#########
conversion=np.array([[-1,0,0,0],[0,-1,0,0],[0,0,1,0],[0,0,0,1]], dtype=np.float64)
T=conversion.dot(T.dot(conversion))
###########################################################################################
return T