-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathnhdr.py
183 lines (159 loc) · 8.32 KB
/
nhdr.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
from json.tool import main
import os
import SimpleITK as sitk
import numpy as np
import matplotlib.pyplot as plt
plt.rcParams['image.cmap'] = 'gray'
plt.rcParams['image.interpolation'] = 'nearest'
import nibabel as nib
from dipy.viz import regtools
from dipy.align.imaffine import (AffineMap,
MutualInformationMetric,
AffineRegistration)
from dipy.align.transforms import (TranslationTransform3D,
RigidTransform3D,
AffineTransform3D)
# $ git clone https://github.com/SuperElastix/SimpleElastix
# $ mkdir build
# $ cd build
# $ cmake ../SimpleElastix/SuperBuild
# $ make -j4
# set default parameter
new_spacing = [1, 1, 1]
registrated_type = 'affine'
# See the Dipy registration tutorial for the details of what these parameters mean:
nbins = 32
sampling_prop = None
level_iters = [10, 10, 5]
sigmas = [3.0, 1.0, 0.0]
factors = [4, 2, 1]
dirname = r'F:\BaiduNetdiskDownload\shuju\shuju\226630'
filename = r'F:\BaiduNetdiskDownload\shuju\shuju\226630\ep2d_diff_tra_b800_b1000_p2_160 - as a 3 frames MultiVolume by Siemens.B-value.raw.gz'
filename2 = r'F:\BaiduNetdiskDownload\shuju\ep2d_diff_tra_b800_b1000_p2_160 - as a 3 frames MultiVolume by Siemens.B-value.raw.gz'
def registrate_valid():
moving = nib.load(r'F:\BaiduNetdiskDownload\shuju\shuju\226630\8 ep2d_diff_tra_b800_b1000_p2_160_ADC_1_resampled.nii')
fixed = nib.load(r'F:\BaiduNetdiskDownload\shuju\shuju\226630\4 t2_tse_tra_p2_4_resampled.nii')
moving_data = moving.get_data()
fixed_data = fixed.get_data()
moving_affine = moving.affine
fixed_affine = fixed.affine
identity = np.eye(4)
affine_map = AffineMap(identity, fixed_data.shape, fixed_affine, moving_data.shape, moving_affine)
resampled = affine_map.transform(moving_data) # 3-4 minutes
# regtools.overlay_slices(fixed_data, resampled, None, 0, 'fixed', 'moving')
# regtools.overlay_slices(fixed_data, resampled, None, 1, 'fixed', 'moving')
# regtools.overlay_slices(fixed_data, resampled, None, 2, 'fixed', 'moving')
metric = MutualInformationMetric(nbins, sampling_prop)
affreg = AffineRegistration(metric=metric, level_iters=level_iters, sigmas=sigmas, factors=factors)
# optimize the translations
transform = TranslationTransform3D()
params0 = None
translation = affreg.optimize(fixed_data, moving_data, transform, params0, fixed_affine, moving_affine)
# translation.affine
transformed = translation.transform(moving_data)
# regtools.overlay_slices(fixed_data, transformed, None, 0, 'fixed', 'transformed')
# regtools.overlay_slices(fixed_data, transformed, None, 1, 'fixed', 'transformed')
# regtools.overlay_slices(fixed_data, transformed, None, 2, 'fixed', 'transformed')
# optimize a rigid-body transform
transform = RigidTransform3D()
rigid = affreg.optimize(fixed_data, moving_data, transform, params0, fixed_affine, moving_affine, starting_affine=translation.affine)
transformed = rigid.transform(moving_data)
# full affine registration
transform = AffineTransform3D()
affreg.level_iters = [1000, 1000, 100]
affine = affreg.optimize(fixed_data, moving_data, transform, params0, fixed_affine, moving_affine, starting_affine=rigid.affine)
transformed = affine.transform(moving_data)
regtools.overlay_slices(fixed_data, transformed, None, 0, 'fixed', 'transformed')
regtools.overlay_slices(fixed_data, transformed, None, 1, 'fixed', 'transformed')
regtools.overlay_slices(fixed_data, transformed, None, 2, 'fixed', 'transformed')
transformed_image = sitk.GetImageFromArray(transformed.transpose(2,1,0))
resampled_fix = sitk.ReadImage(r'F:\BaiduNetdiskDownload\shuju\shuju\226630\4 t2_tse_tra_p2_4_resampled.nii')
transformed_image.SetOrigin(resampled_fix.GetOrigin())
transformed_image.SetSpacing(resampled_fix.GetSpacing())
transformed_image.SetDirection(resampled_fix.GetDirection())
sitk.WriteImage(transformed_image, r'F:\BaiduNetdiskDownload\shuju\shuju\226630\8 ep2d_diff_tra_b800_b1000_p2_160_ADC_1_registrated.nii')
# write a function for z-score
def z_score_process_image(sitk_image):
sitk_arr = sitk.GetArrayFromImage(sitk_image) # ndarray
volume = sitk_arr != 0 # type bool
sitk_arr[volume] = (sitk_arr[volume] - np.mean(sitk_arr[volume])) / np.std(sitk_arr[volume])
return sitk.GetImageFromArray(sitk_arr)
def resample_process_image(sitk_image, new_spacing=[1,1,1]):
ori_size = np.array(sitk_image.GetSize(), dtype=np.int)
ori_origin = sitk_image.GetOrigin()
ori_spacing = np.array(sitk_image.GetSpacing(), dtype=np.float)
ori_direction = sitk_image.GetDirection()
new_spacing = np.array(new_spacing, dtype=np.float)
new_size = ori_size * (ori_spacing / new_spacing)
new_size = np.ceil(new_size).astype(np.int) # Image dimensions are in integers
new_size = [int(s) for s in new_size] # SimpleITK expects lists, not ndarrays
resampler_filter = sitk.ResampleImageFilter()
resampler_filter.SetReferenceImage(sitk_image)
resampler_filter.SetSize(new_size)
resampler_filter.SetInterpolator(sitk.sitkBSpline)
resampler_filter.SetOutputSpacing(new_spacing)
resampler_filter.SetOutputOrigin(ori_origin)
resampler_filter.SetOutputDirection(ori_direction)
resampled_sitk_image = resampler_filter.Execute(sitk_image)
return resampled_sitk_image
def registration_process_image(fixed_image, moving_image, type='affine'):
elastixImageFilter = sitk.ElastixImageFilter()
elastixImageFilter.SetFixedImage(fixed_image)
elastixImageFilter.SetMovingImage(moving_image)
elastixImageFilter.SetParameterMap(sitk.GetDefaultParameterMap("affine"))
elastixImageFilter.Execute()
registed_image = elastixImageFilter.GetResultImage()
return registed_image
def z_score_resample():
under_preprocess = {}
for file in os.listdir(dirname):
if 'zscore' in file:
continue
if 'resampled' in file:
continue
if 't2' in file:
under_preprocess['T2WI'] = os.path.join(dirname, file)
if 'ADC' in file:
under_preprocess['ADC'] = os.path.join(dirname, file)
if 'DWI' in file:
under_preprocess['DWI'] = os.path.join(dirname, file)
# Step1: z-score each image
for key, value in under_preprocess.items():
z_score_name = value.split('.nii')[0] + '_zscore.nii'
if not os.path.exists(z_score_name):
image = sitk.ReadImage(value)
image = sitk.Cast(image, sitk.sitkFloat64)
original_origin = image.GetOrigin()
original_spacing = image.GetSpacing()
original_direction = image.GetDirection()
z_score_image = z_score_process_image(image)
z_score_image.SetOrigin(original_origin)
z_score_image.SetSpacing(original_spacing)
z_score_image.SetDirection(original_direction)
sitk.WriteImage(z_score_image, z_score_name)
# Step2: Resample each image
for key, value in under_preprocess.items():
z_score_name = value.split('.nii')[0] + '_zscore.nii'
resampled_name = value.split('.nii')[0] + '_resampled.nii'
if os.path.exists(z_score_name) and not os.path.exists(resampled_name):
z_score_image = sitk.ReadImage(z_score_name)
resampled_image = resample_process_image(z_score_image)
sitk.WriteImage(resampled_image, resampled_name)
# Step3: Registration (Affine)
fixed_name = under_preprocess['T2WI'].split('.nii')[0] + '_resampled.nii'
fixed_image = sitk.ReadImage(fixed_name)
# fixed = under_preprocess.pop(under_preprocess['T2WI'])
for key, value in under_preprocess.items():
resampled_name = value.split('.nii')[0] + '_resampled.nii'
regis_name = value.split('.nii')[0] + '_registed.nii'
if os.path.exists(resampled_name) and not os.path.exists(regis_name):
if key == 'T2WI':
continue
moving_image = sitk.ReadImage(resampled_name)
registed_image = registration_process_image(fixed_image, moving_image)
sitk.WriteImage(registed_image, regis_name)
if __name__ == '__main__':
# z_score_resample()
registrate_valid()
# readdata, header = nrrd.read(filename2)
# b = 1