-
Notifications
You must be signed in to change notification settings - Fork 0
/
bulk_process_GM_flywheel.py
executable file
·324 lines (288 loc) · 12.1 KB
/
bulk_process_GM_flywheel.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
import numpy as np
import nibabel as nb
from glob import glob
import commonly as c
import math
from collections import defaultdict
#import matplotlib.pyplot as plt
import sys
from scipy.stats import norm
import os
from subprocess import Popen
from subprocess import PIPE
import scipy.stats as stats
from scipy.optimize import curve_fit
from avgim import avgim
##just to apply warps to cord and mask iamges##
import General_reg_par as GRP
import logging as log
from henrygce.logging import log_gm_job_status
def sigmoid(x,x0,k,y0):
y = 1 / (1 + np.exp(-k*(x-x0))) + y0
return y
def hist_j(image):
d=defaultdict(int)
for i in image.flatten():
d[i]=d[i]+1
return d
def quantile_transform(image):
dat_array=image
nonzer_dat=dat_array[np.where(dat_array>0)]
unique_vals=sorted(set(nonzer_dat))
step=1/len(unique_vals)
dist=norm(0,1)
normal_data=[dist.pdf(step*(-1*int(len(unique_vals)/2)+x)) for x in range(len(unique_vals))]
new_nums=[]
#print(len(normal_data))
#print(len(unique_vals))
d={}
for i in range(len(unique_vals)):
d[unique_vals[i]]=normal_data[i]
sh=dat_array.shape
for i in range(sh[0]):
for j in range(sh[1]):
if dat_array[i,j]==0:
continue
dat_array[i,j]=d[dat_array[i,j]]
return dat_array
def quantile_transform(image):
dat_array=image
new_dats=np.zeros(dat_array.shape)
nonzer_dat=dat_array[np.where(dat_array!=0)]
unique_vals=sorted(set(nonzer_dat))
#print(len(unique_vals))
#input()
new_vals=[]
dist=stats.norm(0,1)
d={}
for i in unique_vals:
perc=stats.percentileofscore(nonzer_dat, i, kind='strict')
if perc==0:
d[i]=-3
continue
d[i]=dist.ppf(perc*.01)
sh=dat_array.shape
for i in range(sh[0]):
for j in range(sh[1]):
if dat_array[i,j]==0:
continue
new_dats[i,j]=d[dat_array[i,j]]
return new_dats
def z_score(image):
nonzer=image[np.where(image!=0)]
mean=np.mean(nonzer)
std=np.std(nonzer)
sh=image.shape
r_image=np.zeros(sh)
for i in range(sh[0]):
for j in range(sh[1]):
if image[i,j]==0:
continue
r_image[i,j]=(image[i,j]-mean)/std
return r_image
def func_l(x,a,b):
return (a*x)+b
def rounds(x):
if x%1>=.75:
return int(x)+1
else:
return int(x)
def scanner(x):
if 'SKYRA' in x:
return '_SKYRA'
if 'GE' in x:
return '_GE'
if 'PHILIPS' in x:
return '_PHILIPS'
else:
return ''
def create_prob_seg_iteration3(template_grays,templates,image,file_handl):
a=nb.load(image)
adat_raw=a.get_data()
adat=quantile_transform(a.get_data())
#print(adat.dtype)
#input()
distributions_raw=[]
distributions=[]
fgs=[]
data_dict={}
for i in template_grays:
#print(c.get_ms(os.path.basename(i)),i)
data_dict[c.get_ms(os.path.basename(i))]=i
for i in templates:
temp=nb.load(i).get_data()
temp=quantile_transform(temp)
#print('here')
try:
z=z_score(temp)
except:
print(sys.exc_info())
file_handl.write(str(sys.exc_info())+'\n')
continue
try:
#print(data_dict[c.get_ms(os.path.basename(i))],i)
fg=nb.load(data_dict[c.get_ms(os.path.basename(i))]).get_data()
except:
print(sys.exc_info())
file_handl.write(str(sys.exc_info())+'\n')
continue
distributions.append(z)
fgs.append(fg)
return fgs,distributions,a,adat,adat_raw
def run_this(static,outputs_path,pth,pth2,cycle_size,subj, sess, protocol,prefix=0):
file_handl=open(outputs_path+'prints.txt','w')
errors=[]
pass_on=[]
if not(prefix):
if 'retest' in static:
mse_static=c.get_mse(static)+'PSIR_retest'+scanner(static)
else:
mse_static=c.get_mse(static)+'PSIR_'+scanner(static)
else:
mse_static=prefix
try:
os.remove(os.path.join(outputs_path,'registrations1/warped/synslice_avggmsegs.nii.gz'))
except:
pass
output_path=os.path.join(outputs_path, 'registrations1/')
dim=2
static_path=static
log.info("about to start first round of registrations!")
GRP.SimpleRegister(pth,output_path,static_path,file_handl,cycle_size,subj, sess, protocol, pth2=pth2).Syn(gilroy=True)
file_handl.write(str(static)+'\n')
if True:
log_gm_job_status("quality assurance check", subj, sess, protocol)
#gray matter transformed images#
template_grays=glob(os.path.join(output_path, 'warped/*ms*'))
#cord transforms#
templates=glob(os.path.join(output_path,'warped1/*ms*'))
fgs,distributions,a,adat,adat_raw=create_prob_seg_iteration3(template_grays,templates,static,file_handl)
file_handl.close()
avgim(os.path.join(output_path, 'warped/'))
aff=nb.load(static)
adat_raw=nb.load(static).get_data()
adat_z=z_score(adat)
#print(adat.dtype)
#mask=nb.load(glob('/data/henry4/jjuwono/data/henry4/jjuwono/'+prefix+'/'+mse_static+'/warped/syn*')[0]).get_data()
#mask=np.where(mask>.6,mask,0)
#bar=np.mean(adat_z[np.where(mask>0)])
sh=adat.shape
slope=np.zeros(sh)
intercept=np.zeros(sh)
confidences=np.zeros(sh)
confidences1=np.zeros(sh)
confidences2=np.zeros(sh)
t_map=np.zeros(sh)
mean_templates=np.zeros(sh)
new_image=np.zeros(sh)
new_image_logi=np.zeros(sh)
original_line_fit=np.zeros(sh)
color_im=np.zeros(sh)
print('length fg:{}'.format(len(fgs)))
print('length dist:{}'.format(len(distributions)))
distributions=np.asarray(distributions)
fgs=np.asarray(fgs)
for i in range(sh[0]):
for j in range(sh[1]):
if adat_raw[i,j]==0:
continue
##insert A block here for polynomial degree fitting ###
##average method##
a = np.array(distributions[:,i,j])[np.newaxis]
if np.sum(distributions[:,i,j])==0:
params=[0,np.mean(fgs[:i,j])]
else:
params=np.polyfit(distributions[:,i,j],fgs[:,i,j],1)
original_line_fit[i,j]=(adat_z[i,j]*params[0])+params[1]
#if len(np.where(fgs[:,i,j]==0)[0])<40 or len(np.where(fgs[:,i,j]==1)[0])<40:
# assign=(adat_z[i,j]*params[0])+params[1]
#else:
#print('logi')
# group0=np.mean(distributions[:,i,j][np.where(fgs[:,i,j]==0)])
# group1=np.mean(distributions[:,i,j][np.where(fgs[:,i,j]==1)])
# grouped=np.where(fgs[:,i,j]!=0,distributions[:,i,j],group0)
# grouped=np.where(fgs[:,i,j]!=1,grouped,group1)
# params=np.polyfit(grouped,fgs[:,i,j],1)
# logi_slope=params[0]
# logi_inter=params[1]
# assign=(adat_z[i,j]*params[0])+params[1]
# if 0<=assign<=1:
# assign=assign
# elif assign<0:
# assign=0
# else:
# assign=1
#if params[0]!=0:
#plt.plot(grouped,fgs[:,i,j],'o',label=str((i,j)))
#plt.plot([group0,group1],[group0*params[0]-params[1],group1*params[0]-params[1]],'r')
#plt.legend()
#plt.show()
#input([group0,group1])
#new_image_logi[i,j]=assign
##reverse method##
#if len(np.where(fgs[:,i,j]==0)[0])<25 or len(np.where(fgs[:,i,j]==1)[0])<25:
# params=np.polyfit(distributions[:,i,j],fgs[:,i,j],1)
# new_image[i,j]=(adat_z[i,j]*params[0])+params[1]
# if len(np.where(fgs[:,i,j]==0)[0])<60:
# color_im[i,j]=1
#else:
#plt.plot(distributions[:,i,j],fgs[:,i,j],'o',label=str((i,j)))
#plt.legend()
#plt.show()
#input()
# params,covs=np.polyfit(fgs[:,i,j],distributions[:,i,j],1,cov=True)
#input(covs)
# slope[i,j]=params[0]
# intercept[i,j]=params[1]
# if abs(params[0])<2*(covs[0,0]**0.5):
# new_image[i,j]=stats.mode(fgs[:,i,j],axis=None)[0][0]
# color_im[i,j]=2
# else:
# new_image[i,j]=(adat_z[i,j]-params[1])/params[0]
# if np.isnan(covs[0,1]):
#input('whyyyyyyyyyyyyuyuyyyyyyyy')
# covs[0,1]=0
# confidence=(abs((covs[1,1]/((adat_z[i,j]-params[1])**2))+(covs[0,0]/((params[0])**2))-(2*((abs(covs[0,1]))**0.5)/((adat_z[i,j]-params[1])*params[0]))))**0.5
# confidences[i,j]=confidence/new_image[i,j]
# if confidence>=2:
# new_image[i,j]=-1000
# color_im[i,j]=3
# if i>136:
#plt.plot(fgs[:,i,j],distributions[:,i,j],'o',label=str((i,j)))
#plt.plot(sorted(fgs[:,i,j]),[x*params[0]+params[1] for x in sorted(fgs[:,i,j])],'r')
#plt.legend()
#plt.show()
#plt.close()
#input((params[0],params[1]))
#plt.plot(distributions[:,i,j],fgs[:,i,j],'o',label=str((i,j)))
#plt.plot([adat_z[i,j]],[(adat_z[i,j]-params[1])/params[0]],'+',
# label='{:.3g},{:.3g}'.format(adat_z[i,j],(adat_z[i,j]-params[1])/params[0]))
#plt.plot(sorted(distributions[:,i,j]),[(x-params[1])/params[0] for x in sorted(distributions[:,i,j])],'r')
#plt.ylim([-0.01,1.01])
#plt.legend()
#plt.show()
#plt.close()
#input((params[1],params[0]+params[1]))
# pass
# print(params[0],params[1])
mean_template=np.mean(distributions[:,i,j])
std_template=np.std(distributions[:,i,j])
mean_templates[i,j]=mean_template
t_map[i,j]=(adat_z[i,j]-mean_template)/std_template
#input()
#nb.save(nb.Nifti1Image(mean_templates,aff.affine),'/data/henry4/jjuwono/mean_templates.nii.gz')
#nb.save(nb.Nifti1Image(slope,a.affine),'/data/henry4/jjuwono/slopes.nii.gz')
#nb.save(nb.Nifti1Image(intercept,a.affine),'/data/henry4/jjuwono/intercepts.nii.gz')
try:
os.mkdir(os.path.join(outputs_path, 'quality_assurance'))
except:
pass
nb.save(nb.Nifti1Image(confidences,aff.affine), os.path.join(outputs_path, 'quality_assurance/confidence.nii.gz'))
nb.save(nb.Nifti1Image(new_image,aff.affine), os.path.join(outputs_path, 'quality_assurance/new_image.nii.gz'))
nb.save(nb.Nifti1Image(new_image_logi,aff.affine), os.path.join(outputs_path, 'quality_assurance/new_image_logi.nii.gz'))
nb.save(nb.Nifti1Image(t_map,aff.affine), os.path.join(outputs_path, 'quality_assurance/t_map.nii.gz'))
nb.save(nb.Nifti1Image(color_im,aff.affine), os.path.join(outputs_path, 'quality_assurance/color_im.nii.gz'))
nb.save(nb.Nifti1Image(original_line_fit,aff.affine), os.path.join(outputs_path, 'quality_assurance/original_line_fit.nii.gz'))
return 1
else:
return 0