def check_finished(ind, subj, sess, protocol, cycle):
log.info('cycle:{}'.format(cycle))
printer = []
sleep(90)
out = check_output(["ps", "-aux"])
# log.info(out)
fin = 0
for i in ind:
printer.append(i.poll())
if i.poll() == None:
fin = fin + 1
log.info(printer)
log_gm_job_status(
"in progress; on second set of registrations: {0}/{1} finished".format(
len(ind) - fin, len(ind)), subj, sess, protocol)
log.info('{}/{} processes finished'.format(len(ind) - fin, len(ind)))
if fin > 0:
return check_finished(ind, subj, sess, protocol, cycle + 1)
return True
def check_finished(self, ind, cycle):
# import pdb
# pdb.set_trace()
log.info('cycle:{}'.format(cycle))
printer = []
sleep(90)
out = check_output(["ps", "-aux"])
log.info(out)
fin = 0
# import pdb
# pdb.set_trace()
count = 0
for i in ind:
out = []
if i.stdout:
for n in np.arange(20):
if n % 2 == 0:
sleep(1)
out.append(i.stdout.readline())
# pdb.set_trace()
f = open('/flywheel/v0/output/{}_error.txt'.format(count), 'w')
# print(out)
for line in out:
# print(line)
f.write(line.decode('utf-8'))
f.close()
printer.append(i.poll())
if i.poll() == None:
fin = fin + 1
count += 1
log.info(printer)
log_gm_job_status(
"in progress; first cycle {}/{} finished".format(
len(ind) - fin, len(ind)), self.subj, self.sess, self.protocol)
log.info('{}/{} processes finished'.format(len(ind) - fin, len(ind)))
# pdb.set_trace()
if fin > 0:
return self.check_finished(ind, cycle + 1)
return True
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
def run_this(static, outputs_path, subj, sess, protocol, prefix=0):
file_handl = open(os.path.join(outputs_path, 'papers.txt'), 'a')
apply_warps = False
if not (prefix):
if 'retest' in static:
subject = c.get_mse(static) + 'retest' + scanner(static)
else:
subject = c.get_mse(static) + scanner(static)
else:
subject = prefix
log.info('#########{}######{}'.format(subject, subject))
mse = subject
try:
os.remove(
os.path.join(outputs_path,
'registrations2/warped/synslice_avggmsegs.nii.gz'))
except:
pass
output_path = os.path.join(outputs_path, 'registrations1/')
log_gm_job_status("final set of warps", subj, sess, protocol)
if apply_warps == True:
dim = 2
static_path = static
files = sorted(glob('/flywheel/v0/input/pth/*'))
files2 = sorted(glob('/flywheel/v0/input/pth2/*'))
if not os.path.exists(os.path.join(output_path, 'warped1')):
os.makedirs(os.path.join(self.output_path, 'warped1'))
for i in range(len(files)):
if 'syn' in files[i]:
continue
fls = os.path.basename(files[i])
fls2 = os.path.basename(files2[i])
log.info((fls, fls2))
cmd1 = [
'/opt/ants-2.3.1/WarpImageMultiTransform',
str(dim), files2[i],
os.path.join(os.path.join(output_path, 'warped/'),
fls2.split('.')[0] + '.nii.gz'),
os.path.join(output_path,
'warp' + fls.split('.')[0] + '1Warp.nii.gz'),
os.path.join(output_path, 'warp' + fls.split('.')[0] +
'0GenericAffine.mat'), '-R', static_path, '\n'
]
cmd2 = [
'/opt/ants-2.3.1/WarpImageMultiTransform',
str(dim), files[i],
os.path.join(os.path.join(output_path, 'warped1/'),
fls.split('.')[0] + '.nii.gz'),
os.path.join(output_path,
'warp' + fls.split('.')[0] + '1Warp.nii.gz'),
os.path.join(output_path, 'warp' + fls.split('.')[0] +
'0GenericAffine.mat'), '-R', static_path, '\n'
]
proc = Popen(cmd1, stdout=PIPE)
proc.wait()
proc = Popen(cmd2, stdout=PIPE)
proc.wait()
###run process to grab distributions##
template_grays = glob(
os.path.join(outputs_path, 'registrations2/warped/ms*.nii.gz'))
templates = glob(
os.path.join(outputs_path, 'registrations1/warped1/ms*.nii.gz'))
fgs, distributions, a, adat, adat_raw = create_prob_seg_iteration3(
template_grays, templates, static, file_handl)
avgim(os.path.join(outputs_path, 'registrations2/warped/'))
###run process to fit lines###
aff = nb.load(static)
adat_raw = nb.load(static).get_data()
first_tmap = nb.load(
glob(os.path.join(outputs_path,
'quality_assurance/t_map.nii.gz'))[0]).get_data()
adat_z = z_score(adat, mapt=first_tmap)
mask = nb.load(
glob(os.path.join(outputs_path,
'registrations2/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)
screwed = 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)
file_handl.write(str(len(fgs)) + '\n')
file_handl.write(str(len(distributions)) + '\n')
distributions = np.asarray(distributions)
fgs = np.asarray(fgs)
file_handl.close()
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]
try:
params = np.polyfit(distributions[:, i, j], fgs[:, i, j], 1)
except:
screwed[i, j] = 1
#plt.plot(distributions[:,i,j],fgs[:,i,j],'o',label=str((i,j)))
#plt.legend()
#plt.show()
#input()
continue
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] = statss.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, 'final_output'))
except:
pass
#nb.save(nb.Nifti1Image(confidences,aff.affine),'/data/henry4/jjuwono/new_GM_method/'+mse+'/confidence.nii.gz')
#nb.save(nb.Nifti1Image(new_image,aff.affine),'/data/henry4/jjuwono/new_GM_method/'+mse+'/new_image.nii.gz')
#nb.save(nb.Nifti1Image(new_image_logi,aff.affine),'/data/henry4/jjuwono/new_GM_method/'+mse+'/new_image_logi.nii.gz')
nb.save(nb.Nifti1Image(t_map, aff.affine),
os.path.join(outputs_path, 'final_output/rereg_t_map.nii.gz'))
#nb.save(nb.Nifti1Image(color_im,aff.affine),'/data/henry4/jjuwono/new_GM_method/'+mse+'/color_im.nii.gz')
nb.save(
nb.Nifti1Image(original_line_fit, aff.affine),
os.path.join(outputs_path,
'final_output/rereg_original_line_fit.nii.gz'))
return 1
def run_this(static, outputs_path, subj, sess, protocol, prefix=0):
file_handl = open(os.path.join(outputs_path, 'papers.txt'), 'a')
apply_warps = False
if not (prefix):
if 'retest' in static:
subject = c.get_mse(static) + 'retest' + scanner(static)
else:
subject = c.get_mse(static) + scanner(static)
else:
subject = prefix
mse = subject
try:
os.remove(
os.path.join(outputs_path,
'registrations2/warped/synslice_avggmsegs.nii.gz'))
except:
pass
log_gm_job_status("final set of warps", subj, sess, protocol)
###run process to fit lines###
aff = nb.load(static)
adat_raw = nb.load(static).get_data()
print(outputs_path)
###run process to grab distributions##
#quick control helper
controls = glob(
os.path.join(outputs_path, 'registrations2/warped/[0-9]*.nii.gz'))
controls1 = glob(
os.path.join(outputs_path, 'registrations1/warped1/[0-9]*.nii.gz'))
control_flag = False
control1_flag = False
for i in controls:
if control_flag == False:
control_par = os.path.dirname(i)
control_flag = True
il = control_par + '/ms' + os.path.basename(i)
os.rename(i, il)
for i in controls1:
if control1_flag == False:
control1_par = os.path.dirname(i)
control1_flag = True
il = control1_par + '/ms' + os.path.basename(i)
os.rename(i, il)
#control helper done
template_grays = glob(
os.path.join(outputs_path, 'registrations2/warped/ms*.nii.gz'))
templates = glob(
os.path.join(outputs_path, 'registrations1/warped1/ms*.nii.gz'))
fgs, distributions, a, adat, adat_raw = create_prob_seg_iteration3(
template_grays, templates, static, file_handl)
avgim(os.path.join(outputs_path, 'registrations2/warped/'))
#initialize images to write
sh = adat.shape
slope = np.zeros(sh)
screwed = 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)
file_handl.write(str(len(fgs)) + '\n')
file_handl.write(str(len(distributions)) + '\n')
distributions = np.asarray(distributions)
print('distributions:{}'.format(distributions.shape))
fgs = np.asarray(fgs)
file_handl.close()
adat_list = [z_score_a_dat(adat)]
count = 0
for adat_z in adat_list:
count += 1
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]
try:
params = np.polyfit(distributions[:, i, j], fgs[:, i, j],
1)
except:
screwed[i, j] = 1
continue
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
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:
try:
params = np.polyfit(distributions[:, i, j], fgs[:, i,
j], 1)
except:
continue
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:
try:
params, covs = np.polyfit(fgs[:, i, j],
distributions[:, i, j],
1,
cov=True)
except:
continue
slope[i, j] = params[0]
intercept[i, j] = params[1]
if abs(params[0]) < 2 * (covs[0, 0]**0.5):
new_image[i, j] = statss.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
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
try:
os.mkdir(os.path.join(outputs_path, 'final_output'))
except:
pass
#nb.save(nb.Nifti1Image(confidences,aff.affine),'/data/henry4/jjuwono/new_GM_method/'+mse+'/confidence.nii.gz')
#nb.save(nb.Nifti1Image(new_image,aff.affine),'/data/henry4/jjuwono/new_GM_method/'+mse+'/new_image.nii.gz')
#nb.save(nb.Nifti1Image(new_image_logi,aff.affine),'/data/henry4/jjuwono/new_GM_method/'+mse+'/new_image_logi.nii.gz')
nb.save(nb.Nifti1Image(t_map, aff.affine),
os.path.join(outputs_path, 'final_output/rereg_t_map.nii.gz'))
#nb.save(nb.Nifti1Image(color_im,aff.affine),'/data/henry4/jjuwono/new_GM_method/'+mse+'/color_im.nii.gz')
nb.save(
nb.Nifti1Image(original_line_fit, aff.affine),
os.path.join(outputs_path,
'final_output/rereg_original_line_fit.nii.gz'))
return 1