def gen_module_roi(root_dir):
"""Make ROI file based on Power264 atlas."""
roi_info = open('sel_emotion_rois.csv').readlines()
roi_info.pop(0)
roi_info = [line.strip().split(',') for line in roi_info]
mask = np.zeros((91, 109, 91))
for line in roi_info:
i = int((90.0 - float(line[3])) / 2)
j = int((float(line[4]) + 126) / 2)
k = int((float(line[5]) + 72) / 2)
label = int(line[1])
for n_x in range(i-2, i+3):
for n_y in range(j-2, j+3):
for n_z in range(k-2, k+3):
try:
if mask[n_x, n_y, n_z]>0:
mask[n_x, n_y, n_z] = 1000
else:
mask[n_x, n_y, n_z] = label
except:
pass
mask[mask==1000] = 0
mni_vol = os.path.join(os.environ['FSL_DIR'], 'data', 'standard',
'MNI152_T1_2mm_brain.nii.gz')
aff = nib.load(mni_vol).affine
outfile ='sel_emotion_modules.nii.gz'
nibase.save2nifti(mask, aff, outfile)
def get_mean_cope(root_dir, subj):
# dir config
work_dir = os.path.join(root_dir, 'workshop', 'searchlight')
# load nii data list
print 'Load nii files ...'
cope_list = get_subj_cope_list(root_dir, subj)
# get trial sequence info
print 'Load trial sequence info ...'
tag_list = get_subj_cope_tag(root_dir, subj)
for i in range(10):
mean_cope = np.zeros((91, 109, 91, 4))
copes = cope_list[i]
tag = np.array(tag_list[i][:72])
for c in range(4):
cond_idx = tag == (c + 1)
cond_cope = copes[..., cond_idx]
mean_cope[..., c] = np.mean(cond_cope, axis=3)
# save to nifti
fsl_dir = os.getenv('FSL_DIR')
template_file = os.path.join(fsl_dir, 'data', 'standard',
'MNI152_T1_2mm_brain.nii.gz')
aff = nib.load(template_file).affine
nibase.save2nifti(
mean_cope, aff,
os.path.join(work_dir, subj + '_mean_copes_%s.nii.gz' % (i + 1)))
def get_mean_emo_activation(root_dir, subj):
"""Get mean activation for each emotion category based on single run's data.
"""
work_dir = os.path.join(root_dir, 'workshop', 'searchlight')
# load nii data list
print 'Load nii files ...'
act_list = get_subj_act_list(root_dir, subj)
# get trial sequence info
print 'Load trial sequence info ...'
tag_list = get_subj_trial_seq(root_dir, subj)
for i in range(10):
mean_act = np.zeros((91, 109, 91, 4))
act = act_list[i]
tag = tag_list[i]
for c in range(4):
tmp_idx = []
for k in tag:
if tag[k][1]==(c+1):
tmp_idx = tmp_idx + tag[k][0]
tmp_act = act[..., np.array(tmp_idx)]
mean_act[..., c] = np.mean(tmp_act, axis=3)
# save to nifti
fsl_dir = os.getenv('FSL_DIR')
template_file = os.path.join(fsl_dir, 'data', 'standard',
'MNI152_T1_2mm_brain.nii.gz')
aff = nib.load(template_file).affine
nibase.save2nifti(mean_act, aff,
os.path.join(work_dir, subj+'_mean_act_%s.nii.gz'%(i+1)))
def mat2nii(mat_structure, data_name, template_file, out_file):
"""save `data_name` from `mat_structure` to nii file.
`data_name` can be one of `r2_train`, `r2_val`, `beta_train`, `beta_val`
and `hrfs`.
"""
data = mat_structure[data_name]
img = nib.load(template_file)
aff = img.affine
save2nifti(data, aff, out_file)
def save2nifti(data, output_file):
"""
Save a nifti file.
"""
fsl_dir = os.getenv('FSL_DIR')
template_file = os.path.join(fsl_dir, 'data', 'standard',
'MNI152_T1_2mm_brain.nii.gz')
header = nib.load(template_file).get_header()
mybase.save2nifti(data, header, output_file)
def random_svm_cope_searchlight(root_dir, subj):
"""SVM based searchlight analysis."""
# dir config
work_dir = os.path.join(root_dir, 'workshop', 'searchlight')
# read mask file
print 'Load mask data ...'
mask_file = os.path.join(work_dir, 'mask', 'func_mask.nii.gz')
mask_data = nib.load(mask_file).get_data()
mask_data = mask_data > 0
# load nii data list
print 'Load nii files ...'
cope_list = get_subj_cope_list(root_dir, subj)
# get trial sequence info
print 'Load trial sequence info ...'
tag_list = get_subj_cope_tag(root_dir, subj)
for i in range(100):
# svm results var
clf_results = np.zeros((91, 109, 91, 4))
# for loop for voxel-wise searchlight
mask_coord = niroi.get_roi_coord(mask_data)
ccount = 0
for c in mask_coord:
ccount += 1
print ccount
cube_roi = np.zeros((91, 109, 91))
cube_roi = niroi.cube_roi(cube_roi, c[0], c[1], c[2], 2, 1)
cube_coord = niroi.get_roi_coord(cube_roi)
[train_x, train_y, test_x,
test_y] = get_roi_cope_mvps(cope_list, tag_list, cube_coord)
clf = svm.SVC(kernel='sigmoid')
train_y = np.random.permutation(train_y)
clf.fit(train_x, train_y)
pred = clf.predict(test_x)
for e in range(4):
acc = np.sum(pred[test_y == (e + 1)] ==
(e + 1)) * 1.0 / np.sum(test_y == (e + 1))
print acc
clf_results[c[0], c[1], c[2], e] = acc
# save to nifti
fsl_dir = os.getenv('FSL_DIR')
template_file = os.path.join(fsl_dir, 'data', 'standard',
'MNI152_T1_2mm_brain.nii.gz')
aff = nib.load(template_file).affine
nibase.save2nifti(
clf_results, aff,
os.path.join(
work_dir,
'random_' + subj + '_svm_acc_cope_%s.nii.gz' % (i + 1)))
def z2r():
"""
Do a Fisher r-to-z transform.
"""
source_dir = r'/nfs/h1/workingshop/huanglijie/uni_mul_analysis/multivariate'
input = os.path.join(source_dir, 'beh_corr', 'rmet', 'merged_data.nii.gz')
output = os.path.join(source_dir, 'beh_corr', 'rmet',
'merged_data_r2z.nii.gz')
data = nib.load(input).get_data()
header = nib.load(input).get_header()
one_ele = np.sum(data == 1)
print one_ele
data[data==1] = 0.999999
data[data==-1] = -0.999999
one_ele = np.sum(data == 1)
print one_ele
zdata = np.log((1+data)/(1-data)) / 2
nibase.save2nifti(zdata, header, output)
def refine_rois(root_dir):
"""Refine ROIs."""
orig_roi_file = os.path.join(root_dir, 'group-level', 'rois', 'neurosynth',
'merged_hfdn_mask_Tmax_s2_lmax_roi_orig.nii.gz')
roi_info_file = os.path.join(root_dir, 'group-level', 'rois', 'neurosynth',
'new_neurosynth_roi_info.csv')
roi_info = open(roi_info_file, 'r').readlines()
roi_info = [line.strip().split(',') for line in roi_info]
roi_info.pop(0)
# refine rois
orig_roi = nib.load(orig_roi_file).get_data()
new_roi = np.zeros_like(orig_roi)
for line in roi_info:
oid = int(line[1])
nid = int(line[0])
new_roi[orig_roi==oid] = nid
# save file
new_roi_file = os.path.join(root_dir, 'group-level', 'rois', 'neurosynth',
'merged_hfdn_mask_Tmax_s2_lmax_roi.nii.gz')
aff = nib.load(orig_roi_file).affine
nibase.save2nifti(new_roi, aff, new_roi_file)
def power264roi(root_dir):
"""Make ROI file based on Power264 atlas."""
roi_info = open('power264.csv').readlines()
roi_info.pop(0)
roi_info = [line.strip().split(',') for line in roi_info]
roi_dict = {}
roi_label_dict = {}
for line in roi_info:
if not line[5] in roi_dict:
roi_dict[line[5]] = {}
roi_label_dict[line[5]] = {}
i = int((90.0 - int(line[2])) / 2)
j = int((int(line[3]) + 126) / 2)
k = int((int(line[4]) + 72) / 2)
roi_dict[line[5]][int(line[0])] = [i, j, k]
roi_label_dict[line[5]][int(line[0])] = line[1]
# create cube roi based on center coord
#for m in roi_dict:
# centers = roi_dict[m]
# mask = np.zeros((91, 109, 91))
# for c in centers:
# mask = niroi.cube_roi(mask, centers[c][0], centers[c][1],
# centers[c][2], 2, c)
# mni_vol = os.path.join(os.environ['FSL_DIR'], 'data', 'standard',
# 'MNI152_T1_2mm_brain.nii.gz')
# aff = nib.load(mni_vol).affine
# outfile ='power264_%s_rois.nii.gz'%(m.replace('/','-').replace(' ','-'))
# nibase.save2nifti(mask, aff, outfile)
#sel_module = ['Salience', 'Visual', 'Subcortical',
# 'Cingulo-opercular Task Control', 'Default mode',
# 'Fronto-parietal Task Control']
sel_module = roi_dict.keys()
froi = open('power264_roi.csv', 'w')
froi.write('RID,FSL_label,X,Y,Z,Module\n')
count = 1
mask = np.zeros((91, 109, 91))
for m in sel_module:
centers = roi_dict[m]
labels = roi_label_dict[m]
for c in centers:
x = centers[c][0]
y = centers[c][1]
z = centers[c][2]
for n_x in range(x-2, x+3):
for n_y in range(y-2, y+3):
for n_z in range(z-2, z+3):
try:
if mask[n_x, n_y, n_z]>0:
mask[n_x, n_y, n_z] = 1000
else:
mask[n_x, n_y, n_z] = count
except:
pass
#mask = niroi.cube_roi(mask, centers[c][0], centers[c][1],
# centers[c][2], 2, count)
froi.write(','.join([str(count), labels[c], str(centers[c][0]),
str(centers[c][1]), str(centers[c][2]),
m])+'\n')
count += 1
froi.close()
mask[mask==1000] = 0
mni_vol = os.path.join(os.environ['FSL_DIR'], 'data', 'standard',
'MNI152_T1_2mm_brain.nii.gz')
aff = nib.load(mni_vol).affine
outfile ='power264_rois.nii.gz'
nibase.save2nifti(mask, aff, outfile)