def roi_barplot(mask, mask_name, contrasts, labels, results_path, title_extra=''):
roi_data = pd.DataFrame(np.zeros(shape=(mask.sum(), len(contrasts))),
columns=labels)
for idx, img in enumerate([contrast_path.format(c) for c in contrasts]):
roi_data[labels[idx]] = rm.extract_data_in_mask(img, mask)
fig = plt.figure()
ax = fig.add_subplot(1, 1, 1)
fig.suptitle(mask_name+'\n'+title_extra)
sns.barplot(data=roi_data, ax=ax)
fig.savefig(os.path.join(results_path, mask_name+'.png'))
plt.close()
def roi_histogram(mask, mask_name, imgs, results_path, title_extra=''):
roi_data = []
for idx, img in enumerate(imgs):
roi_data.append(rm.extract_data_in_mask(img, mask))
roi_data = np.concatenate(roi_data)
fig = plt.figure()
ax = fig.add_subplot(1, 1, 1)
fig.suptitle(mask_name+'\n'+title_extra)
sns.distplot(roi_data, ax=ax)
fig.savefig(os.path.join(results_path, mask_name+'.png'))
plt.close()
def roi_statistical_model(mask, mask_name, contrasts, labels, factors,
results_path, model_desc):
# factors should be a pandas dataframe with named columns
# factors correspond to labels by idx
roi_data = pd.DataFrame(np.zeros(shape=(mask.sum()*len(contrasts),
len(factors.columns)+3)),
columns=['condition']+factors.columns.tolist() + \
['voxel', 'beta']
)
counter = 0
print 'preparing data in ' + mask_name
for idx, img in enumerate([contrast_path.format(c) for c in contrasts]):
data_cat = [labels[idx]] + factors.loc[idx].tolist()
for v_idx, beta in enumerate(rm.extract_data_in_mask(img, mask)):
row = data_cat + [v_idx, beta[0]]
roi_data.loc[counter] = row
counter += 1
print 'computing model: ' + model_desc + ' in ' + mask_name
regress = sm.ols(model_desc, data=roi_data).fit()
f = open(os.path.join(results_path, mask_name+'.txt'), 'w')
f.write(regress.summary().as_text())
f.close()
contrasts = [20, 40]
assert(len(contrasts) == len(labels))
gm_img = os.path.join('pipelines', 'preprocessing', subject,
'highres001_graymatter_mask_bold_native.nii')
gm_mask = rm.get_roi_mask(gm_img)
diff_map = contrast_path.format(40)
inter_map = contrast_path.format(20)
df2 = 219*8 - 32
pedges = [1, 0.01, 0.009, 0.008, 0.007, 0.006, 0.005, 0.004, 0.003, 0.002, 0.001, 0]
pedges.reverse()
for roi in rm.get_rois_mask_in_native_space(sub_id):
if not os.path.exists(os.path.join(results_path, *roi[2])):
os.makedirs(os.path.join(results_path, *roi[2]))
gm_adj_roi = gm_mask * roi[0]
diff = np.ravel(rm.extract_data_in_mask(diff_map, gm_adj_roi))
inter = np.ravel(rm.extract_data_in_mask(inter_map, gm_adj_roi))
diffp = [-np.log10(sst.f.sf(x, 15, df2)) for x in diff]
interp = [-np.log10(sst.f.sf(x, 9, df2)) for x in inter]
# fig = plt.figure()
# ax = fig.add_subplot(111)
extra_text = str([math.ceil(x) for x in rm.get_roi_center(roi[3], roi[4])[3]])
g = sns.jointplot(x=np.array(diffp), y=np.array(interp), kind='scatter', stat_func=None, xlim=(0,np.max(diffp)+1), ylim=(0,np.max(interp)+1)).set_axis_labels("-log10(p_value) number difference", "-log10(p_value) interaction")
# , xlim=(0,1), ylim=(0,1)
# plt.show()
# H, xedges, yedges = np.histogram2d(interp, diffp, bins=(pedges, pedges),
# normed=False)
# fig = plt.figure()
# ax = fig.add_subplot(111)
extra_text = str([math.ceil(x) for x in rm.get_roi_center(roi[3], roi[4])[3]])
