def plscorr_eval(train_fmri_ts, train_feat_ts, val_fmri_ts, val_feat_ts,
out_dir, mask_file):
"""Compute PLS correlation between brain activity and CNN activation."""
train_feat_ts = train_feat_ts.reshape(-1, train_feat_ts.shape[3]).T
val_feat_ts = val_feat_ts.reshape(-1, val_feat_ts.shape[3]).T
train_fmri_ts = train_fmri_ts.T
val_fmri_ts = val_fmri_ts.T
# Iteration loop for different component number
#for n in range(5, 19):
# print '--- Components number %s ---' %(n)
# plsca = PLSCanonical(n_components=n)
# plsca.fit(train_feat_ts, train_fmri_ts)
# pred_feat_c, pred_fmri_c = plsca.transform(val_feat_ts, val_fmri_ts)
# pred_fmri_ts = plsca.predict(val_feat_ts)
# # calculate correlation coefficient between truth and prediction
# r = corr2_coef(val_fmri_ts.T, pred_fmri_ts.T, mode='pair')
# # get top 20% corrcoef for model evaluation
# vsample = int(np.rint(0.2*len(r)))
# print 'Sample size for evaluation : %s' % (vsample)
# r.sort()
# meanr = np.mean(r[-1*vsample:])
# print 'Mean prediction corrcoef : %s' %(meanr)
# model generation based on optimized CC number
cc_num = 10
plsca = PLSCanonical(n_components=cc_num)
plsca.fit(train_feat_ts, train_fmri_ts)
from sklearn.externals import joblib
joblib.dump(plsca, os.path.join(out_dir, 'plsca_model.pkl'))
plsca = joblib.load(os.path.join(out_dir, 'plsca_model.pkl'))
# calculate correlation coefficient between truth and prediction
pred_fmri_ts = plsca.predict(val_feat_ts)
fmri_pred_r = corr2_coef(val_fmri_ts.T, pred_fmri_ts.T, mode='pair')
mask = vutil.data_swap(mask_file)
vxl_idx = np.nonzero(mask.flatten() == 1)[0]
tmp = np.zeros_like(mask.flatten(), dtype=np.float64)
tmp[vxl_idx] = fmri_pred_r
tmp = tmp.reshape(mask.shape)
vutil.save2nifti(tmp, os.path.join(out_dir, 'pred_fmri_r.nii.gz'))
pred_feat_ts = pls_y_pred_x(plsca, val_fmri_ts)
pred_feat_ts = pred_feat_ts.T.reshape(96, 14, 14, 540)
np.save(os.path.join(out_dir, 'pred_feat.npy'), pred_feat_ts)
# get PLS-CCA weights
feat_cc, fmri_cc = plsca.transform(train_feat_ts, train_fmri_ts)
np.save(os.path.join(out_dir, 'feat_cc.npy'), feat_cc)
np.save(os.path.join(out_dir, 'fmri_cc.npy'), fmri_cc)
feat_weight = plsca.x_weights_.reshape(96, 14, 14, cc_num)
#feat_weight = plsca.x_weights_.reshape(96, 11, 11, cc_num)
fmri_weight = plsca.y_weights_
np.save(os.path.join(out_dir, 'feat_weights.npy'), feat_weight)
np.save(os.path.join(out_dir, 'fmri_weights.npy'), fmri_weight)
fmri_orig_ccs = get_pls_components(plsca.y_scores_, plsca.y_loadings_)
np.save(os.path.join(out_dir, 'fmri_orig_ccs.npy'), fmri_orig_ccs)
def reg_cca(train_fmri_ts, train_feat_ts, val_fmri_ts, val_feat_ts, out_dir):
"""Conduct CCA between brain activity and CNN activation."""
train_feat_ts = train_feat_ts.reshape(-1, train_feat_ts.shape[3]).T
val_feat_ts = val_feat_ts.reshape(-1, val_feat_ts.shape[3]).T
train_fmri_ts = train_fmri_ts.T
val_fmri_ts = val_fmri_ts.T
#-- model training
# for reduce complexity, a linear kernel is used
#cca = rcca.CCACrossValidate(numCCs=[7, 8, 9, 10, 11, 12, 13],
# kernelcca=True)
CCnum = 7
#cca = rcca.CCA(kernelcca=True, reg=0.007743, numCC=CCnum)
#cca.train([train_feat_ts, train_fmri_ts])
#cca.validate([val_feat_ts, val_fmri_ts])
#cca.compute_ev([val_feat_ts, val_fmri_ts])
#print 'Best CC number : %s' %(cca.best_numCC)
#print 'Best reg : %s' %(cca.best_reg)
out_file = os.path.join(out_dir, 'CCA_results_%s.hdf5' % (CCnum))
#cca.save(os.path.join(out_file))
#-- model exploring
mask_file = r'/Users/sealhuang/brainDecoding/S1_mask.nii.gz'
cca = rcca.CCA()
cca.load(out_file)
# model prediction performance
fmri_pred_r = cca.corrs[1]
feat_pred_r = cca.corrs[0].reshape(96, 11, 11)
vutil.plot_cca_fweights(feat_pred_r, out_dir, 'pred_feat_r_CC%s' % (CCnum))
mask = vutil.data_swap(mask_file)
vxl_idx = np.nonzero(mask.flatten() == 1)[0]
tmp = np.zeros_like(mask.flatten(), dtype=np.float64)
tmp[vxl_idx] = fmri_pred_r
tmp = tmp.reshape(mask.shape)
vutil.save2nifti(
tmp, os.path.join(out_dir, 'pred_fmri_r_CC%s.nii.gz' % (CCnum)))
# model weights visualization
feat_weights = cca.ws[0]
feat_weights = feat_weights.reshape(96, 11, 11, feat_weights.shape[1])
fmri_weights = cca.ws[1]
vutil.plot_cca_fweights(feat_weights, out_dir,
'feat_weight_CC%s' % (CCnum))
vutil.save_cca_volweights(fmri_weights, mask_file, out_dir,
'cca_component')
feat_cc = cca.comps[0]
parallel_corr2_coef(train_feat_ts.T,
feat_cc.T,
os.path.join(out_dir, 'feat_cc_corr.npy'),
block_size=7,
n_jobs=1)
feat_cc_corr = np.load(os.path.join(out_dir, 'feat_cc_corr.npy'))
feat_cc_corr = feat_cc_corr.reshape(96, 11, 11, 7)
vutil.plot_cca_fweights(feat_cc_corr, out_dir, 'feat_cc_corr')
def prf2visual_angle(prf_mtx, img_size, out_dir, base_name):
"""Generate retinotopic mapping based on voxels' pRF parameters.
`prf_mtx` is a #voxel x pRF-features matrix, pRF features can be 2 columns
(row, col) of image or 3 columns which adding a third pRF size parameters.
"""
feature_size = prf_mtx.shape[1]
pos_mtx = prf_mtx[:, :2]
# eccentricity
ecc = retinotopy.coord2ecc(pos_mtx, img_size, 20)
vol = ecc.reshape(18, 64, 64)
vutil.save2nifti(vol, os.path.join(out_dir, base_name + '_ecc.nii.gz'))
# angle
angle = retinotopy.coord2angle(pos_mtx, img_size)
vol = angle.reshape(18, 64, 64)
vutil.save2nifti(vol, os.path.join(out_dir, base_name + '_angle.nii.gz'))
# pRF size
if feature_size > 2:
size_angle = retinotopy.get_prf_size(prf_mtx, 55, 20)
vol = size_angle.reshape(18, 64, 64)
vutil.save2nifti(vol, os.path.join(out_dir,
base_name + '_size.nii.gz'))
def retinotopic_mapping(corr_file, data_dir, vxl_idx=None, figout=False):
"""Make the retinotopic mapping using activation map from CNN."""
if figout:
fig_dir = os.path.join(data_dir, 'fig')
check_path(fig_dir)
# load the cross-correlation matrix from file
corr_mtx = np.load(corr_file, mmap_mode='r')
# set voxel index
if not isinstance(vxl_idx, np.ndarray):
vxl_idx = np.arange(corr_mtx.shape[0])
elif len(vxl_idx) != corr_mtx.shape[0]:
print 'mismatch on voxel number!'
return
else:
print 'voxel index loaded.'
pos_mtx = np.zeros((73728, 2))
pos_mtx[:] = np.nan
for i in range(len(vxl_idx)):
print 'Iter %s of %s' % (i + 1, len(vxl_idx)),
tmp = corr_mtx[i, :]
tmp = np.nan_to_num(np.array(tmp))
# significant threshold for one-tail test
tmp[tmp <= 0.019257] = 0
if np.sum(tmp):
tmp = tmp.reshape(96, 27, 27)
mmtx = np.max(tmp, axis=0)
print mmtx.min(), mmtx.max()
# get indices of n maximum values
max_n = 20
row_idx, col_idx = np.unravel_index(
np.argsort(mmtx.ravel())[-1 * max_n:], mmtx.shape)
nmtx = np.zeros(mmtx.shape)
nmtx[row_idx, col_idx] = mmtx[row_idx, col_idx]
if figout:
fig_file = os.path.join(fig_dir,
'v' + str(vxl_idx[i]) + '.png')
imsave(fig_file, nmtx)
# center of mass
x, y = ndimage.measurements.center_of_mass(nmtx)
pos_mtx[vxl_idx[i], :] = [x, y]
else:
print ' '
#receptive_field_file = os.path.join(data_dir, 'receptive_field_pos.npy')
#np.save(receptive_field_file, pos_mtx)
#pos_mtx = np.load(receptive_field_file)
# eccentricity
dist = retinotopy.coord2ecc(pos_mtx, (27, 27))
# convert distance into degree
# 0-4 degree -> d < 5.5
# 4-8 degree -> d < 11
# 8-12 degree -> d < 16.5
# 12-16 degree -> d < 22
# else > 16 degree
ecc = np.zeros(dist.shape)
for i in range(len(dist)):
if np.isnan(dist[i]):
ecc[i] = np.nan
elif dist[i] < 2.7:
ecc[i] = 1
elif dist[i] < 5.4:
ecc[i] = 2
elif dist[i] < 8.1:
ecc[i] = 3
elif dist[i] < 10.8:
ecc[i] = 4
else:
ecc[i] = 5
#dist_vec = np.nan_to_num(ecc)
#vol = dist_vec.reshape(18, 64, 64)
vol = ecc.reshape(18, 64, 64)
vutil.save2nifti(
vol, os.path.join(data_dir, 'train_max' + str(max_n) + '_ecc.nii.gz'))
# angle
angle_vec = retinotopy.coord2angle(pos_mtx, (27, 27))
#angle_vec = np.nan_to_num(angle_vec)
vol = angle_vec.reshape(18, 64, 64)
vutil.save2nifti(
vol, os.path.join(data_dir,
'train_max' + str(max_n) + '_angle.nii.gz'))
def ridge_retinotopic_mapping(corr_file, vxl_idx=None, top_n=None):
"""Make the retinotopic mapping using activation map from CNN."""
data_dir = os.path.dirname(corr_file)
# load the cross-correlation matrix from file
corr_mtx = np.load(corr_file, mmap_mode='r')
# corr_mtx.shape = (3025, vxl_num)
if not isinstance(vxl_idx, np.ndarray):
vxl_idx = np.arange(corr_mtx.shape[1])
if not top_n:
top_n = 20
pos_mtx = np.zeros((73728, 2))
pos_mtx[:] = np.nan
for i in range(len(vxl_idx)):
print 'Iter %s of %s' % (i, len(vxl_idx)),
tmp = corr_mtx[:, i]
tmp = np.nan_to_num(np.array(tmp))
# significant threshold
# one-tail test
#tmp[tmp <= 0.17419] = 0
if np.sum(tmp):
tmp = tmp.reshape(55, 55)
print tmp.min(), tmp.max()
# get indices of n maximum values
row, col = np.unravel_index(
np.argsort(tmp.ravel())[-1 * top_n:], tmp.shape)
mtx = np.zeros(tmp.shape)
mtx[row, col] = tmp[row, col]
# center of mass
x, y = ndimage.measurements.center_of_mass(mtx)
pos_mtx[vxl_idx[i], :] = [x, y]
else:
print ' '
#receptive_field_file = os.path.join(data_dir, 'receptive_field_pos.npy')
#np.save(receptive_field_file, pos_mtx)
#pos_mtx = np.load(receptive_field_file)
# eccentricity
dist = retinotopy.coord2ecc(pos_mtx, (55, 55))
# convert distance into degree
# 0-4 degree -> d < 5.5
# 4-8 degree -> d < 11
# 8-12 degree -> d < 16.5
# 12-16 degree -> d < 22
# else > 16 degree
ecc = np.zeros(dist.shape)
for i in range(len(dist)):
if np.isnan(dist[i]):
ecc[i] = np.nan
elif dist[i] < 5.445:
ecc[i] = 1
elif dist[i] < 10.91:
ecc[i] = 2
elif dist[i] < 16.39:
ecc[i] = 3
elif dist[i] < 21.92:
ecc[i] = 4
else:
ecc[i] = 5
vol = ecc.reshape(18, 64, 64)
vutil.save2nifti(vol, os.path.join(data_dir, 'ecc_max%s.nii.gz' % (top_n)))
# angle
angle_vec = retinotopy.coord2angle(pos_mtx, (55, 55))
vol = angle_vec.reshape(18, 64, 64)
vutil.save2nifti(vol, os.path.join(data_dir,
'angle_max%s.nii.gz' % (top_n)))