def create_physio_brf(physiological_params,
response_dt=.5,
response_duration=25.,
return_brf_q_v=False):
"""
Generate a BOLD response function by integrating a physiological model and
setting its driving input signal to a single impulse.
Args:
- physiological_params (dict (<pname (str)> : <pvalue (float)>)):
parameters of the physiological model.
In jde.sandbox.physio see PHY_PARAMS_FRISTON00, PHY_PARAMS_FMRII ...
- response_dt (float): temporal resolution of the response, in second
- response_duration (float): duration of the response, in second
Return:
- np.array(nb_time_coeffs, float)
-> the BRF (normalized)
- also return brf_not_normalized, q, v when return_prf_q_v=True
(for error checking of v and q generation in calc_hrfs)
"""
p = Paradigm({'c': [np.array([0.])]}, [response_duration],
{'c': [np.array([1.])]})
n = np.array([[1.]])
s, f, v, q = create_evoked_physio_signals(physiological_params, p, n,
response_dt)
brf = create_bold_from_hbr_and_cbv(physiological_params, q[:, 0], v[:, 0])
if return_brf_q_v:
return brf / (
brf**2).sum()**.5, q, v, s, f #WARNING!! Added to compute figures
#return brf, q, v, s, f #WARNING!! Added to compute figures
else:
return brf / (brf**2).sum()**.5
def create_physio_prf(physiological_params,
response_dt=.5,
response_duration=25.,
return_prf_q_v=False):
"""
Generate a perfusion response function by setting the input driving signal
of the given physiological model with a single impulse.
Args:
- physiological_params (dict (<pname (str)> : <pvalue (float)>)):
parameters of the physiological model.
In jde.sandbox.physio see PHY_PARAMS_FRISTON00, PHY_PARAMS_FMRII ...
- response_dt (float): temporal resolution of the response, in second
- response_duration (float): duration of the response, in second
Return:
- np.array(nb_time_coeffs, float)
-> the PRF
- also return brf_not_normalized, q, v when return_prf_q_v=True
(for error checking of v and q generation in calc_hrfs)
"""
p = Paradigm({'c': [np.array([0.])]}, [response_duration],
{'c': [np.array([1.])]}) # response_dt to match convention
# in JDE analysis
n = np.array([[1.]])
s, f, v, q = create_evoked_physio_signals(physiological_params, p, n,
response_dt)
prf = f[:, 0] - f[0, 0] #remove y-intercept
if return_prf_q_v:
return prf / (prf**2).sum()**.5, q, v
#return prf, q, v
else:
return prf / (prf**2).sum()**.5
def __init__(self,
onsets,
bold,
tr,
sessionsScans,
roiMask,
graphs=None,
stimDurations=None,
meta_obj=None,
simulation=None,
backgroundLabel=0,
data_files=None,
data_type=None,
edge_lengths=None,
mask_loaded_from_file=False,
extra_data=None):
logger.info('Creation of FmriData object ...')
if isinstance(bold, xndarray):
logger.info('bold shape: %s', str(bold.data.shape))
else:
logger.info('bold shape: %s', str(bold.shape))
logger.info('roi mask shape: %s', str(roiMask.shape))
logger.info('unique(mask): %s', str(np.unique(roiMask)))
sessionsDurations = [len(ss) * tr for ss in sessionsScans]
self.paradigm = Paradigm(onsets, sessionsDurations, stimDurations)
self.tr = tr
self.sessionsScans = sessionsScans
self.extra_data = extra_data or {}
if backgroundLabel is None:
backgroundLabel = 0
self.backgroundLabel = backgroundLabel
logger.info('backgroundLabel: %d', self.backgroundLabel)
self.store_mask_sparse(roiMask)
m = self.np_roi_mask
if isinstance(bold, xndarray):
if bold.data.ndim > 2: # volumic data
print '----------------'
if data_type is None:
data_type = 'volume'
assert roiMask.ndim == 3
if TIME_AXIS == 0:
bold = bold.data[:, m[0], m[1], m[2]]
else: # TIME_AXIS = 3
bold = bold.data[m[0], m[1], m[2], :].transpose()
else: # surfacic or already flatten data
assert bold.data.ndim == 2
if len(m[0]) != bold.data.shape[1]:
bold = bold.data[:, m[0]]
else:
bold = bold.data
else:
if bold.ndim > 2: # volumic data
if data_type is None:
data_type = 'volume'
assert roiMask.ndim == 3
if TIME_AXIS == 0:
bold = bold[:, m[0], m[1], m[2]]
else: # TIME_AXIS = 3
bold = bold[m[0], m[1], m[2], :].transpose()
else: # surfacic or already flatten data
assert bold.ndim == 2
if len(m[0]) != bold.shape[1]:
bold = bold[:, m[0]]
self.data_type = data_type
logger.info('extracted bold shape (without background): %s',
str(bold.shape))
self.bold = bold
self.bold_full = bold
self.bold_avg = None
self.graphs = graphs
self.edge_lengths = edge_lengths
self.graphs_full = graphs
self.graphs_avg = None
self.meta_obj = meta_obj
self.simulation = simulation
self.nbConditions = len(self.paradigm.stimOnsets)
self.nbSessions = len(sessionsScans)
self.mask_loaded_from_file = mask_loaded_from_file
if data_files is None:
data_files = []
self.data_files = data_files
