def test_cosine_drift():
# add something so that when the tests are launched
#from a different directory
spm_drifts = DESIGN_MATRIX['cosbf_dt_1_nt_20_hcut_0p1']
tim = np.arange(20)
P = 10 # period is half the time, gives us an order 4
nistats_drifts = _cosine_drift(P, tim)
assert_almost_equal(spm_drifts[:, 1:], nistats_drifts[:, :-1])
def test_cosine_drift():
# add something so that when the tests are launched
#from a different directory
spm_drifts = DESIGN_MATRIX['cosbf_dt_1_nt_20_hcut_0p1']
tim = np.arange(20)
P = 10 # period is half the time, gives us an order 4
nistats_drifts = _cosine_drift(P, tim)
assert_almost_equal(spm_drifts[:, 1:], nistats_drifts[:, : - 1])
def test_cosine_drift():
# add something so that when the tests are launched
#from a different directory
spm_drifts = DESIGN_MATRIX['cosbf_dt_1_nt_20_hcut_0p1']
frame_times = np.arange(20)
high_pass_frequency = .1
nistats_drifts = _cosine_drift(high_pass_frequency, frame_times)
assert_almost_equal(spm_drifts[:, 1:], nistats_drifts[:, :-2])
def _run_preproc_in_parallel(run, event, conf, func, ricor, task, space, conf_vars, mask, hp_cutoff,
add_motion_outliers, smoothing_fwhm, df_filter, slice_time_ref, tr, logger):
logger.info(f"Preprocessing run {run+1} ...")
if 'fs' in space:
func_ts = np.vstack([d.data[np.newaxis, :] for
d in nib.load(func).darrays])
else:
if not np.all(nib.load(func).affine == mask.affine):
print(f"Resampling run {run+1} to mask affine, because they are different")
func = image.resample_img(func, target_affine=mask.affine, target_shape=mask.shape)
func_ts = masking.apply_mask(
imgs=func,
mask_img=mask,
smoothing_fwhm=smoothing_fwhm
)
n_vols = func_ts.shape[0]
start_time = slice_time_ref * tr
end_time = (n_vols - 1 + slice_time_ref) * tr
frame_times = np.linspace(start_time, end_time, n_vols)
event = pd.read_csv(event, sep='\t')
if df_filter is not None:
event = df_filter(event)
conf = pd.read_csv(conf, sep='\t')
all_conf_vars = conf_vars
if add_motion_outliers:
mo = [col for col in conf.columns if 'motion_outlier' in col]
if len(mo) > 0:
logger.info(f"Adding {len(mo)} motion outliers to design for run {run+1}.")
all_conf_vars += mo
if hp_cutoff is None:
all_conf_vars += [col for col in conf.columns if 'cosine' in col]
conf = conf.loc[:, all_conf_vars].fillna(0)
conf = pd.concat((conf, cos_df), axis=1)
else:
conf = conf.loc[:, all_conf_vars].fillna(0)
cos = _cosine_drift(hp_cutoff, frame_times)[:, :-1]
cos_names = [f'cosine{ic}'.zfill(3) for ic in range(cos.shape[1])]
cos_df = pd.DataFrame(data=cos, columns=cos_names, index=conf.index)
conf = pd.concat((conf, cos_df), axis=1)
if ricor is not None:
ricor_df = pd.read_csv(ricor, sep='\t')
conf = pd.concat((conf, ricor_df), axis=1)
return func_ts, conf, event