Multi-dimensional Searchlights

Note

This tutorial part is also available for download as an IPython notebook: [ipynb]

This is a little addendum to Event-related Data Analysis where we want to combine what we have learned about event-related data analysis and, at the same time, take a little glimpse on the power of PyMVPA for “multi-space” analysis.

First let’s re-create the dataset with the spatio-temporal features from Event-related Data Analysis:

>>> from mvpa2.tutorial_suite import *
>>> ds = get_raw_haxby2001_data(roi=(36,38,39,40))
>>> poly_detrend(ds, polyord=1, chunks_attr='chunks')
>>> zscore(ds, chunks_attr='chunks', param_est=('targets', 'rest'))
>>> events = find_events(targets=ds.sa.targets, chunks=ds.sa.chunks)
>>> events = [ev for ev in events if ev['targets'] in ['house', 'face']]
>>> event_duration = 13
>>> for ev in events:
...     ev['onset'] -= 2
...     ev['duration'] = event_duration
>>> evds = eventrelated_dataset(ds, events=events)

From the Looking here and there – Searchlights we know how to do searchlight analyses and it was promised that there is more to it than what we already saw. And here it is:

>>> cvte = CrossValidation(GNB(), NFoldPartitioner(),
...                        postproc=mean_sample())
>>> sl = Searchlight(cvte,
...                  IndexQueryEngine(voxel_indices=Sphere(1),
...                                   event_offsetidx=Sphere(2)),
...                  postproc=mean_sample())
>>> res = sl(evds)

Have you been able to deduce what this analysis will do? Clearly, it is some sort of searchlight, but it doesn’t use sphere_searchlight(). Instead, it utilizes Searchlight. Yes, you are correct this is a spatio-temporal searchlight. The searchlight focus travels along all possible locations in our ventral temporal ROI, but at the same time also along the peristimulus time segment covered by the events. The spatial searchlight extent is the center voxel and its immediate neighbors and the temporal dimension comprises of two additional time-points in each direction. The result is again a dataset. Its shape is compatible with the mapper of evds, hence it can also be back-projected into the original 4D fMRI brain space.

Searchlight is a powerful class that allows for complex runtime ROI generation. In this case it uses an IndexQueryEngine to look at certain feature attributes in the dataset to compose sphere-shaped ROIs in two spaces at the same time. This approach is very flexible and can be extended with additional query engines to algorithms of almost arbitrary complexity.

>>> ts = res.a.mapper.reverse1(1 - res.samples[0])
>>> # need to put the time axis last for export to NIfTI
>>> ts = np.rollaxis(ts, 0, 4)
>>> ni = nb.Nifti1Image(ts, ds.a.imgaffine).to_filename('ersl.nii')

After you are done and want to tidy up after yourself, you can easily remove unneeded generated files from within Python:

>>> os.unlink('ersl.nii')