Hi,
On Wed, Dec 9, 2015 at 6:20 AM, Helmut Nebl <h....@gm...> wrote:
> Dear everyone,
>
> How does Marsbar deal with roi.mat files in case their resolution differs
> from that of the analysis / niftis of the model? Say my ROIs were imported
> from a brain atlas with 1x1x1 mm^3 but the analysis is based on images with
> 1.5x1.5x1.5 mm^3. Does Marsbar do some interpolation on the ROI files then,
> e.g. nearest neighbour, which is then applied in an all-or-none fashion onto
> the data? Or will it weight voxels of the images to be analyzed accordingly,
> e.g. voxels falling entirely inside the ROI counting 100% and some
> "boundardy" voxels counting to a lesser extent depending on overlap between
> ROI and image?
It just resamples the data at the locations defined by the ROI points:
https://github.com/matthew-brett/marsbar/blob/master/marsbar/%40maroi/getdata.m#L97
In your example, for every 1x1x1mm voxel location in the ROI, it will
calculate the (probably non-integer) voxel coordinate in the
1.5x1.5x1.5mm image, and resample from that voxel coordinate.
Best,
Matthew
|
