We have processed DWI data in two resolutions using QSIPrep version 1.0.0, which leaves the T1W and DWI data in ACPC space orientation but at their respective native dimensions.
Next, we have segmented T1W images using Freesurfer and THOMAS segmentation techniques to obtain the respective segmentation ROIs. We have obtained
Query: Can we use the Mrtrix based tckgen to generate tractograms between two ROIs with the WhiteMatter_FODs from either of the DWI data at their native resolution but providing ROIs from segmented T1W (in native T1W resolution)?
As such we were able to perform tckgen to generate tracts between two ROIs successfully. We are curious to know if the dimensions mismatch will lead to any undesired results.
The tractography in MRtrix3 is very intentionally designed to not require that different input images be defined on the same image grid. The streamline vertices are defined in “real” / “scanner” space; as they propagate, image data from any input source—whether it be DWI / FOD data, 5TT tissue segmentations, or ROIs—are sampled by interpolating at the corresponding sub-voxel location in the image. This is done independently for those different possible input image sources. So it does not matter if each of them is defined on a different voxel grid, or even if they cover different fields of view. All that matters is that the underlying content of those images overlaps in “real” / “scanner” space.
