Hi All,
I’m quite new to using MRTrix3 and have started running an analysis using tckgen. I’ve successfully generated a tractogram, however my analysis requires studying specific tracks like the superior longitudinal fasciculus and accordingly I am given these coordinates x = −35, y = −27, z = 44 mm. How would I go about achieving this output for visualisation?
Hi Yuscah,
There’s a few layers of both complexity and experimental flexibility here, so I’ll have to do my best to limit the scope of my response…
If I interpret your question more generally based on this specific extract, I would note that there are ways in which this could be done that would not involve the use of MNI coordinates. So you don’t necessarily want to invest huge amounts of effort into MNI coordinate transformation if it’s actually a bit of a red herring.
Let’s assume here that you simply want to determine the position in the image data of your individual subject that “corresponds” to a specific MNI coordinate. What this requires is:
Some form of image registration in order to achieve spatial correspondence between subject data and the MNI template;
Transformation of MNI coordinates from template space to subject space.
Point 1 will likely be best-posed if you have performed adequate distortion correction of your DWI data, and inter-modal registration, such that you can use the subject’s T1-weighed image for registration with the MNI template.
For Point 2, the esoteric details here become important. For the majority of experiments, one is content with simply being able to take image data that are defined on one voxel grid (e.g. the subject’s native image data), and resample those intensities onto another voxel grid (e.g. the template space), taking into account localised differences in brain geometry (i.e. the non-linear warping). In your case it sounds like you’re looking for something different: you want to determine an XYZ location in subject space that, if the non-linear warping were to be applied, it would arrive at location [-35, -27, 44] in template space. This information is somewhat already accessible from the results of non-linear registration, but requires an understanding of the information actually contained in non-linear warp fields (specifically what we refer to as deformation fields). A warp field for transforming data from subject space to template space is actually defined on the voxel grid of the template image, and stores, for every voxel in template space, the sub-voxel location in space from which image data should be “pulled” from the subject image. So if there is a voxel in the template image that is centred at [-35, -27, 44], the triplet of values within that voxel contains the position in real space that nominally applies to the equivalent anatomy within the subject’s image data.
This is obviously raw data; there is not any kind graphical user interface that allows you to type an MNI location into a text box and automagically highlights the equivalent location in the subject’s image. But I hope it at least gives you a sense of in which direction you may wish to invest your efforts.
Cheers
Rob