where the WholeBrain (WB) tractogram is a 20-Million-fiber whole-brain tractogram.
• My first question is whether using the 20M-fiber WB tractogram is a good approach or I should decrease the number of fibers of the tractogram from 20M to 2M and then run tckedit. I opted for the 20M-fiber tractogram since I get more fibers between the two ROIs, but I am not sure whether this is right. Since the threshold to consider that a tract exists at an individual level is usually set at 10 fibers, I am concerned that my results (the existence/not existence of that tract) are biased by the number of fibers of my WB tractogram. In other words, it is easier to find 10 fibers that pass through my ROIs with a 20M WB tractogram than with a 2M tractogram.
• My second question is whether I should use tckedit (with the WB tractogram) or tckgen (just considering my ROIs).
I apologize if my questions are too basic and thank you very much for the help you provide everyday.
I would use “true” (debatable) targeted tractography, i.e. tckgen directly every single day without a doubt for that scenario, especially if your aim is mostly “dissecting” (i.e. segmentation, potentially to create a ROI for further quantification, but not even per se). You’ll have much faster access and control of parameters which are only available during the tckgen step itself, such as step size, curvature constraints, even fancy things like the iFOD2 power parameter, etc… A lot of targeted tractography scenarios involve some level of supervision (which, I reckon, is what ultimately enables them to be reliable), and with that, a bit of trial and error. The best results are those that are semi-automated (i.e. semi-driven by the data), but also consciously inspected by someone with anatomical knowledge of what structure they are after. And the latter bit may lead to feedback to improve the parameters, both direct streamline tracking parameters (e.g. curvature and the like) as well as ROI induced constraints (inclusion, exclusion, masks, etc…).