Abnormal cerebellar connections with nearby regions

Hi MRtrix experts,

I am getting unusually high connections between the cerebellum (as a single node) and a region in the temporooccipital area that is near the cerebellum. The tracks are not going through peduncles, but rather directly from CSF (see the cursor). The search criterion in tck2connectome was assigned with radial search = 2mm. I am looking for any suggestions in this regard.

Processing pipeline and data:
Data [b=0,1000 in 64 dir, 2mm isotropic voxel-size], Preprocessing using b0 generated by Synb0-DISCO, wmfod msmt_csd -shells 0,1000 [for only wm and csf], mtnormalise, 5ttgen hsvs, registered to dMRI with bbregister, tckgen with ACT and dynamic seeding [-cutoff 0.06, maxlength = 300], tcksift2, tck2connectome using Schaefer400+Melbourne’s subcortical+cerebellum mapped/registered to subject space using ANTs/FreeSurfer


p.s. As a suggestion, I think when implementing denser atlases (e.g. Shaefer’s>200) requires another fiber assignment method compared to radial search, as it may falsely assign nodes in highly folded cortical areas (esp. occipital). I don’t know whether there are any solutions -maybe trying to assign fibers in folding-less spaces such as inflated. I also saw MACT, but I couldn’t find any implementation of it inside MRtrix.