search

sotnir / NENAH-BIDS

A collaborate platform for streamlining a BIDS workflow for the NENAH study, on the linux workstation based at University of Southampton.
https://www.southampton.ac.uk/cndng
Creative Commons Zero v1.0 Universal
2 stars 2 forks source link

Misclassification of WM hypointensities in FS to 5TT pathological tissue? #161

Open RioPhillips opened 5 months ago

RioPhillips commented 5 months ago

Using 5ttgen to generate 5TT image from FS segmentation in https://github.com/sotnir/NENAH-BIDS/blob/1a876927a97560a0e1922c6178cb0f3caca49e41/dMRI/5tt.sh#L84-L88

This is the result for NENAH002 which shows misclassification around ventricles with 5TT = pathological tissue. image

When checking in FS segmentation, it corresponds to 77 (=WM-hypointensities) in FS aparc+aseg image

RioPhillips commented 5 months ago

In 5ttgen, the conversion of WM hypointensities is to pathological tissue image So this can be manually changed as a solution

RioPhillips commented 5 months ago

However, when looking at the FOD (estimated with dhollander, i.e. not using ACT information) then this area describes coherent WM.

In this image with FODs and an overlaid 5TTvis-image image

So, should be kept.

RioPhillips commented 5 months ago

But when using ACT in tckgen, we have streamlines going through here... image