Hi Knut Jørgen - The isosurfaces are created after thresholding a probability distribution. This means that even a few voxels with very low probability, which would have very little effect on the stats, can have a big effect on the "look" of the isosurface if they're included, making it look much rougher/noisier. In that sense the look of the isosurface can be tricky to evaluate. It can help to look at the heat map of the probability distribution in the slice views in deciding if the tracts are really noisy or not.

Best,
Anastasia.

From: freesurfer-bounces@nmr.mgh.harvard.edu [freesurfer-bounces@nmr.mgh.harvard.edu] on behalf of Knut J Bjuland [knutjorgen@outlook.com]
Sent: Monday, November 21, 2016 11:28 AM
To: Freesurfer support list
Subject: [Freesurfer] quality control of tracula tractography

Hi,

We have done tracula tractography on 180 subjects, and most of the isosurfaces look a lot noisier than the example on the Tracula wiki (see fminor in attached figure). Is there any way to gauge the noisiness of a reconstructed path? And can that information be used for quality control? For instance, here (https://neuro.ebioscience.amc.nl/portal/web/nsg/tracula) they do not accept a noisy fminor (not unlike ours).

Another source of information on outliers is the log files that are produced when running -trac-all -stat (based on the shape of the tract). It doesn't seem to flag tracts that are (almost) completely missing (see attached figure, which was not flagged in lh.cab_PP.avg33_mni_bbr.log). Do they fall under the minimum threshold and are therefore not evaluated? Is it recommended to automatically exclude the subjects that are flagged as outliers here?

Thank you!

Knut Jørgen Bjuland, PHD

Sørlandet sykehus