If you run
mri_segstats --qa-stats subject subject.qa.dat
It will generate the QA .dat file. This will have 20 numbers in it. Run mri_segstats --help to see what they are. I usually use #20 (CNR)
Run it for all your subjects, then sort by CNR, then examine the worse subjects

On 10/4/2022 10:24 AM, Blujus, Jenna wrote:

External Email - Use Caution

Hi all,

I am processing a large longitudinal dataset of T1-weighted anatomical images through the recon-all pipeline (N=680). I know that the gold standard for quality control following recon-all is visual inspection of the output surfaces and parcellations. However, given the large sample size, I was writing to see if the community has any recommendations for automatic QC pipelines.

I read a few papers discussing the use of Euler numbers to detect outlier datasets for exclusion (see citations below). Does anyone have any other recommendations?

Thank you in advance!

Best,

Jenna

References:

1. Rosen, A. F., Roalf, D. R., Ruparel, K., Blake, J., Seelaus, K., Villa, L. P., ... & Satterthwaite, T. D. (2018). Quantitative assessment of structural image quality. Neuroimage, 169, 407-418.

2. de Lange, A. M. G., Kaufmann, T., Quintana, D. S., Winterton, A., Andreassen, O. A., Westlye, L. T., & Ebmeier, K. P. (2021). Prominent health problems, socioeconomic deprivation, and higher brain age in lonely and isolated individuals: A population-based study. Behavioural Brain Research, 414, 113510.
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer