External Email - Use Caution
Dear experts,
I recently performed recon-all on a set of subjects and the output (aparcstats2table and asegstats2table) will be part of a meta-analysis of other labs' cohorts. When assessing the asymmetry index direction (relative size of left vs right hemisphere across multiple rois), our dataset shows an inversion of the mean asymmetry index in several structures compared to the other cohorts.
For example, in our dataset, the lateral ventricle volume is on average larger in the right hemisphere compared to the left, while the opposite is true for all other cohorts. The same inversion of asymmetry is visible in other structures that in general are more asymmetrical.
In order to check this, I went through each T1 to make sure the orientation was correct and indeed they are.
Is there any possible way that there was a left-right inversion somewhere during the recon-all pipeline? I want to make sure I did everything correctly on our end and that the results are due to the biology and not a technical mistake!
I found this website on Left-right reversal, and it seems relevant to my issue, but I'm not fully understanding what this means (how would bias occur during segmentation?) - https://surfer.nmr.mgh.harvard.edu/fswiki/LeftRightReversal
Many thanks for your input, Ellen
---
Ellen Ji, PhD Postdoctoral Research Fellow Psychiatric University Hospital University of Zürich ellen.ji@bli.uzh.ch homanlab.github.io/ellen/