I hope my email finds you well. I am running into a major issue when I run recon-all -all on the OASIS1 cohort. I am getting an inconsistent number of ROIs for the same subject between the lh and rh, resulting in different sizes of connectivity matrices across subjects (various numbers of ROIs {27, 28, 29, 30, 31,32}. For all subjects from other cohorts, patients and controls, it was 31 ROIs). Please find below a quick example, subject OAS1_0384 is fine, whereas OAS1_0167 and OAS1_0159 are not.

I tried inspecting which ROIs usually cause this difference between the two hemispheres, it differs from subject to subject, but here is a comprehensive list of all the problematic ROIs that came up. Please not this is a cohort of healthy controls this issue is coming up for. I ran this on a set of patients with Multiple Sclerosis, and I got a consistent number of 31 ROIs.

{'insula', 'parahippocampal', 'caudalmiddlefrontal', 'parsorbitalis', 'paracentral', 'transversetemporal',

'parsopercularis', 'superiortemporal', 'postcentral', 'isthmuscingulate', 'inferiortemporal', 'superiorparietal',

 'parstriangularis', 'middletemporal', 'rostralanteriorcingulate', 'lateraloccipital', 'entorhinal', 'posteriorcingulate',

'lingual', 'pericalcarine', 'inferiorparietal', 'precentral', 'cuneus', 'fusiform', 'lateralorbitofrontal',

'caudalanteriorcingulate', 'superiorfrontal', 'medialorbitofrontal', 'precuneus', 'supramarginal', 'temporalpole',

'rostralmiddlefrontal'}

I attached a few examples. Any idea what this might be due to? This is quite a big issue for our pipeline so any help would be appreciated. Thanks, and looking forward to your response.

Hajer Karoui

Data Manager & Researcher | MS & PD Tissue Bank

Imperial College London

Department of Brain Sciences

LinkedIn