Description: Is it left or is it right? A machine learning framework for studying hemispheric differences

This title appears in the Scientific Report : 2022

Is it left or is it right? A machine learning framework for studying hemispheric differences

The comparison between regions or tracts in the left and right hemispheres grants insight into localasymmetries as one characteristic feature of brain organization. These pairwise comparisons,however, are reminiscent of a localists’ view and come with several caveats when assessing multipleasymmetri...

Personal Name(s):	Friedrich, Patrick
	Patil, Kaustubh / Mochalski, Lisa / Li, Xuan / Camilleri, Julia / Kröll, Jean-Philippe / Wiersch, Lisa / Vickery, Sam / Hopkins, William D. / Hoffstaedter, Felix / Eickhoff, Simon / Weis, Susanne
Contributing Institute:	Gehirn & Verhalten; INM-7
Imprint:	2022
Conference:	8th North Sea Laterality Conference on Brain Asymmetry, Bergen (Norway), 2022-08-24 - 2022-08-27
Document Type:	Conference Presentation
Research Program:	Multilevel Brain Organization and Variability
	Publikationsportal JuSER

The comparison between regions or tracts in the left and right hemispheres grants insight into localasymmetries as one characteristic feature of brain organization. These pairwise comparisons,however, are reminiscent of a localists’ view and come with several caveats when assessing multipleasymmetries. To overcome some of the limitations set by conventional statistical comparisons, werecently introduced a novel machine learning-based framework for studying hemispheric differences.In a recent proof-of-principle study, we showed the capability of machine learning-based classificationto distinguish the hemispheres based on voxel-wise features. Using a Boruta feature-selectionalgorithm allowed the mapping of voxels that were important for correctly classifying a givenhemisphere. Furthermore, relating these maps of hemisphere-determining voxels with volumetricasymmetries validated our approach for mapping lateralized brain structure. In this talk, I will highlightour framework for studying hemispheric differences and present ongoing work on possibleapplications.