Talk from
Emma Robinson – King's College London
Short abstract:
The human cortex varies significantly across individuals in ways that confound the detection of reliable biomarkers predictive of cognition or disease. In our lab we have been trying to resolve this through both advancing classical approaches - based on surface registration - as well as through developing learning-based frameworks for modelling static and dynamic signals directly on cortical surface manifolds. In this talk, I will focus on recent work in which we have been seeking to characterize the scale of cortical shape variability, then model where it comes from and disentangle it from phenotypically relevant sources of variability. I'll also discuss recent work in which we used surface vision transformers to train fMRI decoding models that generalize across brains.