Keynotes | K&Ts | GACs | Talks | Posters | Search

Poster D in Poster Session D: Wednesday, August 5, 2:00 – 3:45 pm, Kimmel Center, Shorin & Rosenthal Rooms

Automated Individual Subject Functional Parcellation of the Reading Circuitry in the Ventral Occipitotemporal Cortex

Álvaro Sánchez-Paniagua Ríos1, Yongning Lei1, Siwen Wang1, Bharath Kumar Santhi Rajan2, Pedro M. Paz-Alonso1,3, Garikoitz Lerma-Usabiaga1; 1Basque Center on Cognition, Brain and Language, 2Universidad del País Vasco, 3IKERBASQUE, Basque Foundation for Science

Presenter: Garikoitz Lerma-Usabiaga

Mapping the computational architecture of reading requires precise, subject-specific localization within the ventral occipitotemporal cortex (vOTC). However, establishing a reliable topography is currently hindered by a lack of standardized delineation criteria, the use of disparate functional localizers across different laboratories, considerable individual differences in cortical anatomy, inter-session variability, signal dropout, and group-level averaging that blurs functional boundaries. To overcome these limitations, we operate on the premise that despite the variability, functional organization exhibits fundamental invariabilities anchored to structural geometry (e.g. V2 always posterior to V1). Leveraging dense fMRI sampling (10 subjects; 100 sessions), we introduce an automated framework to delineate functional regions of interest (ROIs) in individual space. Our approach extracts candidate clusters on the native cortical surface and establishes cross-subject correspondence using the Fused Gromov-Wasserstein (FGW) distance, jointly aligning geometry and functional profiles. By assembling these couplings into a global graph and applying community detection, we isolate maximally consistent subregions. Applied to the vOTC, our method reveals distinct, reproducible functional communities organized along the posterior-to-anterior axis, conserving spatial relationships despite individual anatomical differences. Bypassing rigid template registration, this framework provides a robust, mathematically grounded solution for subject-specific mapping of human functional architecture.

Topic Area: Methods, Tools, Theory & Neural Coding