Leonardo Bonilha

The paper ablates attention heads in vision-language models to study picture naming accuracy, finding that head removal disrupts naming but does not confirm the head computes the behavior.

This study shows white matter signals during seizures reflect communication between brain regions and differentiate epilepsy pathophysiologies, challenging past assumptions about their limited value.

The paper examines associations between epilepsy-related polygenic risk scores and brain morphology in children, finding PRS-related cortical thinning linked to structural patterns in TLE-HS patients.

This study investigates the neural separation of verbal working memory and syntactic comprehension, finding distinct brain networks associated with each function.

This study examines age-related changes in grey and white matter in TLE patients across the adult lifespan, finding accelerated aging effects after age 55.

This study uses resting-state fMRI and graph theory to examine how disruptions in the dual-stream speech processing model relate to aphasia impairments in stroke patients.

The paper uses connectome-based lesion-symptom mapping to investigate word and sentence comprehension deficits in stroke patients, finding that middle-posterior temporal lobe regions, not double disconnection, are primarily responsible.

The paper maps structural white matter disconnections in apraxia by analyzing diffusion tensor imaging data from stroke patients, identifying key fronto-temporo-parietal and subcortical networks involved.

This study identifies divergent patterns of atypical cortical asymmetry and regional atrophy in temporal lobe epilepsy using MRI data from a large multi-site cohort.

This study links structural network changes in epilepsy syndromes to risk gene expression patterns using ENIGMA data.

This study identifies white matter microstructural differences across epilepsy syndromes using diffusion MRI data from a large multicenter cohort.

This study identifies network associations with atrophy patterns in epilepsy using neuroimaging and connectome analysis across 1,021 patients and 1,564 controls.

This study examines word error patterns in connected speech and object naming after left hemisphere stroke, finding distinct neural substrates for different types of paraphasias.