Mert Rory Sabuncu

This study uses naturalistic movie-watching fMRI to predict cognitive scores and sex through a deep neural network framework, showing improved performance over resting-state FC.

The paper introduces Krakencoder, a tool that translates and fuses structural and functional brain connectomes across atlases and processing strategies, achieving higher identifiability than existing models.

This paper introduces BrainSurfCNN, a surface-based CNN that predicts task-based contrast maps from resting-state fMRI, achieving state-of-the-art accuracy and generalization.

The paper quantifies regional structural-functional connectome coupling in adults and finds it varies with age, sex, cognitive scores, and is highly heritable.

This study shows deep neural networks can predict cortical responses to naturalistic stimuli across large brain areas beyond visual and auditory regions.

This study investigates the heritability of individualized cortical network topography using twin and family data, finding genetic influences on network size and spatial organization.

This study identifies sex differences in brain regions using temporal dependence and volume measures from resting-state fMRI data, achieving high classification accuracy.

This study compares deep neural networks and kernel regression in predicting behavior and demographics from resting-state functional connectivity data, finding comparable performance with kernel regression being more computationally efficient.

This paper identifies three atrophy-cognitive factors in Alzheimer’s disease using a hierarchical Bayesian model, showing they are stable across cohorts and associated with distinct Tau deposition patterns.

This study finds that global signal regression enhances the association between resting-state functional connectivity and behavioral measures in young healthy adults.

This study uses machine learning to predict motor improvement in chronic stroke patients after therapy, finding that pre-intervention assessments and motor thresholds are key predictors.

This study identifies shared genetic factors between educational attainment and cerebral cortical morphology using UK Biobank data and novel statistical methods.

This study introduces a hierarchical Bayesian model to estimate individual-specific cortical networks and shows that their topography can predict human behavior with moderate accuracy.

The paper shows that joint analysis of cortical area and thickness is more sensitive than volume for studying brain anatomy, and proposes a permutation-based method to replace volume analysis.