Predicting functional topography of the human visual cortex from cortical anatomy at scale

Topographic organization is a fundamental principle of the brain, whereby adjacent cortical locations represent adjacent features in sensory or cognitive space. Current methodologies for obtaining individual-specific topographic maps either require resource-intensive functional neuroimaging data or, when using population atlases, lack precision for individual-level inference, limiting large-scale studies of individual differences. Here, we developed deepRetinotopy toolbox , a deep learning-based application for predicting topographic organization of the human visual cortex from cortical anatomy. We show that our toolbox accurately predicts retinotopic organization using only standard structural MRI scans acquired under diverse conditions, including varying imaging sites and scanner types. We then demonstrate that predicted maps can be leveraged to automatically generate individual-specific visual area boundaries, overcoming common biases in manual annotations. Finally, we estimate age-related changes in visual cortex organization by retrospectively applying our method to 11,060 anatomical scans, revealing that group differences in primary visual cortex organization are predictable from cortical anatomy.