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

Area of Science:

• Neuroscience
• Computational Neuroscience
• Medical Imaging

Background:

• Topographic organization is key to sensory processing in the brain.
• Current methods for mapping visual cortex are often resource-intensive or lack individual precision.
• Population atlases have limitations for individual-level inference.

Purpose of the Study:

• To develop a deep learning tool for predicting functional topographic organization of the human visual cortex using only anatomical data.
• To enable accurate, individual-specific retinotopic mapping.
• To overcome limitations of existing neuroimaging and atlas-based approaches.

Main Methods:

• Introduction of the 'deepRetinotopy toolbox', a deep learning application.
• Prediction of retinotopic maps from cortical anatomy.
• Validation across diverse experimental conditions, imaging sites, and scanner types.

Main Results:

• The deepRetinotopy toolbox accurately predicts retinotopic maps.
• Predicted maps facilitate automatic generation of individual-specific visual area boundaries, reducing manual annotation bias.
• Application to 11,060 scans quantified age-related changes in visual cortex organization.

Conclusions:

• DeepRetinotopy toolbox provides a scalable, anatomy-based method for functional brain mapping.
• The tool enables precise, individual-specific retinotopic mapping without functional neuroimaging.
• This approach has broad utility for large-scale studies of brain organization and aging.