CRCNS: A computational approach to map visual cortex organization in the human brain

PROJECT SUMMARY About a quarter of the human cerebral cortex is visual, with dozens of retinotopic maps and category selective regions. Localizing these areas is crucial to much of human visual neuroscience research and to clinical care. Area localization, such as delineating the boundaries of primary visual cortex, is a necessary step in studies that measure neural activity or anatomical properties because knowledge can only be aggregated across labs, groups, studies, and the lifespan if researchers can identify the same visual areas in each person. Moreover, brain area boundaries often serve as dependent measures themselves, for instance in studies that track the size of face or word-selective areas during development, or that link the surface area of a visual map to performance–e.g., V1 and acuity or hV4 and recognition in clutter. However, while dozens of functional visual areas have been identified on cortex, many are technically difficult to localize, and few have been carefully characterized across large populations of individuals. The field lacks validated computational methods to parcellate visual cortex into its component retinotopic maps and category-selective regions at the level of individual subjects, and it lacks a quantitative description of how anatomical map properties (size, tissue thickness, myelination, white matter inputs, etc) vary across large populations. The central aim of this proposal is to quantify how the components of the cortical visual system vary (and covary) in large populations of human subjects. We take two complementary approaches. One is to develop machine learning (ML) algorithms to delineate visual areas using various combinations of MRI inputs. The other is to establish descriptive models of the distribution, heritability, and patterns of co-variation of properties of visual cortical areas and their associated white matter tracts. Currently, accurately identifying visual system components in individuals remains challenging, hampering both scientific and clinical progress. The results of this project will greatly increase the ability of basic scientists and clinicians to accurately identify visual areas in individual humans and to quantify how individual variability impacts vision. This in turn furthers the neuroscientific understanding of vision and empowers translational research into the cortical origins of human visual disorders.