Lesion analysis plays a central role in cognitive neuroscience, complementing functional activation studies by identifying brain regions that are necessary for the performance of particular functions. Most existing methods for lesion analysis are inherently univariate, considering the relationship between injury and behavior independently between locations in the brain. We propose to develop and adapt methods for detecting multivariate relationships between brain injury and behavior. These methods will be applied to resolve puzzling results from an existing dataset of aphasic stroke patients, and will be provided to the community in the form of user-friendly cross-platform software.
We propose to develop advanced methods and software for the analysis of brain-behavior relationships in stroke, with application to an existing dataset of patients with language impairments.