To introduce the functional anomaly map (FAM), a novel approach to mapping anatomical and functional brain lesions in individual patients. To demonstrate that FAMs can be used as the basis for lesion-symptom mapping analyses.

The lesion method was foundational to neurology, but a major drawback of lesion studies is an assumption that lesion identification based on the appearance of anatomical tissue completely captures the lesion extent and effects on the brain. This assumption is flawed because one cannot determine degree of tissue function or dysfunction based on anatomical appearance alone. More importantly, this assumption ignores effects from outside the anatomical lesion including diaschisis and compensatory neuroplasticity.

We derive functional anomaly maps (FAMs) using machine learning (voxelwise SVR) on resting state BOLD time-courses in fifty chronic left-hemisphere stroke survivors. We document the reliability of FAMs and validate their sensitivity. We illustrate an example application by performing lesion-symptom mapping on four behaviors commonly affected in left-hemisphere stroke using FAMs and a matching traditional anatomical dataset.

FAMs showed good replicability and were sensitive to anatomical lesion location. Signal in the unlesioned hemisphere related to functional integrity of homotopic locations in the lesioned hemisphere. FAM-based lesion-deficit maps replicated classical patterns of behavioral localization, and also detected relationships between functional integrity and behavior in regions distant from the anatomical lesions.

FAMs identify anatomical lesions and functionally anomalous areas beyond lesion boundaries, providing a direct measure of functional differences between individuals potentially related to diaschisis, compensatory plasticity, or individual differences that explain stroke resilience. Lesion-behavior mapping using these maps replicates classic behavioral localization within the lesion distribution and can identify relationships between spared tissue function and behavioral outcomes. This method provides the first direct brain-wide mapping of functional lesions, which could have wide implications for one of the most important methods in neuroscience.