Robust lesion network mapping reveals genuine symptom-specific networks.

Lesion network mapping (LNM) and its derivatives successfully integrate anatomically distributed brain loci into common symptom-associated functional networks, however, their statistical validity and specificity have recently become topics of debate. Here, we introduce a null model-based robust LNM (rLNM) framework to perform sensitivity and symptom-specificity testing. Across multiple lesion-based (10 conditions, 333 lesions) and task-based (4 conditions, 706 experiments) datasets, rLNM reveals biologically meaningful and symptom-specific networks while effectively controlling for false positives.