Mapping Functional Language Areas with Non-Functional Brain MRI

Abstract

Mapping eloquent brain areas has become a standard of care in brain surgery. Current imaging-based techniques usually rely on functional MRI (fMRI), which measures neural activity via the blood oxygenation level-dependent signal. fMRI protocols are time-intensive, require active patient collaboration, and involve laborious manual post-processing and expertise, making them difficult to implement in some clinical scenarios. In this research, we propose a fully automated deep neural pipeline for the mapping of Broca and Wernicke functional language areas using multiple non-functional MRI modalities. The proposed method is evaluated on a cohort of 30 drug-resistant epilepsy patients, showing encouraging qualitative and quantitative results and suggesting its potential applicability as an effective and practical tool for neurosurgical planning and navigation. Implementation details can be found in our GitHub.