MRI-based quantitative AI approaches for vestibular schwannoma care

Abstract

The diagnosis and management of vestibular schwannoma (VS) rely on longitudinal imaging surveillance and precise tumor measurement. Conventional approaches imply manually measuring every patient, adding a huge workload to the entire healthcare system. This thesis investigates the application of artificial intelligence (AI) in head and neck MRI analysis for data-driven VS care. Based on several deep learning techniques, we developed quantitative approaches for VS segmentation, completion and restoration of missing MRI sequences, and prediction of tumor growth. Comprehensive validations were conducted in both technical and clinical settings.