INTERFACING ARTIFICIAL INTELLIGENCE AND MULTIMODAL BRAIN IMAGING: A NOVEL METHODOLOGY FOR INVESTIGATING TRIGEMINAL NEURALGIA

Abstract

Chronic pain, a critical condition for global health, affects millions and imposes significant economic and social burdens. Despite advancements in neuroimaging, the correlation between central nervous system (CNS) changes and the behavioral and functional consequences of chronic pain remains inadequately understood. This thesis aims to bridge this gap, focusing on Trigeminal Neuralgia (TN), a debilitating form of chronic neuropathic pain. By interfacing artificial intelligence (AI) with magnetic resonance imaging (MRI) and clinical data, this research seeks to unravel the complex CNS changes associated with TN and predict treatment outcomes, enhancing our understanding and management of TN and chronic painin general. In Study 1, machine learning algorithms, combined with gray matter metrics from T1-weighted MR imaging and white matter metrics from Diffusion Tensor Imaging (DTI), were employed to distinguish Classical Trigeminal Neuralgia and Trigeminal Neuropathic Pain from healthy controls. This approach achieved high accuracy, demonstrating the efficacy of AI in derivingimaging signatures of TN pain. Study 2 focused on developing a stratification schema for Classical Trigeminal Neuralgia surgical outcomes using clinical and imaging data processed through machine learning algorithms. This approach allowed us to accurately predict surgical outcomes based on pre-surgical data, suggesting a potential tool for personalized treatment strategies. Study 3 specifically addressed Trigeminal Neuralgia secondary to Multiple Sclerosis (MSTN). By analyzing imaging data from MSTN patients and comparing it with pain-free Multiple Sclerosis subjects, distinct MR imaging signatures of pain were identified, enabling the accurate distinction between the pain-afflicted and pain-free MS cohorts. In summary, this thesis represents a pioneering convergence of neuroimaging and AI in the field of chronic trigeminal pain. Our findings illuminate the intricate CNS abnormalities linked to the experiences of individuals with TN. This research contributes significantly to a more nuanced understanding of chronic pain and its management using TN as a model condition.