Muthukumar Pandaram: Optimisation of the parameter space for deep brain stimulation fiber filtering across neuropsychiatric disorders

BCCN Berlin / Technische Universität Berlin

Abstract

This thesis explores Deep Brain Stimulation (DBS), a surgical intervention for neurological and psychiatric conditions such as Parkinson’s Disease, Dystonia, Alzheimer’s Disease and Obsessive Compulsive Disorder. DBS involves implanting electrodes in specific brain regions connected to an impulse generator, modulating brain activity to alleviate symptoms. Within the DBS research domain, the Lead-DBS platform provides essential tools, including the Fiber Filtering tool, which evaluates DBS’s connectomic impact by identifying fiber tracts linked to clinical outcomes. However, the current version of the Fiber Filtering tool includes multiple settings that require meticulous tuning for optimal predictions. The primary objective of the thesis is to identify the optimal set of parameters or settings within the Fiber Filtering toolbox, generating the most effective model for predicting fibers with clinical improvement. This investigation utilizes datasets from patients with Parkinson’s disease (PD), Alzheimer’s disease (AD), and Obsessive Compulsive Disorder (OCD). Optimal settings are determined using a surrogate optimizer, analyzing datasets across the disorders both individually and collectively. The results demonstrate that a common set of settings yields the most effective models for PD and OCD, while optimizing parameters separately for each cohort produces the best settings for the AD cohort.

 

Additional Information

Master Thesis Defense

 

Organized by

Prof. Dr. Andreas Horn   & Prof. Dr. Stefan Haufe