(22) Comparison of Structural Connectomes for Modeling Deep Brain Stimulation Pathway Activation

Introduction: Structural connectivity models of the brain are commonly employed to identify pathways that are directly activated during deep brain stimulation (DBS). However, various connectomes differ in the technical parameters, parcellation schemes, and methodological approaches used in their construction. As such, it is unclear if the inferred DBS connectivity estimated from different connectomes is consistent across analyses.

Methods: The goal of this project was to compare DBS pathway activation predictions when using different structural connectomes, while using identical electrode placements and stimulation volumes in the brain. Our analysis focused on four popular structural connectomes: 1) Horn normative connectome (whole brain), 2) Yeh population-averaged tract-to-region pathway atlas (whole brain), 3) Petersen histology-based pathway atlas (subthalamic focused), and 4) Majtanik histology-based pathway atlas (anterior thalamus focused). DBS simulations were performed to generate pathway recruitment curves for each connectome, at three generalized locations for DBS electrode placement: 1) subthalamic nucleus (STN), 2) anterior nucleus of thalamus (ANT), and 3) ventral capsule (VC).

Results: The choice of connectome used in the DBS simulations resulted in notably distinct pathway activation predictions, and little congruence in the predicted patterns of brain network connectivity. The Horn connectome exhibited wide-ranging streamline activation at all DBS locations, but lacks both pathway annotation and anatomical validity. The Yeh connectome provided estimates of DBS connectivity for any stimulation location in the brain with anatomically annotated pathways, but lacks finer anatomical details. The Petersen and Majtanik histology-based connectomes are likely the most anatomically realistic, but are only applicable to specific DBS targets because of their limited representation of pathways.

Conclusion: While each connectome has unique strengths and weaknesses, they yield widely-varying and inconsistent inferences of DBS network connectivity. This raises substantial concern regarding the general reliability of connectomic DBS studies, especially those that lack anatomical and/or electrophysiological validation in their analyses.