Objective: This study aims to evaluate different strategies to map DBS‑induced side‑effects regarding their ability to create anatomically plausible stimulation maps and to predict side‑effect occurrence in out‑of‑sample data.

Background: Voxel‑wise mapping of DBS effects has been used in a variety of research. While beneficial DBS effects can be expected to originate close to a well-placed DBS lead, side‑effects might often occur due to stimulation spread into adjacent structures. Consecutively, it can be assumed that only a small part of the stimulation – the part that spreads into adjacent structures – actually causes a side‑effect, which makes side‑effects difficult to assess in voxel-wise DBS mapping.

Method: In this project, we explore different approaches to the voxel‑wise mapping of DBS‑induced side‑effects in a large retrospective dataset of monopolar reviews in subthalamic nucleus DBS. The training dataset consists of data from n=30 patients, while data from n=10 patients was used as the validation dataset. The approaches consist of classic VTA‑based stimulation mapping, “shell-mapping”, and “safezone-mapping”. Using models based on receiver operating characteristics (ROC) we also evaluate how well the maps generated by these different approaches can predict the occurrence of different side‑effects in out‑of‑sample data.

Results: Using close to 4000 stimulation settings associated with either the occurrence of DBS‑induced side‑effects or no side‑effects, we demonstrate that different mapping approaches lead to differences in the resulting maps – some more anatomically plausible than others. While all the resulting maps can be used to predict the occurrence of side‑effects to a certain degree, the predictive capabilities struggle to outperform stimulation amplitude as the most important predictive factor for side‑effect occurrence.

Conclusion: While stimulation amplitude is the most important predictor for the occurrence of DBS‑induced side‑effects, voxel‑wise mapping of side‑effects – if done correctly – can help in predicting which type of side‑effect will occur. In a next step, connectomic data and fiber‑wise mapping of DBS effects should be incorporated to possibly increase predictive capabilities.

« Back to 2022 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/mapping-side-effects-in-deep-brain-stimulation-different-approaches-different-problems/