# A novel methodology for localizing pallidal deep brain stimulation leads

## TL;DR

A new software method improves the accuracy of deep brain stimulation lead placement in Parkinson's disease by using patient-specific anatomical data.

## Contribution

A novel pipeline in Unity™ enables semi-automated, reproducible localization of DBS leads in the globus pallidus internus.

## Key findings

- Using GPi-specific coordinates reduced the volume of active contact locations compared to traditional MCP coordinates.
- The mean distance to the centroid was significantly smaller with GPi-specific coordinates.
- The method allows linking lead positions to clinical outcomes like motor improvement.

## Abstract

Positioning of deep brain stimulation (DBS) leads is paramount for optimizing therapeutic efficacy in Parkinson’s disease (PD) and dystonia. Quantitative determination of lead position remains essential; however, current atlas-based targeting and stereotactic coordinate methods, while informative, limit patient specificity afforded by high-resolution reconstruction and introduce subjective variability.

We developed a novel pipeline in Unity™ to ingest 7T MRI-based reconstructions of DBS leads within the globus pallidus internus (GPi). Using anatomical landmarks and structure-specific algorithms, the GPi was parcellated into 12 anatomically-based subregions in a semi-automated, reproducible manner. Active contact positions were localized relative to a novel coordinate system developed from a GPi-based bounding box. A novel distance-to-border metric remapped contacts onto a common atlas (PD25) for population comparison against the same contacts mapped onto a common left GPi space using mid-commissural point (MCP) coordinates (Schaltenbrand-Wahren atlas).

Fifteen leads from 10 PD subjects were used for ellipsoid fitting of active contact locations, resulting in an elliptical volume of 38.94 mm3 when using MCP coordinates, compared to a volume of 5.08 mm3 with our GPi-specific coordinates. The mean distance-to-ellipse centroid was 3.45 ± 1.57 mm for MCP coordinates and 2.03 ± 0.82 for our GPi-specific coordinates. Our distance-to-border remapping metric yielded mean adjustments of 0.81 mm (y-axis) and 1.61 mm (z-axis). A subset of six GPi active contacts were plotted with post-DBS motor improvement scores, demonstrating the ability to link lead location with clinical outcomes.

Our novel software provides a quantifiable lead location with respect to the anatomical target, enhancing patient-specific lead localization by avoiding some of the pitfalls of either structure-to-atlas normalization or traditional stereotactic coordinates.