Comparison of thalamic atlases and segmentation techniques in defining motor and sensory nuclei for deep brain stimulation targeting in essential tremor

Introduction

Many thalamic atlases are available for deep brain stimulation (DBS) applications, but their usage and clinical validation vary. This study investigated the effectiveness of six atlases in DBS targeting for essential tremor (ET) through structural differences, impact of two thalamic segmentation methods, and correspondence with clinical outcomes via individualized tissue activation modeling.

Methods

22 ET patients with unilateral VIM DBS were retrospectively analyzed. Volume of tissue activation (VTA) models were linked to tremor reduction (n = 22) and sustained paresthesia (n = 32). Six atlases were co-registered for pairwise comparison. Patient thalami were segmented using atlas-based segmentation (ABS) and diffusion tensor imaging-based segmentation (DTIBS). Geometric properties and VTA overlap with motor and sensory regions were statistically assessed.

Results

Atlas comparisons revealed significant differences in motor and sensory subnuclei delineation (p < 0.001). ABS generally produced larger motor and smaller sensory volumes than DTIBS, with both showing significant geometric variability. For therapeutic VTAs, ABS consistently showed greater motor activation across all atlases, while DTIBS demonstrated this in only half. Sensory activation was more often greater for paresthesia than therapeutic VTAs using ABS. The Jakab atlas showed the strongest correspondence with clinical outcomes using both segmentation methods.

Conclusion

Atlas choice and segmentation method can potentially influence DBS targeting accuracy. A segmentation approach that performs well with one atlas may not generalize to others, underscoring the need for clinical validation to evaluate applicability in surgical planning. DTIBS may enable more individualized targeting, but requires further refinement to consistently outperform traditional methods.