Localization of Deep Brain Stimulation Contacts Using Corticospinal/Corticobulbar Tracts Stimulation.

Background: Successful deep brain stimulation (DBS) in Parkinson's disease (PD) requires optimal electrode placement. One technique of intraoperative electrode testing is determination of stimulation thresholds inducing corticospinal/corticobulbar tracts (CSBT) motor contractions. Objective: This study aims to analyze correlations between DBS electrode distance to CSBT and contraction thresholds, with either visual or electromyography (EMG) detection, to establish an intraoperative tool devoted to ensure safe distance of the electrode to the CSBT. Methods: Twelve PD patients with subthalamic nucleus DBS participated. Thresholds of muscular contractions were assessed clinically and with EMG, for three different sets of stimulation parameters, all monopolar: 130 Hz high-frequency stimulation (HFS); 2 Hz low-frequency stimulation with either 60 or 210 mu s (LFS-60, LFS-210). The anatomical distance of electrode contacts to CSBT was measured from fused CT-MRI. Results: The best linear correlation was found for thresholds of visually detected contractions with HFS (r(2) = 0.63, p < 0.0001) when estimated stimulation currents rather than voltages were used. This correlation was found in agreement with an accepted model of electrical spatial extent of activation (r(2) = 0.50). When using LFS, the correlation found remained lower than for HFS but increased when EMG was used. Indeed, the detection of contraction thresholds with EMG versus visual inspection did allow more frequent detection of face contractions, contributing to improve that correlation. Conclusion: The correlation between electrode distance to the CSBT and contraction thresholds was found better when estimated with currents rather than voltage, eliminating the variance due to electrode impedance. Using LFS did not improve the precision of that evaluation, but EMG did. This technique provides a prediction band to ensure minimum distance of the electrode contacts to the CSBT, integrating the variance that can be encountered between prediction of models and practice.