StimFit—A Data‐Driven Algorithm for Automated Deep Brain Stimulation Programming

Abstract

BackgroundFinding the optimal deep brain stimulation (DBS) parameters from a multitude of possible combinations by trial and error is time consuming and requires highly trained medical personnel.ObjectiveWe developed an automated algorithm to identify optimal stimulation settings in Parkinson's disease (PD) patients treated with subthalamic nucleus (STN) DBS based on imaging‐derived metrics.MethodsElectrode locations and monopolar review data of 612 stimulation settings acquired from 31 PD patients were used to train a predictive model for therapeutic and adverse stimulation effects. Model performance was then evaluated within the training cohort using cross‐validation and on an independent cohort of 19 patients. We inverted the model by applying a brute‐force approach to determine the optimal stimulation sites in the target region. Finally, an optimization algorithm was established to identify optimal stimulation parameters. Suggested stimulation parameters were compared to the ones applied in clinical practice.ResultsPredicted motor outcome correlated with observed outcome (R = 0.57, P < 10−10) across patients within the training cohort. In the test cohort, the model explained 28% of the variance in motor outcome differences between settings. The stimulation site for maximum motor improvement was located at the dorsolateral border of the STN. When compared to two empirical settings, model‐based suggestions more closely matched the setting with superior motor improvement.ConclusionWe developed and validated a data‐driven model that can suggest stimulation parameters leading to optimal motor improvement while minimizing the risk of stimulation‐induced side effects. This approach might provide guidance for DBS programming in the future. © 2021 The Authors. Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society

Towards biomarker-based optimization of deep brain stimulation in Parkinson’s disease patients

“…Other emerging tools to guide programming are fMRI combined with machine learning ( Phibbs et al, 2014 ) and biophysical model-based programming ( Howell et al, 2021 ). Furthermore, a recent study ( Roediger et al, 2022 ) even developed a fully automated algorithm to help guide the programming of individual patients, termed “Stimfit.” Adding electrophysiological data to image-derived information could improve the programming of individual patients in a complementary manner to optimize DBS programming. Also, image-based programming with commercial software has also proven an effective programming tool that leads to non-inferior motor symptom control compared to standard programming ( Ewert et al, 2018 ).…”

Local Field Potentials Predict Motor Performance in Deep Brain Stimulation for Parkinson's Disease

“…Directional adaptive stimulation could mitigate this limitation by increasing the number of available contacts, for example, in patients with suboptimal lead placement. Second, previous studies introduced automated programming algorithms that predict stimulation settings based on electrode positioning 9,41‐44 . These imaging‐based programming algorithms may be further improved by the integration of easily acquirable electrophysiological data 33 .…”

“…Second, previous studies introduced automated programming algorithms that predict stimulation settings based on electrode positioning. 9,[41][42][43][44] These imagingbased programming algorithms may be further improved by the integration of easily acquirable electrophysiological data. 33 This might especially hold true for longer follow-up periods, because changes at the electrode tissue interface, 45,46 impedance changes, 35 alterations of the spatial distribution and extent of oscillatory activity 47 or changes in clinical symptoms might warrant adaptation of stimulation settings.…”

Programming of subthalamic nucleus deep brain stimulation for Parkinson’s disease with sweet spot-guided parameter suggestions

“…The algorithm presented here is only capable of suggesting monopolar stimulation with one contact as the cathode and the implanted pulse generator as the return. A few in-silico studies have proposed algorithms to calculate multicathodic or multipolar suggestions and may be added to our algorithm in the future ( Anderson et al, 2018 ; Pena et al, 2018 ; Cubo et al, 2019 ; Connolly et al, 2021 ; Roediger et al, 2022 ). Interestingly, Waldthaler et al (2021) generated multipolar suggestions manually with the software Guide XT but had the dorsolateral part of the subthalamic nucleus as the only target.…”