Objective: During deep brain stimulation (DBS) surgery for the treatment of advanced Parkinson's disease (PD), microelectrode recording (MER) in conjunction with functional stimulation techniques are commonly applied for accurate electrode implantation. However, the development of automatic methods for clinical decision making has to date been characterized by the absence of a robust single-biomarker approach. Moreover, it has only been restricted to the framework of MER without encompassing intraoperative macrostimulation. Here, we propose an integrated series of novel single-biomarker approaches applicable to the entire electrophysiological procedure by means of a stochastic dynamical model.
Approach: The methods are applied to MER data pertinent to ten DBS procedures. Considering the presence of measurement noise, we initially employ a multivariate phase synchronization index for automatic delineation of the functional boundaries of the subthalamic nucleus (STN) and determination of the acceptable MER trajectories. By introducing the index into a nonlinear stochastic model, appropriately fitted to pre-selected MERs, we simulate the neuronal response to periodic stimuli (130 Hz), and examine the Lyapunov exponent as an indirect indicator of the clinical effectiveness yielded by stimulation at the corresponding sites.
Main results: Compared with the gold-standard dataset of annotations made intraoperatively by clinical experts, the STN detection methodology demonstrates a false negative rate of 4.8% and a false positive rate of 0%, across all trajectories. Site eligibility for implantation of the DBS electrode, as implicitly determined through the Lyapunov exponent of the proposed stochastic model, displays a sensitivity of 71.43%.
Significance: The suggested comprehensive method exhibits remarkable performance in automatically determining both the acceptable MER trajectories and the optimal stimulation sites, thereby having the potential to accelerate precise target finalization during DBS surgery for PD.