An automated navigation system for deep brain stimulator placement using hidden Markov models

Area of Science:

• Neurosurgery
• Computational Neuroscience
• Medical Technology

Background:

• Deep brain stimulator (DBS) placement currently relies on manual interpretation of image-based stereotaxy and intraoperative microelectrode recording (MER) data.
• Integrating MER data with anatomical information is a manual and time-consuming process.
• Hidden Markov Models (HMMs) are established algorithms with broad applications in signal processing and biological systems.

Purpose of the Study:

• To develop and evaluate an automated navigation system for subthalamic nucleus (STN) deep brain stimulator (DBS) surgery using Hidden Markov Models (HMMs).
• To assess the feasibility of using HMMs to interpret intraoperative microelectrode recording (MER) data and integrate it with anatomical information for improved surgical guidance.

Main Methods:

• A 6-state Hidden Markov Model (HMM) was designed and trained.
• The HMM system was evaluated using simulated surgical data for subthalamic nucleus (STN) deep brain stimulator (DBS) placement.
• The system's performance was assessed based on accuracy, sensitivity, specificity, and anatomical localization error.

Main Results:

• The automated system achieved 98.5% accuracy in identifying the correct brain location.
• Sensitivity for detecting microelectrode recording (MER) passes intersecting the subthalamic nucleus (STN) was 100%, with a specificity of 84.9%.
• The mean error in calculating the anatomical location of MER passes was 0.06 mm in the medial-lateral axis.

Conclusions:

• The prototype automated deep brain stimulator (DBS) navigation system using Hidden Markov Models (HMMs) shows promising results.
• Automated intraoperative navigation for DBS surgery using HMMs appears feasible, potentially enhancing surgical precision and efficiency.
• Further development and validation are warranted to translate this technology into clinical practice.