Adapting the listening time for micro-electrode recordings in deep brain stimulation interventions

Area of Science:

• Neurosurgery
• Computational Neuroscience
• Medical Device Technology

Background:

• Deep brain stimulation (DBS) relies on precise electrode placement in subcortical structures like the subthalamic nucleus.
• Micro-electrode recordings (MERs) guide surgeons by analyzing auditory signals to identify anatomical locations.
• Current automated MER analysis lacks flexibility, limiting adaptive data collection based on certainty.

Purpose of the Study:

• To develop and evaluate a flexible algorithm for MER signal collection in DBS surgery.
• To enable adaptive termination of MER data acquisition based on model confidence.
• To improve the efficiency of DBS electrode implantation procedures.

Main Methods:

• An algorithm was developed to terminate MER signal collection when sufficient confidence is reached.
• The algorithm's performance was evaluated using three underlying models: a neural network and two Bayesian extensions.
• Parameterization of the adaptive approach was explored.

Main Results:

• A Bayesian model incorporating network certainty demonstrated superior performance.
• This model showed relative insensitivity to parameterization.
• Early signal classification was achieved without an increase in error rates.

Conclusions:

• A novel algorithm enables early termination of MER data collection by monitoring neural network confidence.
• This approach has the potential to significantly reduce the time needed for anatomical identification during DBS surgery.
• Improved efficiency in DBS electrode implantation is a key potential benefit.