Predicting in vivo MRI Gradient-Field Induced Voltage Levels on Implanted Deep Brain Stimulation Systems Using Neural...

Area of Science:

• Biomedical Engineering
• Computational Neuroscience
• Medical Imaging

Background:

• Magnetic Resonance Imaging (MRI) gradient fields can induce hazardous voltages in implanted deep brain stimulation (DBS) systems.
• These induced voltages may lead to unintended stimulation or system malfunction.
• Current electromagnetic (EM) simulation methods for predicting these voltages are accurate but resource-intensive.

Purpose of the Study:

• To develop a computationally efficient and accurate predictive model for MRI gradient-field induced voltages in DBS systems.
• To integrate machine learning with computational modeling to overcome limitations of traditional simulation approaches.
• To enhance the safety and reliability of DBS systems during MRI scans.

Main Methods:

• Simulated MRI gradient-field induced voltage levels across six adult human anatomical models with relevant DBS trajectories.
• Trained predictive artificial neural network (ANN) regression models on the generated dataset.
• Validated ANN model performance using leave-one-out cross-validation.

Main Results:

• Generated over 180,000 unique gradient-induced voltage levels.
• Selected a two-fully connected layer ANN for superior generalizability over other machine learning algorithms.
• Achieved rapid predictions (thousands per second) with a mean-squared error under 200 mV.

Conclusions:

• Successfully integrated machine learning with computational modeling to create an accurate predictive tool.
• The developed ANN model offers a fast and reliable method for assessing MRI-related risks in DBS systems.
• This approach facilitates safer clinical use of DBS systems in patients requiring MRI.