Personalizing Deep Brain Stimulation Therapy for Parkinson's Disease With Whole-Brain MRI Radiomics and Machine...

Area of Science:

• Medical Imaging
• Machine Learning
• Neurology

Background:

• Deep brain stimulation (DBS) is a key treatment for advanced Parkinson's disease (PD).
• White matter alterations in the brain can indicate PD progression and predict DBS outcomes.
• Predictive biomarkers are needed for optimal patient selection for DBS therapy.

Purpose of the Study:

• To develop a machine learning (ML) model for predicting DBS therapy outcomes in PD patients.
• To utilize whole-brain white matter radiomics and clinical variables for patient selection.
• To assess the predictive performance of various ML algorithms.

Main Methods:

• 120 PD patients undergoing subthalamic nucleus DBS were analyzed.
• One-year follow-up assessed motor response using the Unified Parkinson's Disease Rating Scale-part III (UPDRS-III).
• Whole-brain white matter radiomics and clinical data were used with logistic regression and random forest models.

Main Results:

• The random forest model demonstrated superior performance with an AUC of 0.99, accuracy 0.95, sensitivity 0.93, and specificity 0.97.
• The logistic regression model achieved an AUC of 0.93, accuracy 0.88, sensitivity 0.84, and specificity 0.91.
• Radiomics and clinical data integration improved predictive accuracy for DBS response.

Conclusions:

• ML models can serve as clinical decision support tools for personalized PD therapy.
• A trade-off exists between model performance and interpretability.
• Further large-scale clinical trials are necessary to build clinician and patient trust in AI for DBS selection.