Predictive Clinical Decision Support System Using Machine Learning and Imaging Biomarkers in Patients With...

Area of Science:

• Neuroscience
• Medical Imaging
• Pain Management

Background:

• Chronic pain is linked to brain structure and function changes.
• Selecting patients for spinal cord stimulation (SCS) is challenging, with high explant rates.
• Current selection relies on functional variables and pain scores, lacking predictive accuracy.

Purpose of the Study:

• To evaluate imaging biomarkers, specifically functional connectivity (FC) and brain volumetry, in Failed Back Surgery Syndrome (FBSS) patients.
• To develop a clinical decision support system (CDSS) integrating neuroimaging and clinical data for predicting SCS therapy effectiveness post-trial.

Main Methods:

• Prospective, observational, single-center study using resting-state functional MRI (rs-fMRI).
• Region of interest (ROI) to ROI and seed-to-voxel analyses compared FC and volume changes.
• Machine learning models assessed clinical variables and imaging biomarkers to predict therapy responders (R-G).

Main Results:

• Volumetric differences identified in the left putamen; significant FC differences found in four brain area pairs, including the right insular cortex.
• Linear Discriminant Analysis (LDA) achieved the highest performance in the CDSS.
• The CDSS improved diagnostic accuracy for predicting SCS responders from 29% to 96%.

Conclusions:

• The left putamen and specific brain regions play a crucial role in FBSS.
• A CDSS incorporating imaging biomarkers can significantly enhance patient selection for SCS.
• This approach promises to improve long-term outcomes for patients undergoing SCS therapy.