Novel Metric Using Laplacian Eigenmaps to Evaluate Ischemic Stress on the Torso Surface

Area of Science:

• Cardiology
• Biomedical Engineering
• Machine Learning Applications

Background:

• Myocardial ischemia pathophysiology remains incompletely understood, complicating diagnosis.
• Current diagnostic methods for ischemia have limitations in early and robust detection.
• A data-driven approach using electrocardiogram (ECG) patterns is needed to link statistical data to disease states.

Purpose of the Study:

• To extend the Laplacian Eigen-maps (LE) dimensionality reduction technique to body surface potential mapping (BSPM) for ischemia detection.
• To evaluate the efficacy of LE in identifying acute, transient ischemia episodes using BSPM data from animal and human studies.
• To assess the temporal performance and robustness of LE in detecting ischemia compared to traditional markers.

Main Methods:

• Applied Laplacian Eigen-maps (LE), a machine learning dimensionality reduction method, to body surface potential mapping (BSPM) data.
• Utilized BSPM recordings from 10 induced acute ischemia episodes in animals.
• Analyzed 200 human recordings during percutaneous transluminal coronary angioplasty (PTCA) procedures.

Main Results:

• The LE algorithm successfully detected ischemic events from BSPM data by identifying changes in trajectory morphology.
• LE demonstrated superior temporal performance in detecting ischemia compared to previous studies and clinical markers.
• The approach proved sensitive to spatiotemporal electrocardiographic consequences of ischemia-induced stress on the body surface.

Conclusions:

• Laplacian Eigen-maps applied to BSPM effectively detect acute myocardial ischemia in both animal models and human patients.
• This data-driven method offers an improved, earlier, and more robust detection of ischemic events.
• The findings support the expansion of LE for clinical application in diagnosing myocardial ischemia.