Application of artificial intelligence on magnetocardiology

Area of Science:

• Cardiology
• Biomedical Engineering
• Medical Imaging

Background:

• Ischemic heart disease (IHD) causes myocardial dysfunction due to impaired microcirculation.
• Current diagnostic methods like PET/SPECT and CMR involve radiation and reproducibility concerns.
• Early and precise localization of ischemic regions is vital for effective IHD treatment.

Purpose of the Study:

• To evaluate the efficacy of machine learning algorithms using Magnetocardiography (MCG) data for diagnosing IHD.
• To assess the capability of MCG in localizing ischemic regions within specific coronary arteries.
• To identify key MCG parameters crucial for accurate ischemia detection.

Main Methods:

• Support Vector Machines, XGBoost, Multiple Layer Perceptron Networks, and Random Forest algorithms were applied.
• MCG data from patients with confirmed IHD were utilized.
• Feature selection was performed to identify the most informative MCG parameters for each coronary artery.

Main Results:

• High prediction accuracies were achieved: 0.80 for left anterior descending artery, 0.72 for ramus circumflexus, and 0.70 for right coronary artery ischemia.
• The study identified specific MCG parameters that are more indicative of ischemia.
• Optimal feature subsets were determined for improved diagnostic performance in each arterial territory.

Conclusions:

• MCG, analyzed with machine learning, presents a non-invasive and effective tool for early IHD diagnosis.
• This approach aids in the precise localization of ischemic regions, guiding treatment strategies.
• MCG offers a radiation-free alternative to conventional imaging modalities for IHD assessment.