Artificial intelligence-based accurate myocardial infarction mapping using 12-lead electrocardiography

Area of Science:

• Cardiology
• Artificial Intelligence
• Medical Imaging

Background:

• Assessing myocardial fibrosis (MF) post-myocardial infarction (MI) is critical for patient prognosis.
• Artificial intelligence-assisted electrocardiography (AI-ECG) shows promise for detecting MF.
• Training AI-ECG models requires extensive data; biosimulation offers a potential solution.

Purpose of the Study:

• To develop and validate a novel AI-assisted method, AI-MI-12ECG, for assessing MF using 12-lead ECG.
• To utilize a biosimulation model for training the AI-MI-12ECG system.
• To evaluate the accuracy of AI-MI-12ECG in detecting MF presence, location, and size in post-MI patients.

Main Methods:

• A biosimulation model was used to train the AI-MI-12ECG system.
• 182 post-MI patients were enrolled in a prospective study.
• AI-MI-12ECG findings were compared against late gadolinium-enhanced (LGE) cardiac magnetic resonance (CMR) imaging.

Main Results:

• AI-MI-12ECG demonstrated a strong correlation (R=0.955) with CMR-LGE in identifying MI location.
• The method achieved high accuracy in detecting fibrosis across coronary artery territories: LAD (0.95), RCA (0.97), and LCX (0.91).
• Receiver operating characteristic curves showed excellent performance for AI-MI-12ECG: 0.95 (LAD), 0.95 (RCA), and 0.89 (LCX).

Conclusions:

• The AI-MI-12ECG method, trained via biosimulation, aligns well with CMR-LGE in post-MI patients.
• This AI-ECG approach shows significant potential for accurate myocardial fibrosis detection.
• AI-MI-12ECG can effectively identify patients with substantial infarct burdens, aiding in prognosis.