Work

Functional myocardial ischemia assessment based on HF-ECG microstructural features

Year: 2026

Type: article

Abstract: Early detection and stratification of functional myocardial ischemia are critical for timely intervention but remain constrained by the limited sensitivity of conventional electrocardiography (ECG) and the cost and accessibility of imaging-based diagnostics. We present a non-invasive ischemia assessment framework based on a portable 12-lead high-frequency ECG (HF-ECG) acquisition system, which captures microelectrical conduction dynamics at a finer spatiotemporal resolution than standard ECG. HF-ECG analysis of the 150–250 Hz QRS band captures microvolt-level alterations in high-frequency amplitude and morphology that may be associated with myocardial ischemia. The proposed method integrates multidimensional time–frequency analysis with anatomically guided signal decomposition to extract physiologically grounded features from the QRS complex. These features are designed to quantify key electrophysiological signatures of ischemia, including high-frequency attenuation, temporal desynchronization, and signal irregularity. Validation was performed using myocardial perfusion imaging, implemented in this study with SPECT/CT, a clinically accepted reference standard for functional ischemia. At the global level, HF-ECG features showed strong correlations with continuous ischemic burden (maximum r = 0.733, p < 0.001), and significantly differentiated between patients with low and moderate-to-severe ischemic burden ( p = 0.003). A binary classification model achieved an AUC of 0.841 under cross-validation. At the regional level, we further demonstrated the feasibility of ischemia localization across five myocardial territories, with region-wise AUCs ranging from 0.456 (apex) to 0.889 (septal). These results establish HF-ECG as a promising modality for non-invasive, scalable, and real-time functional assessment of myocardial ischemia. The analysis framework offers a foundation for future applications in ischemia screening.

Source: Biomedical Signal Processing and Control

DOI: 10.1016/j.bspc.2026.111566