Deep Learning-Based Automated Pressure-Strain Loop Analysis

Noninvasive left ventricular pressure-strain loop analysis estimates myocardial work from longitudinal strain, brachial systolic blood pressure, and cardiac valve-event timing. It provides global work index, constructive work, wasted work, and work efficiency. This study evaluates a deep-learning method that automates myocardial work and loop-shape analysis. The method may reduce analysis time, improve reproducibility, and identify patterns associated with heart failure.

Inclusion Criteria: Heart failure participants: * Age ≥18 years * Able to provide informed consent * Previous myocardial infarction * Left ventricular ejection fraction ≤40% after myocardial infarction Healthy volunteers: * Age ≥18 years * Able to provide informed consent Exclusion Criteria: Heart failure participants: * Unstable disease * Contraindication to exercise testing or echocardiography, as judged by an experienced cardiologist Healthy volunteers: * Hypertension * Diabetes mellitus * Dyslipidemia * Current smoking * Pregnancy * Significant systemic disease * Cardioactive medication affecting heart rate, myocardial contractility, or vascular tone * Echocardiographic findings suggesting structural heart disease