Abstract 15401: Artificial Intelligence-Enabled Exercise-CMR for Evaluation of Cardiac Functional Reserve in Heart Failure

Abstract

Introduction: Exercise unmasks pathophysiological markers in heart failure (HF), not evident at rest. Exercise CMR (Ex-CMR) has the potential to offer an accurate and reproducible assessment of cardiac functional reserve in HF. Objective: To study the cardiac functional reserve in HF using an artificial intelligence (AI) enabled exercise CMR imaging protocol. Method: We prospectively recruited patients with an established clinical diagnosis of HF and healthy adults. Subjects were imaged at rest and immediately post-exercise using a supine ergometer positioned on the CMR scan. The imaging protocol included free-breathing, real-time, highly accelerated cine short-axis images covering the entire LV. Cine images were collected with a spatial resolution of 1.9 х 1.9 mm 2 and temporal resolution of 38.8 ms using a 14.8-fold acceleration using an in-house inline deep learning reconstruction. Pre- and post-exercise biventricular volumetric and functional parameters and their corresponding percentage changes were calculated. Results: The study included 77 subjects, consisting of 37 HF patients (24 males; 60 ± 12 years) [HFpEF, n=13; HFmrEF, n=5; HFimpEF, n=13; and HFrEF, n=6], and 40 healthy adult (20 males; 45 ± 16 years). Changes in LV end-diastolic and end-systolic volume among subgroups revealed distinct clusters (Fig 1). These findings corresponded to a blunted to a modest rise in stroke volume (SV) observed in HFpEF and HFmrEF patients, respectively, while HFimpEF and HFrEF patients exhibited a mildly to excessively higher SV rise compared to other subgroups (Fig 1). HFpEF and HFmrEF showed no improvement in SV index beyond 10 ml/m 2 . Patients with known post capillary pulmonary hypertension demonstrated an inadequate right ventricle response in preload recruitment, contributing to a blunted increase in RV SV (Fig 1). Conclusion: Our AI-enabled, high-accelerated Ex-CMR imaging protocol reveals a distinctive biventricular functional reserve response in HF.