Needle-free myocardial blood flow and reserve quantification using artificial intelligence-enhanced coronary sinus...

Quantification of coronary sinus (CS) flow has been used with pharmacologic stress as a non-invasive surrogate of global myocardial blood flow and coronary flow reserve (CFR). Whether CS flow assessment can be extended to physiological exercise stress remains uncertain. Accurate measurement during exercise is technically challenging due to the small caliber of the CS and its rapidly varying flow dynamics, particularly under exercise conditions. In this study, we evaluated the feasibility of a high-resolution, high-frame-rate cardiovascular magnetic resonance (CMR) approach for measuring post-exercise CS flow and CFR and compared these measures with quantitative myocardial perfusion imaging.

We implemented a phase-contrast sequence with non-interleaved velocity-compensated and velocity-encoded k-space acquisition and truncated phase encoding. Generative artificial intelligence (AI) synthesized high-resolution images from the low-resolution inputs and interpolated intermediate frames, effectively doubling temporal resolution. In a prospective exercise CMR study, patients with stable coronary artery disease (n = 13, 50 ± 20 years) underwent AI-enabled CS flow imaging at 1.1 × 1.1 mm² spatial and 27 ms temporal resolution, performed twice at rest for scan/re-scan repeatability and once after exercise. Quantitative perfusion imaging was performed before and post-exercise. Scan/re-scan repeatability of rest CS flow, and inter-observer repeatability of rest and post-exercise CS flow and CS flow-derived CFR were assessed using intraclass correlation coefficients (ICC). CS flow and CFR were compared with perfusion-derived myocardial blood flow and myocardial perfusion reserve (MPR) using linear regression and Pearson correlation (r).

Analysis was successful in all rest and 11 of 13 stress scans; two were excluded due to electrocardiogram (ECG) mis-gating. CS flow showed excellent scan/re-scan (ICC = 0.97 [0.91-0.99]) and inter-observer repeatability (ICC = 0.97 [0.92-0.99]). CS flow showed good correlation with perfusion-derived myocardial blood flow (y = 0.95×, r = 0.61, P = 0.002). CS flow-based CFR also correlated well with perfusion-derived MPR (y = 1.02×, r = 0.67, P = 0.025).

We demonstrate the feasibility of a high-resolution, high-frame-rate CMR technique for quantifying post-exercise CS flow and CFR, with excellent repeatability and good agreement with perfusion-derived measures. This approach shows promise for assessing global myocardial perfusion after physiological exercise without pharmacologic stress, warranting further validation.