AI-Derived Mitral Annular Relaxation on Cardiac MRI Accurately Identifies Diastolic Dysfunction in Hypertrophic Cardiomyopathy

Background: Hypertrophic cardiomyopathy (HCM) is an inherited cardiomyopathy leading to myocardial hypertrophy, myofibril disarray, and fibrosis. Functional impairment due to HCM affects at least 1 in 3000 individuals in the United States and may be due to left ventricular outflow tract (LVOT) obstruction, myocardial ischemia, arrhythmias, valvular abnormalities, and diastolic dysfunction. Traditionally, diastolic function is assessed by transthoracic echocardiography (TTE) (i.e., tissue doppler imaging, mitral inflow on spectral doppler). Diastolic function can also be assessed via cardiac MRI; however, existing methods have been labor-intensive, utilizing modalities such as transmitral velocities using phase-contrast sequences and left ventricular strain. Here, we present an AI-derived measure of mitral annular relaxation (e’) on cardiac MRI, demonstrating a novel and easily assessable approach to evaluating diastolic dysfunction in HCM.

Methods: A total of 14 patients with confirmed HCM underwent a cardiac MRI on a 3 Tesla cardiac MRI scanner (Siemens, Germany). Cardiac function cine images were obtained in multiple planes via gated steady state free precession sequences in the short axis, 4-, 3-, and 2-chamber planes. Parametric mapping (T1 pre- and post-gadolinium contrast, T2), myocardial perfusion, and late gadolinium enhancement sequences were also acquired. MRI image analysis was performed on SuiteHeart (Neosoft, Pewaukee, Wisconsin) and septal and lateral e’ values were derived by the analytic software after a line, denoting the mitral annular plane, was drawn slightly distal to the mitral annulus (Figure 1). This line was not adjusted if it adequately tracked the endomyocardial motion.

Results: Patients were stratified according to normal versus reduced mitral e’ values on TTE (Table 1). Abnormal diastolic e′ parameters were observed in 9 of 14 HCM patients (64%). Among patients with reduced diastolic e′ parameters on TTE, 7 of 9 (78%) demonstrated obstructive physiology while only 1 of 5 (20%) demonstrated obstructive physiology in patients with normal diastolic parameters. Compared with those with normal diastolic function, patients with abnormal diastolic function had higher Valsalva LVOT gradients (52.4 ± 25.7 vs 17.8 ± 24.3 mmHg, p = 0.03) and a larger left ventricular end-diastolic volume index on MRI (88.5 ± 16.4 vs 69.8 ± 10.9 ml/m², p = 0.04).

On MRI, lateral e′ values were significantly reduced in HCM patients with diastolic dysfunction compared with those without diastolic dysfunction (4.7 ± 2.5 vs 8.1 ± 2.7 cm/s, p = 0.03). Septal e′ values trended lower but did not achieve statistical significance (4.0 ± 1.5 vs 7.7 ± 7.3 cm/s, p = 0.15).

Conclusion: In HCM patients, AI-derived mitral annular relaxation (e′) on cardiac MRI showed strong concordance with tissue Doppler imaging on TTE. These findings support the potential of AI to streamline and enhance diastolic assessment with MRI, warranting further validation in larger studies.

Figure 1. 4 chamber view shown at (A) End-Diastole and (B) End-Systole. The green line denotes the mitral annular plane and the purple and orange shapes denote the right and left atria, respectively.  
Table 1. Cardiac MRI and Transthoracic Echocardiographic Characteristics of Hypertrophic Cardiomyopathy Patients.