Contour regularized left ventricular strain analysis from cine MRI

Abstract

Quantitative measurements of left ventricular (LV) strain are important in the diagnosis and management of patients with heart disease and tracking the efficacy of treatments over time. Tagged cardiac magnetic resonance imaging (MRI) is an established method for non-invasively measuring LV strains and strain rates, but non-tagged (or cine) MRI is the most commonly-used cardiac MRI protocol. In this paper, we propose a new algorithm for computing 2-D strains and strain rates from cine MRI that incorporates information from contours drawn at end-diastole and end-systole. These contours are typically drawn anyway at many institutions to compute LV volumes and contain important information on heart's deformation. Validation results on 18 normal human volunteers and 20 patients with myocardial infarction show good agreement between strains computed from cine MRI and those computed from tagged MRI using HARP analysis. Further validation on 45 patients with hypertension demonstrate that diastolic strains computed from cine MRI can measure changes in diastolic function common in hypertensive patients.