Reproducibility assessment of a biomechanical model-based elasticity imaging method for identifying changes in left ventricular mechanical stiffness

Abstract

. Purpose Cardiotoxicity of antineoplastic therapies is increasingly a risk to cancer patients treated with curative intent with years of life to protect. Studies highlight the importance of identifying early cardiac decline in cancer patients undergoing cardiotoxic therapies. Accurate tools to study this are a critical clinical need. Current and emerging methods for assessing cardiotoxicity are too coarse for identifying preclinical cardiac degradation or too cumbersome for clinical implementation. Approach In the previous work, we developed a noninvasive biomechanical model-based elasticity imaging methodology (BEIM) to assess mechanical stiffness changes of the left ventricle (LV) based on routine cine cardiac magnetic resonance (CMR) images. We examine this methodology to assess methodological reproducibility. We assessed a cohort of 10 participants that underwent test/retest short-axis CMR imaging at baseline and follow-up sessions as part of a previous publicly available study. We compare test images to retest images acquired within the same session to assess within-session reproducibility. We also compare test and retest images acquired at the baseline imaging session to test and retest images acquired at the follow-up imaging session to assess between-session reproducibility. Results We establish the within-session and between-session reproducibility of our method, with global elasticity demonstrating repeatability within a range previously demonstrated in cardiac strain imaging studies. We demonstrate increased repeatability of global elasticity compared to segmental elasticity for both within-session and between-session. Within-subject coefficients of variation for within-session test/retest images globally for all modulus directions and a mechanical fractional mechanical stiffness anisotropy metric ranged from 11% to 28%. Conclusions Results suggest that our methodology can reproducibly generate estimates of relative mechanical elasticity of the LV and provides a threshold for distinguishing true changes in myocardial mechanical stiffness from experimental variation. BEIM has applications in identifying preclinical cardiotoxicity in breast cancer patients undergoing antineoplastic therapies.

Myocardial Elasticity Imaging Correlates with Histopathology in a Model of Anthracycline-Induced Cardiotoxicity

“…To assess region-based changes that correspond to collected pathological tissue samples, the two septal segments, two posterior segments, and two anterior segments were each combined to create septal, posterior, and anterior segments, respectively. Note that previous work shows minimal intraobserver differences in BIEM-calculated elasticity 15 .…”

“…We utilize a previously developed biomechanical model-based elasticity imaging methodology (BEIM) 14 to estimate spatial elasticity maps of the LV in AGMs. While a detailed description of the method can be found in our prior work 14, 15 , a brief description follows. Observed deformation representing inherent cardiac motion during passive diastole was obtained from non-rigid image registration.…”

“…The result of the method is spatial circumferential and radial elastic modulus maps of the LV in the circumferential and radial directions. As described in previous work 15 , since fully-constrained displacement-based boundary condition solutions are indeterminate, we normalize elasticity solutions by a time-stress integral calculated over the endocardium 28 with the time-stress-integral serving as a normalization factor to allow for image comparisons across subjects and time points and representative of the stress on the endocardium throughout passive diastole based on the spatial elasticity solution. Normalized circumferential, radial, and shear modulus maps were then compared across image time point acquisitions.…”

“…However, there remains a need for the development of a validated non-invasive imaging methodology to directly measure myocardial elasticity prior to its utilization in future studies to determine if it may be an early predictor. To address this need, we developed a non-invasive biomechanical model-based elasticity imaging methodology (BEIM) to estimate LV elasticity utilizing routinely acquired cine CMR images 14 and have characterized the methodological reproducibility 15 . Briefly, BEIM employs an inverse biomechanics computational modeling framework to estimate a direct intrinsic measurement of LV tissue elasticity that is not dependent upon uniform stress assumptions and not confounded by changes in LV shape or size.…”