Abstract 11438: Texture Analysis Identifies Differences in Cardiac Magnetic Resonance T1 Maps of Pediatric Heart Transplant Patients with Rejection

Abstract

Introduction: Graft rejection is a significant contributor to morbidity and mortality in pediatric heart transplant recipients, despite surveillance by endomyocardial biopsy (EMB). Cardiac Magnetic Resonance (CMR) T1 mapping has shown potential for noninvasive diagnosis of rejection. However, variability in normative T1 values across field strengths, vendors, and individual scanners limits the utility of absolute T1 values to diagnose rejection across institutions. Texture analysis is an emerging method for quantifying spatial distribution of relative voxel intensities rather than using the absolute voxel intensities themselves and may serve as an additional method for evaluating T1 maps. Hypothesis: We hypothesized that T1 maps from patients with rejection would exhibit differences in texture features from patients without rejection. Methods: This study included 19 cases from 17 pediatric transplant patients (mean age 11.8 ± 5 year, graft age 5.9 ± 4.7 years, 41% female) who underwent CMR imaging and EMB. 8 short axis slices of breath-held T1 mapping using Modified Look-Locker Inversion (MOLLI) recovery on 1.5T Siemens scanner were obtained. Left ventricular myocardium was semiautomatically segmented from T1 maps, excluding the blood pool and epicardial myocardium. Image texture analysis was performed on a single mid-ventricular slice from each case. Rejection was defined as acute cellular rejection (biopsy grade 1R or higher) or negative biopsy with decreased function requiring treatment. T-test analysis was performed to compare differences in texture analysis between patients with and without rejection. Results: Analysis of T1 maps showed differences (p-value ≤ 0.10) between patients with and without rejection in texture features of energy (0.062 ± 0.012 vs 0.091 ± 0.051), entropy (4.59 ± 0.23 vs 4.23 ± 0.60), and variance (0.031 ± 0.005 vs 0.025 ± 0.009). Discussion: In this small pilot study, texture analysis identified differences in tissue character between patients with and without graft rejection. Evaluation of tissue character via texture analysis may provide an additional method for diagnosis of rejection, given the variability in normative T1 values across institutions.