Patient-specific variability in breath-hold positions during cardiac magnetic resonance imaging has a negligible effect on measures of cardiac mechanics

Abstract

Background: Cardiac magnetic resonance (CMR) can be used to quantify cardiac mechanics from images that are generally acquired during an end-expiratory breath-hold. Unfortunately, it is difficult for subjects to hold their breath at the exact same position when undergoing a series of breath-holds during a typical CMR study. The effect of patient-specific variability in breath-hold positions on measures of cardiac mechanics has not been investigated. We hypothesized that normal variability in breathhold positions would significantly affect estimates of left ventricular strains and torsion. Methods: Ten healthy volunteers (Age: 29±10 years, 60% female) and 6 patients with a history of cardiovascular disease or myocardial infarction (Age: 58±9 years, 50% female) were consented. A 3T Siemens Tim Trio was used to measure the diaphragm position during ten 10-second breath-holds to determine each subject's breath-hold range. The diaphragm was measured at a sampling rate of 3Hz with the respiratory navigator sequence built into the DENSE sequence. A navigator feedback system enabled subjects to view their diaphragm position in real-time during image acquisition by using an angled mirror and projector screen placed at the back of the scanner bore. We acquired navigator-gated 2-chamber, 4-chamber, basal, mid-ventricular, and apical slices of 2D cine DENSE at the subject-specific maximum, middle and minimum breath-hold positions (with repeated middle position for inter-test quantification). The 2D DENSE parameters were: 6 spiral interleaves, voxel size = 2.8x2.8x8 mm, TE/TR = 1.08/17, flip angle = variable 20°, navigator acceptance window = ±3mm. Radial, circumferential, and longitudinal strains and torsion were calculated for each subject and compared between diaphragm locations using a repeated measures ANOVA. The inter-test 95% limits of agreement were calculated for strains and torsion using the Bland-Altman method. Since torsion is quantified from three independent short-axis images acquired during separate breath-holds, and we acquired each short-axis image at 3 different breath-hold locations, torsion was computed from each of 27 possible slice combinations (9 unique combinations). Results: One healthy subject was excluded due to patient movement. The average breath-hold range was 10 mm for both the healthy and patient groups. The breath-hold range varied from 4 to 19 mm for both groups combined. There were no clear trends or significant differences in strains and torsion measured at the different diaphragm positions. Furthermore, the differences were smaller than the inter-test 95% limits of agreement (Fig 1). Most torsion permutations fell within the inter-test 95% limits of agreement, suggesting that diaphragm position has minimal effect on the calculation of torsion (Fig 2). Conclusions: Different patient-specific breath-hold positions had no significant effect on the quantification of peak left ventricular cardiac strains and torsion from two-dimensional DENSE CMR. (Figure Presented).