22 Inter-field strength agreement of left ventricular strain and strain rate using Tissue Tracking and AI derived global longitudinal shortening

Background The aim of this study was to assess the inter-field strength agreement of left ventricular (LV) strain, strain rates and artificial intelligence (AI) landmark-based global longitudinal shortening at 1.5T and 3T, using routinely acquired cardiovascular magnetic resonance (CMR) cine images. Methods This was a prospective, randomised cross-over observational study in which healthy individuals completed two CMR scans at 1.5T and 3T in a randomised order within 30 minutes. Short and long-axis balanced steady state free precession cine imaging was acquired with retrospective electrocardiogram gating and scanning conditions standardised between field strengths. Analysis was undertaken offline by a single experienced observer blinded to participant details. CMR Tissue Tracking (cvi42, Circle Cardiovascular Imaging, Calgary, Canada) was used to derive LV global longitudinal, circumferential and long and short axis radial systolic strain and peak systolic, early diastolic and late diastolic strain rates. Global longitudinal shortening was measured from long axis images using a fully automated method with an AI deep neural network model. All strain values are expressed as positive numbers. Results Twenty-two individuals (mean age 36±12 years; 45% male) completed the study. There were no significant differences in blood pressure or heart rate during scanning between field strengths. Figure 1 shows scatterplots and Bland-Altman plots for LV systolic strain measurements and global longitudinal shortening at 1.5T and 3T. Strain and strain rate measurements derived from short axis images (circumferential and short axis radial) showed good to excellent agreement between field strengths (intraclass correlation co-efficient (ICC) range 0.78–0.91). However, strain and strain rate measurements derived from long axis images (longitudinal and long axis radial) showed poor to fair agreement (ICC range 0.39–0.71). Global longitudinal shortening showed good agreement (ICC = 0.81). Minimal bias was seen in all measurements between field strengths. Conclusions Inter-field strength agreement of short axis derived LV strain and strain rate measurements at 1.5T and 3T was good to excellent. By contrast, inter-field strength agreement of long axis derived measurements was poor to fair, but for AI landmark-based global longitudinal shortening agreement was good. These findings should be considered when assessing strain values from different field strengths.