The interfield strength agreement of left ventricular strain measurements at 1.5 T and 3 T using cardiac MRI feature tracking

Abstract

Background: Left ventricular (LV) strain measurements can be derived using cardiac MRI from routinely acquired balanced steady-state free precession (bSSFP) cine images. Purpose: To compare the interfield strength agreement of global systolic strain, peak strain rates and artificial intelligence (AI) landmark-based global longitudinal shortening at 1.5 T and 3 T. Study type: Prospective. Subjects: A total of 22 healthy individuals (mean age 36 ± 12 years; 45% male) completed two cardiac MRI scans at 1.5 T and 3 T in a randomized order within 30 minutes. FIELD STRENGTH/SEQUENCE: bSSFP cine images at 1.5 T and 3 T. Assessment: Two software packages, Tissue Tracking (cvi42, Circle Cardiovascular Imaging) and QStrain (Medis Suite, Medis Medical Imaging Systems), were used to derive LV global systolic strain in the longitudinal, circumferential and radial directions and peak (systolic, early diastolic, and late diastolic) strain rates. Global longitudinal shortening and mitral annular plane systolic excursion (MAPSE) were measured using an AI deep neural network model. Statistical tests: Comparisons between field strengths were performed using Wilcoxon signed-rank test (P value < 0.05 considered statistically significant). Agreement was determined using intraclass correlation coefficients (ICCs) and Bland-Altman plots. Results: Minimal bias was seen in all strain and strain rate measurements between field strengths. Using Tissue Tracking, strain and strain rate values derived from long-axis images showed poor to fair agreement (ICC range 0.39-0.71), whereas global longitudinal shortening and MAPSE showed good agreement (ICC = 0.81 and 0.80, respectively). Measures derived from short-axis images showed good to excellent agreement (ICC range 0.78-0.91). Similar results for the agreement of strain and strain rate measurements were observed with QStrain. Conclusion: The interfield strength agreement of short-axis derived LV strain and strain rate measurements at 1.5 T and 3 T was better than those derived from long-axis images; however, the agreement of global longitudinal shortening and MAPSE was good. Evidence level: 2 TECHNICAL EFFICACY: Stage 2.