Reproducibility of native and contrast-enhanced CMR techniques to measure lesion size following acute myocardial infarction

Purpose: To analyse the reproducibility of native and contrast-enhanced CMR techniques to measure lesion size after acute myocardial infarction (AMI). Methods and Materials: Lesions were independently quantified by 2 experienced observers in total of 120 consecutive CMRs obtained in 30 patients within the first 6 months after AMI using native and contrast-enhanced sequences. A threshold method (cutoff > 2SD of remote normal myocardium) was used on basal, midventricular and apical short-axis left ventricular slices. Bland-Altman analysis was performed to determine the agreement between the two observers. Non-parametric Levene's test was used to compare the variances of the relative differences. Results: The relative median difference of the native CMR techniques were - 1.95% (-12.7% and 9.8%) for T2w, -5.3% (-19.6% and 14.8%) for native T1 and -4.0% (-23.9% and 9.9%) for native T2 (Figure 1). Results for contrast- enhanced CMR imaging were: 2.9% (-4.5% and 10.5%) for LGE, 7.5% (-2.4% and 21.5%) for post-contrast T1 and -2.9% (- 11.4% and 9%) for ECV measurement. Bland-Altman analysis revealed a better agreement for all post- contrast sequences indicated by lower limits of agreement (Figure 1). The increased variability of native imaging was caused by higher interobserver differences in small lesions. This bias was not observed for post-contrast imaging. Conclusion: There was a good agreement between the two observers to measure lesion size after AMI. However, all post-contrast sequences had a better agreement compared to the native sequences. The low agreement of native imaging was mainly caused by higher interobserver differences in small lesions compared to lager lesions.