Ultrasound speckle tracking is frequently used to quantify myocardial strain, and magnetic resonance imaging (MRI) feature tracking is rapidly gaining interest. Our aim is to validate cardiac MRI feature tracking by comparing it with the gold standard method (i.e., MRI tagging) in healthy subjects and patients. Furthermore, we aim to perform an indirect validation by comparing ultrasound speckle tracking with MRI feature tracking. Forty-two subjects (17 formerly preeclamptic women, three healthy women, and 22 left bundle branch block patients of both sexes) received 3-T cardiac MRI and echocardiography. Cine and tagged MRI, and B-mode ultrasound images, were acquired. Intrapatient global and segmental left ventricular circumferential (MRI tagging vs. MRI feature tracking) and longitudinal (MRI feature tracking vs. ultrasound speckle tracking) peak strain and time to peak strain were compared between the three techniques. Intraclass correlation coefficient (ICC) (< 0.50 = poor, 0.50-0.75 = moderate, > 0.75-0.90 = good, > 0.90 = excellent) and Bland-Altman analysis were used to assess correlation and bias; pless than 0.05 indicates a significant ICC or bias. Global peak strain parameters showed moderate-to-good correlations between methods (ICC = 0.71-0.83, p< 0.01) with no significant biases. Global time to peak strain parameters showed moderate-to-good correlations (ICC = 0.56-0.82, p< 0.01) with no significant biases. Segmental peak strains showed significant biases in all parameters and moderate-to-good correlation (ICC = 0.62-0.77, p< 0.01), except for lateral longitudinal peak strain (ICC = 0.23, p= 0.22). Segmental time to peak strain parameters showed moderate-to-good correlation (ICC = 0.58-0.74, p< 0.01) with no significant biases. MRI feature tracking is a valid method to examine myocardial strain, but there is bias in absolute segmental strain values between imaging techniques. MRI feature tracking shows adequate comparability with ultrasound speckle tracking.
Read full abstract