Abstract 13673: DICOM-Based Strain Shows Low Variability Compared to Strain from Polar Source Images in a Heterogenous Population

Abstract

Introduction: Two-dimensional Strain (2D S) is a powerful technique to evaluate myocardial deformation. We compared S from images in raw Polar (P) format with the same images in DICOM (D) in adult & pediatric patients to test agreement.

Methods: 110 patients (55/110 pediatric, 80/110 normal LV function) had echocardiograms with a General Electric (GE) Vivid 7 or E9. 4,3 & 2-chamber LV apical views yielding 330 datasets were analyzed with EchoPAC (GE, v. 10.8.1) in P & Velocity Vector Imaging (VVI, Siemens, v. 3.5) in D by 3 blinded readers. Variability in measurements was tested with 2-way analysis of variance with calculation of intra- & interobserver standard error of measurement (SEMintra, SEMinter) and minimum discernable difference (MDDintra, MDDinter) for 95% confidence interval.

Results: Mean Global Longitudinal S (GLS) by EchoPAC (P) was -18.1 +/- 4.4 & VVI (D) -15.3 +/- 4.1. D yielded lower absolute GS by mean of 2.8 (± SD 2.7, p < 0.0001), Fig. 1. SEMintra & SEMinter were similar for P & D, Fig. 2. The minimum discernable difference required to detect significant variability between 3 observers (MDDinter) was 4.21 & 4.99 S points (P & D respectively). On repeat reads of 20 studies MDDintra was 3.41 & 4.57 for P & D GS respectively. t-Tests did not show statistically significant differences in SEM or MDD between P & D analysis.

Conclusion: DICOM analysis for GLS is reliable & reproducible. There are no significant differences in SEM between analysis systems. A statistically significant change in GS occurs with changes > 3-4 strain points on repeat measurements. D analysis via VVI yields lower absolute GS values compared to source P images, so P & D analyses are not interchangeable.