Abstract 17489: Regional Strain Shows Greater Interplatform Variability Than Global Strain
Introduction: Strain (S) analysis is a useful adjunct to conventional echocardiography but is limited by vendor-specific restrictions. Analysis of DICOM (D) images enables extraction of S regardless of image source. Existing data show good agreement in Global (G) Longitudinal S values between native polar (P) and processed D images. To further evaluate agreement we compared Regional S (RS) values from the same images in P & D formats.
Hypothesis: RS values will have variability between P & D analysis. We will accept good agreement if there is no significant difference in mean RS values in P and D analysis, and if the correlation coefficient (r) is ≥ 0.85.
Methods: 20 patients (54 +/− 12 years) had echocardiograms with a General Electric (GE) Vivid 7 scanner: 7 normals, 13 cardiomyopathies. 4,3 & 2-chamber LV apical views yielding 60 datasets were analyzed using EchoPAC (GE, version 10.8.1) in P format and Velocity Vector Imaging (VVI, Siemens, version 3.5) in D format by 3 blinded observers. RS data was analyzed using an 18-segment LV model.
Results: While agreement for GS was good (r=0.90, p<0.0001), there was low (though generally significant, p<0.05) correlation between P & D RS. 1 segment showed r ≥ 0.85, with 6 segments showing r ≥ 0.75 (Table 1). D images yielded lower S in magnitude in 13 of 18 segments (mean P-D -2.1, range -5.0 to 1.47). There was low inter-observer agreement for RS (Table 2).
Conclusion: LV Longitudinal RS from DICOM images do not agree well with native Polar S for the majority of cardiac segments, with fairly high inter-observer variability. Higher agreement for GS is likely due to averaging of RS across LV segments. Standardization efforts are essential to address this issue.
- © 2011 by American Heart Association, Inc.