Intra- and Inter-Observer Reproducibility of a 2-D Shear Wave Elastography Technique and the Impact of Ultrasound Experience in Achieving Reliable Data

Abstract

The aim of the study described here was to evaluate the reproducibility of a new shear-wave elastography (SWE) technique, 2-D SWE.GE, and the impact of ultrasound experience in acquiring reliable measurements, as no official recommendations are available for this system. Elastographic measurements (EMs) were obtained in 60 patients using 2-D-SWE with the GE Logiq E9. Three examiners with different levels of experience in ultrasound-based elastography performed 10 valid EMs on each subject: a novice (C.P.) who had no experience in liver elastography and had performed fewer than 50 ultrasound examinations; an elastography expert (A.M.S.) who had more than 1 y of liver elastographic experience in four elastographic methods and had performed more than 1000 ultrasonography examinations; and an ultrasound expert (T.M.) who had no experience in liver elastography and had performed more than 1000 ultrasound examinations. Medians and interquartile ranges were calculated (m/s). We used the inter-class correlation coefficient and Bland–Altman plots with 95% lower and upper limits of agreement to assess the inter- and intra-observer reproducibility of 2-D-SWE.GE measurements. The final study group included 60 patients, 56.7% women and 43.3% men, with a mean age of 33.08 ± 13.83 y and mean body mass index of 22.85 ± 4.04 kg/m2. In this group, 73.3% were healthy volunteers and 26.7% had compensated liver cirrhosis. We did not find significant differences between EMs taken by the examiners overall and across study groups. The overall agreement between examiners was excellent: 0.915 (95% confidence interval [CI]: 0.870–0.946). The agreement between the novice and the experienced examiners, respectively, was good to excellent (novice and ultrasound expert: 0.908, 95% CI: 0.846–0.945; novice and elastography expert: 0.885, 95% CI: 0.808–0.931). The intra-observer reproducibility for each of the examiners was excellent; however, the inter-class correlation coefficients were higher for the examiners more experienced in elastography: 0.936 (95% CI: 0.896–0.963) versus 0.966 (95% CI: 0.943–0.980) versus 0.984 (95% CI: 0. 973–0.991). The good ICCs for the median values indicate that 2-D-SWE.GE is a reproducible method. Ultrasound experience did not significantly influence the results.