Author + information
- Tomoko Negishi,
- Kazuaki Negishi,
- Koji Kurosawa,
- Krasimira Hristova,
- Julie Lemieux,
- Svend Aakhus,
- Martin Penicka,
- Bogdan Popescu,
- Dragos Vinereanu,
- Goo-Yeong Cho,
- Paaladinesh Thavendiranathan and
- Thomas Marwick
Background: While attention has focused on between-vendor variability, within-vendor variations in software (SW) version. Many pts require follow-up of LV status over months to years, during which time SW is often updated. We sought the impact of SW versions in cardio-oncology pts.
Methods: Global longitudinal strain (GLS) were measured in patients from 9 international institutions (3 Europe, 2 North America, 3 Asia and 1 Australia) by sites with one single version of software. Same images were blindly measured by one core-lab reader with 3 consecutive versions of software. Concordance was evaluated with intraclass correlation coefficient (ICC), Bland-Altman plot, mean difference with standard deviation, and coefficient of variance. As there was a change in the algorithm from Ver1 to 2, but minimal from 2 to 3, the latter acted as a control.
Results: Among 58 cancer pts (age 52±12, 56 female), site-derived GLS was -20.6±2.1%, close to every version at the core-lab (Ver1; -21.2±2.2 (p=0.71), Ver2; -21.2±2.5% (p=0.78), Ver3; -21.2±2.2% (p=1.00), group p=0.36). ICC between site and core-lab Ver1 was 0.84 [95%CI 0.71-0.82], similar to site and Ver2 (0.86 [95%CI 0.74-0.93]) and site and Ver3 (0.85 [0.74-0.92]). Correlations between site and the 3 versions were good (site vs. Ver1, r=0.78; vs. Ver2, r=0.80; vs. Ver3; r=0.76, all p<0.0001), the differences in strain value were negligible (Table).
Conclusions: The effect of software version on sequential follow-up over the last 3 years seems minimal.
Poster Hall, Hall C
Friday, March 17, 2017, 3:45 p.m.-4:30 p.m.
Session Title: Emerging Applications of Echocardiography
Abstract Category: 28. Non Invasive Imaging: Echo
Presentation Number: 1160-221
- 2017 American College of Cardiology Foundation