IJCS | Volume 31, Nº6, November / December 2018

632 Felix et al. RV function by 2D strain in left-sided valve disease Int J Cardiovasc Sci. 2018;31(6)630-642 Original Article 40-80 frames/sec in patients in sinus rhythm and five consecutive cycles in patients with atrial fibrillation (AF). The data were exported at the end of the test to a workstation (EchoPac BT12, GE Vingmed, Horten, Norway) for further offline analysis. Preliminary analysis was performed online in the ultrasound machine was performed online in the ultrasound machine to check if the image quality of the loops was good enough to permit adequate tracking of the acousticmarkers (speckles) of themyocardiumduring the entire cardiac cycle. STE analysis was performed semi-automatically by the system, after manual setting of 3 points on the endocardial border of the right ventricle by the operator (2 basal and one at the apex). When the region of interest (ROI) included the whole thickness of the right ventricle and excluded other structures such as trabeculae, moderator band and valvular tissue, the processingwas started, and analysis proceededon a frame- to-frame basis using an automatic tracking system (Figure 1). If the tracking was poor, the operator could repeat the the acquisition of loops, readjusting the endocardial tracing (editing) or change software parameters such as ROI width, frame rate or gain, until an adequate tracking of the entire myocardium was achieved. The ROI generated by the software included basal, mid and apical segments of RV free wall and septum, dividing it into 6 segments (Figure 1). Longitudinal peak strain values were measured for each segment, and the RV free wall longitudinal strain (RVFWS) and the RV global longitudinal strain (RVGS), analyzed by 2DS, were calculated by averaging the values from the three segments of the RV free wall and all the six segments along the entire right ventricle, respectively. These initial results were blinded to the investigators until the offline analysis of the remaining parameters was performed. Three-dimensional echocardiography 3DE was performed in all subjects immediately after the two-dimensional echocardiographic examination using the same ultrasound machine, equipped with a 4V probe. RV three-dimensional (3D) images were obtained in a full-volume dataset from the apical four- chamber view, optimized for analysis of RV function. Multi-beat (3-6 beats) data were obtained during apnea, on the multislice (short axis) visualizationmode, tomake sure that the right ventricle was entirely included in the dataset (Figure 2). All the measurements of RV volumes and EF were made off-line, using a dedicated software (TomTec Imaging Systems GmbH, Munich, Germany). Semi- automatic analysis was performed, with manual tracing of the endocardial borders in end-systolic and end-diastolic frames in the sagittal, four-chamber and coronal views, obtained from the full-volume dataset. In addition, end-diastolic volume (EDV), end-systolic volume (ESV), stroke volume and EF were calculated using the software (Figure 2). Assessment of reproducibility Evaluation of inter- and intraobserver reproducibility were performed. Fourteen patients, chosen by simple random selection, were assessed by two observers for analysis of interobserver variability, and for intra- observer variability, the second analysis was performed with a minimum interval of two weeks from the first analysis. The readers were blinded to previous measurements. Interobserver and intraobserver variability were assessed using the intraclass correlation coefficient and Bland-Altman analysis. 21 Statistical analysis Demographic data are presented as mean ± standard deviation (SD) and categorical data are presented as frequencies. Normality of the distribution of numerical variables was tested by the Kolmogorov-Smirnov test; normally distributed variables were expressed as mean ± SD and variables with abnormal distribution as median with interquartile range. We compared all RV function parameters between subgroups of patients according to their predominant valvular lesion. To reduce the occurrence of alpha error, we used one-way ANOVA test with Bonferroni post-hoc correction. Conventional echocardiographic and 2DS parameters were compared with real-time 3DE RVEF using Spearman correlation test. For comparisons between groups of patients (AandB), basedon thepresenceofRVdysfunctiondefined as RVEF (3DE) < 44%, we used nonparametric Mann- Whitney U test. ROC (receiver operating characteristic) curve analysis was used to assess the clinical utility of all RV function variables in defining RV dysfunction. P values < 0.05 were considered statistically significant. Statistical analyses were performed using SPSS version 13 (SPSS Inc, Chicago, IL).