Proceedings of The Physiological Society

Europhysiology 2018 (London, UK) (2018) Proc Physiol Soc 41, PCA033

Poster Communications

The Impact of Intentional Distortion of Image Quality on Left Ventricular Deformation Indices by Three-dimensional Speckle-Tracking Echocardiography

L. K. Al Saikhan1,2, C. Park1, A. Hughes1

1. Institute of Cardiovascular Science, School of Life and Medical Sciences, University College London, London, London, United Kingdom. 2. Department of Cardiac Technology, College of Applied Medial Sciences, Imam Abdulrahman Bin Faisal University, Dammam, Saudi Arabia.


  • CI, confidence interval; CS, circumferential strain; EDV, end-diastolic volume; EF, ejection fraction; ESV, end-systolic volume; GCS, global circumferential strain; GLS, global longitudinal strain; ICC, intra-class correlation coefficient ;LS, longitudinal strain; LV, left ventricular; mean Δ, mean difference; PTS, principle tangential strain; RS, radial strain ; SV, stroke volume. *Averaged based on 16 segments model. §ICC < 0.4 = poor reproducibility, 0.4 ≥ ICC < 0.75 = fair to good reproducibility, and ICC ≥ 0.75 = excellent reproducibility. †ICC represents data from all 4 scans per participant (i.e. good and sub-optimal quality 3DE data sets of the first and second scans). ‡ICC after restricting the analysis to undistorted image quality (i.e. good quality images; Good1 = the first scan and Good2 = the second scan). p < 0.05 was considered statistically significant.

Background: Three-dimensional speckle-tracking echocardiography (3D-STE) is a novel technique that provides a comprehensive quantification of regional and global left ventricular (LV) myocardial deformation. It is increasingly used in clinical practice. 3D-STE is believed to be influenced by image quality, although quantitative evidence on this is limited. Purpose: To assess the impact of intentionally distorted image quality on LV deformation indices measured by 3D-STE. Methods: 18 healthy volunteers (age 28 ± 6 years, 55.5% men) underwent three-dimensional echocardiography (3DE). Participants with poor images were excluded. One optimal and one intentionally impaired 3DE full-volume datasets of the LV were obtained and analysed per participant. Quality was impaired by including images with echo drop out, shadow artefacts, or poor visualization of the endocardium. Both acquisitions were free of stitching artefacts and the frame rate was maintained similar. Obtained 3D-STE parameters were LV volumetric, and global and averaged segmental LV deformation measures. The acquisition protocol was repeated on the same day to assess the test-retest repeatability. Mixed linear modelling was used to estimate the extent of bias and the intraclass correlation coefficient (ICC) was used as a measure of reliability. Results: Sub-optimal image quality resulted in a systematic bias in all global and averaged segmental LV strain measures and LV rotational measures (Table 1). LV ejection fraction and most of LV volumes by 3D-STE were also consistently underestimated. Reproducibility was fair to good for all global and averaged segmental LV strain measures (ICC 0.47 - 0.72) and poor for twist and torsion (ICC 0.39 and 0.33 respectively). LV volumes and ejection fraction showed excellent reproducibility irrespective of image quality (ICC 0.78 - 0.92). Reproducibility was improved for all measures when restricting the analysis to undistorted quality images [i.e. good image quality] (Table 1). Conclusions: Sub-optimal image quality adversely affects the reproducibility of 3D-STE derived measures and introduces a moderate systematic bias (<10%). Image quality should be assessed and accounted for in 3DSTE studies.

Where applicable, experiments conform with Society ethical requirements