A new approach to artefact detection and its effect on the normal range of segmental strain and strain rate values

• Published in: European Heart Journal - Cardiovascular Imaging
• Main Authors: Kornev, M, Rosner, A, Caglayan, HA
• Format: Article in Journal/Newspaper
• Language: English
• Published: Oxford University Press (OUP) 2022
• Subjects: Cardiology and Cardiovascular Medicine; Radiology, Nuclear Medicine and imaging; General Medicine; Tromso
• Online Access: http://dx.doi.org/10.1093/ehjci/jeab289.046; https://academic.oup.com/ehjcimaging/article-pdf/23/Supplement_1/jeab289.046/42395277/jeab289.046.pdf

Abstract Funding Acknowledgements Type of funding sources: Public grant(s) – National budget only. Main funding source(s): Helse Nord HNF1458-19 Background Definition of normal ranges of myocardial segmental deformation are important in clinical studies and routine echocardiographic practice. It is known that artefacts hamper segmental strain and strain rate (S/SR) analysis. Purpose This study is based on a new approach for artefact detection. We intended to investigate to what degree different types of artefacts and their localization bias the result of measurements and to establish corrected normal ranges for segmental S/SR parameters. Methods The study is based on a collaboration project between the Russian cross-sectional study "Know your heart" and the Norwegian "Tromso-7" study. From 2207 participants we identified 840 individuals (511 females and 329 males, age range 40-79) with normal cardiac function by excluding left ventricular ejection fraction <50%, moderate or severe valvular heart disease, LV- dilatation, pulmonary hypertension, Q-waves and bundle branch block signs on ECG, high blood pressure, history of heart attack, use of antihypertensive drugs or high level of NT-proBNP. End-systolic longitudinal epi- myo- and endocardial strain and peak systolic SR and diastolic SR E and SR A were assessed for the calculation of segmental normal ranges. In addition to noise and foreshortening artefacts, we introduced a concept of "curve artefact" which can be defined as unphysiological strain curve shapes. Panel A of the figure shows examples of these curve artefacts, A: diastolic mismatch B: blunted curve, and C: floating curve. All strain curves were assessed for the presence of one of these unphysiological shapes and 2D images were visually assessed for the presence of noise, reverberations, or missing parts of the myocardium. Results The bar-chart for segmental strain demonstrates significantly reduced strain values at the presence of curve artefacts and increased strain at the presence of foreshortening. Normal ranges for basal septal, basal, medial, and apical segments excluding segments with artefacts and foreshortening are displayed in the table. The presence of curve artefacts matched the presence of noise or other 2D image artefacts in 88% of cases. Conclusions Strain artefacts result in systematically reduced peak-strain values. Thus, the detection and exclusion of curve artefacts seem to be an important measure for the correct interpretation of strain curves for a definition of normalcy and pathology. The selection of artefact-free strain curves reveals high strain and SR gradients increasing from epi to endocardial position and basal-septal towards apical segments. Abstract Figure. normal range Abstract Figure. artefacts