Image Quality Causes Substantial Bias in Three-Dimensional Speckle-Tracking Echocardiography Measures

BACKGROUNDThree-dimensional speckle-tracking echocardiography (3D-STE) is increasingly used to assess left ventricular (LV) mechanics but the quantitative effect of image quality on measurements is not known.\n\nOBJECTIVESTo assess the impact of image quality on 3D-STE derived LV indices.\n\nMETHODSEchocardiography was performed in two groups of 18 healthy participants. In the first study, optimal and intentionally poor-quality images were acquired. In the second study, a sheet of ultrasound-attenuating material (neoprene rubber) of three different thicknesses (2, 3 and 4 mm) was used to mimic mildly, moderately and severely impaired image quality respectively.\n\nRESULTSIn both studies sub-optimal image quality resulted in a systematic underestimation bias in all LV deformation and rotational indices. LV ejection fraction and volumes were also consistently underestimated. The extent of the bias was proportional to the impairment in image quality (i.e. the poorer the image quality the larger the bias). Reproducibility was also less good for sub-optimal images, although LV volumes and ejection fraction showed excellent reproducibility irrespective of image quality.\n\nCONCLUSIONSSub-optimal image quality introduces a substantial systematic bias and impairs the reproducibility of 3D-STE. Bias related to image quality might have important clinical implications since its magnitude is similar to that reported in association with disease and may confound associations between disease and LV mechanics.\n\nCONDENSED ABSTRACTThree-dimensional speckle-tracking echocardiography (3D-STE) is increasingly used in clinical practice. We assessed the impact of sub-optimal image quality on left ventricular (LV) deformation indices either by poor echocardiographic technique or by impairing ultrasound propagation using a neoprene sheet. Sub-optimal image quality introduces a substantial systematic bias that increases with poorer image quality. Bias due to image quality might have important clinical implications since its magnitude is similar to that reported in association with disease.