Reproducibility and interpretation in tissue Doppler echocardiography

Abstract: As cardiovascular disease is the single most common cause ofdeath in the western world, and since there is a closeconnection between cardiovascular disease and left ventricular(LV) function, good methods for the assessment of LV functionis highly needed. A widely used tool for the diagnosis of LVdisease is echocardiography, a technique which today faces twodi_culties; the low reproducibility and the subjectiveinterpretation. The present dissertation aims to quantifyreproducibility, to study the factors that influencereproducibility and to provide tools for simplifying theinterpretation of tissue velocities measured by Dopplerultrasound.The reproducibility has been studied by letting twoindependent observers measure tissue Doppler velocities toinvestigate how well their measurements agree. To improvereproducibility an algorithm for the automatic detection ofpre-defined echocardiographic localizations is presented. Oneof the most difficult skills for the sonographer to master,thus leading to reduced reproducibility, is the transducermanipulation. The effect of poor transducer manipulation hasbeen modeled, and we show that even a poorly placed transducermay yield images which are easily mistaken for good, however,when scanning in two orthogonal planes the transducermisplacement is easily detected.Interpretation of the echocardiograms is influenced byseveral parameters. As the tissue velocities are measured byutilizing the Doppler effect, only the velocity componentdirected towards the transducer can be measured, thus thealignment of the heart within the view of the transduceraffects the tissue velocity measurements. The effect of thishas been investigated, and it is demonstrated that since themyocardium primarily has longitudinal motion and thus thevelocity vectors are mainly longitudinal, imaging in the apicalview will give little error in the velocity measurements.Filtering of the tissue velocity signals have becomecommercially available with the hope that it will improvereproducibility and simplify interpretation. One set of lowpass filters has been tested, and it is seen that there is arisk of overdoing the filtering and cause an underestimation oftissue velocity parameters. A similar effect to low passfiltering is seen when using too low sample rate when recordingthe tissue velocities.Finally a new imaging modality, tissue motion imaging, ispresented, where myocardial displacement, velocity, strain andacceleration may be interpreted from one single image, insteadof the situation today where several measurements must beperformed to get an overview of all these parameters.The thesis concludes that reproducibility can be improved bycurve smoothing and that interpretation can be simplified usingadvanced methods of parametric imaging.