PII S0301-5629(98)00060-X ● Original Contributions BEAT AVERAGING ALTERNATIVES FOR TRANSMITRAL DOPPLER FLOW VELOCITY IMAGES ANDREW F. HALL,SCOTT P. NUDELMAN and S´ ANDOR J. KOV ´ ACS Cardiovascular Biophysics Laboratory, Cardiovascular Division, Washington University School of Medicine, 216 South Kingshighway Blvd., St. Louis, MO 63110, USA (Received 22 December 1997; in final form 7 April 1998) Abstract—To characterize diastolic function from transmitral Doppler data, the image’s maximum velocity envelope (MVE) is fit by a model for flow velocity. To reduce the physiologic beat-to-beat variability of best-fit determined model parameters, averaging of multiple cardiac cycles is indicated. To assess variability mathe- matically, we modeled physiologic noise as a random (normally-distributed) process and evaluated three methods of averaging (1, averaging model parameters from single images; 2, averaging images; and 3, averaging MVEs) using clinical datasets (50 continuous beats from 5 subjects). Method 2 generates a positive bias because low-velocity beats will not contribute to the composite MVE. The difference between Methods 3 and 1 is less than 2.0 E–5 (m/s) 2 for uncorrelated model parameters. Input having 10% beat-to-beat variation yields a bias of <4% for model parameter mean. Hence, Method 1 was, in general, more robust than Method 3. © 1998 World Federation for Ultrasound in Medicine & Biology. Key Words: Diastolic function, Left ventricle, Doppler echocardiography, Mathematical models, Image processing. INTRODUCTION Analysis of echocardiographic Doppler velocity profile (DVP) images of transmitral blood flow continues to be a valuable tool for the assessment of diastolic function. New models of ventricular filling are being proposed (Sun et al. 1995), as well as newer methods for image capture and analysis (Miro et al. 1992). We have previ- ously developed a model for transmitral flow that de- scribes the entire diastolic filling interval (E- and A- waves) (Kova ´cs et al. 1987). This model, which predicts flow velocity as a function of time, accounts for the suction-pump role of the heart and is based on the motion of a damped, harmonic oscillator. For the E-wave, the model parameters to be determined from the Doppler image are: the damping constant ( c ), the spring constant ( k ), and the initial mass displacement ( x 0 ). We have also developed an automated model-based image-processing method for extracting model parameters from DVP im- ages (Hall and Kova ´cs 1994). This method, shown in Fig. 1a–f for the Doppler E-wave (early filling), pro- cesses single DVP digitized images stored on a com- puter. The maximum velocity envelope (MVE), which contains noise, is extracted from the image. The model- predicted solution for velocity is fit to the MVE, resulting in model parameter estimates and estimates for the stan- dard deviation (SD) of each parameter. When fitting the model solution for the E-wave to the MVE, one must decide which MVE data points should be included in the fitting process (i.e., where, in time, do the data start, and where do they end?) Previous work has shown that the model parameter estimates are sensitive to the starting point of the E-wave in the MVE data set (Hall and Kova ´cs 1993, 1994). An example is shown in Fig. 2a– d. The reasons why the parameters are sensitive to the starting point are: 1. At the instant of mitral valve open- ing and E-wave initiation, there may be significant noise in the DVP image due to initially (low velocity) turbulent blood flow though the mitral valve; 2. the kinematic model approximates the rapidly rising atrioventricular pressure gradient as an instantaneous nonzero pressure gradient at the instant of mitral valve opening. Two methods have been devised to overcome the starting point sensitivity problem. In the first method, the data set starting point is treated as a variable. Parameter estimates (and estimated standard deviations) are computed for a series of E-wave starting points, as shown in Fig. 3a– c. The set of model Address correspondence to: Dr. Sa ´ndor J. Kova ´cs, Cardiovascu- lar Biophysics Laboratory, Cardiovascular Division Barnes-Jewish Hospital at Washington University Medical Center, 216 South King- shighway Blvd. St. Louis, MO. 63110. E-mail: sjk@howdy.wustl.edu Ultrasound in Med. & Biol., Vol. 24, No. 7, pp. 971–979, 1998 Copyright © 1998 World Federation for Ultrasound in Medicine & Biology Printed in the USA. All rights reserved 0301-5629/98 $19.00 + .00 971