J. W. Trobaugh and R. M. Arthur, "Estimation of Surface Pose With a Physically-Based Ultrasonic Image Model" IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control, vol. 51, pp. 1128-1136, 2004.

Abstract

State-of-the-art approaches to shape analysis in medical images use a variety of sophisticated models for object shape.  We have developed an image model that permits the application of these approaches to ultrasonic images, with detailed methods for representing rough surfaces.  Our image model is probabilistic and based on a physical model for image formation that incorporates the system point-spread function (PSF), the tissue microstructure and the gross tissue shape.  At each image pixel, the amplitude mean and variance is computed, characterizing the combined influence of shape, microstructure and system PSF.  Calculation of the SNR₀ permits further classification of each pixel as Rayleigh- or non-Rayleigh-distributed.  This characterization was used here to generate a Rayleigh/Gaussian data likelihood for any set of image data, conditioned on surface shape and pose. Maximum likelihood pose estimation was performed using derivative-based optimization algorithms for images of a cadaveric vertebra.  Successful results were achieved in the simulation environment for a data set of only three images.  With a quasi-Newton BFGS algorithm, error in 15 of 20 trials was less than 0.4 degrees in rotation and 0.2 mm in translation.  Estimation was inaccurate in only 1 of 20 trials.  These results illustrate the potential of the model and the approach and serve as an example of quantitative assessment of the model via performance in a specific application.

Acknowledgement: Supported in part by NIH grant R21-CA90531 from the National Cancer Institute and the Wilkinson Trust at Washington University in St. Louis.