Markov models of specular and diffuse scattering in restoration ofmedical ultrasound images

摘要

Observed medical ultrasound images are degraded representations ofnthe true tissue reflectance. The specular reflections at boundariesnbetween regions of different tissue types are blurred, and the diffusenscattering within such regions also contains speckle. This reduces thendiagnostic value of such images. In order to remove both blur andnspeckle, the authors develop a maximum a posteriori deconvolutionnalgorithm for two-dimensional (2-D) ultrasound radio frequency (RF)nimages based on a new Markov random field image model incorporatingnspatial smoothness constraints and physical models for specularnreflections and diffuse scattering. During stochastic relaxation, thenalgorithm alternates steps of restoration and segmentation, and includesnestimation of reflectance parameters. The smoothness constraintsnregularize the overall procedure, and the algorithm uses the specularnreflection model to locate region boundaries. The resulting restorationsnof some simulated and real RF images are significantly better than thosenproduced by Wiener filtering