A new Markov random field model based on κ-distribution for textured ultrasound image

The aim of this paper is to propose a new Markov Random Field (MRF) for textured ultrasound image which use is more relevant than the use of the classic MRF, such as the gaussian markovian model. The main difference is that our model is based on κ-distribution. We have built this κ-MRF with reference to the Product Model. This latter means that the observed intensities of ultrasound image are the product of a degraded perfect image by a multiplicative noise called speckle. When the construct of κ-MRF is already described, we propose in this paper a validation on synthetic and medical B-scan textured image. The synthetic textures are obtained by stimulating the κ-MRF. For medical texture, we estimate the parameters of the model from tissues. The estimated parameters are simulated and compared to medical texture. The resemblance is a first validation of the κ-MRF and the tissue can be then characterized by the parameters of the model.

[1]  B. Goldberg,et al.  Comparisons of the Rayleigh and K-distribution models using in vivo breast and liver tissue. , 1998, Ultrasound in medicine & biology.

[2]  Michael Brady,et al.  Segmentation of ultrasound B-mode images with intensity inhomogeneity correction , 2002, IEEE Transactions on Medical Imaging.

[3]  J. Besag.  Spatial Interaction and the Statistical Analysis of Lattice Systems , 1974 .

[4]  Andrew H. Gee,et al.  Decompression and speckle detection for ultrasound images using the homodyned k-distribution , 2003, Pattern Recognit. Lett..

[5]  E. Jakeman,et al.  Generalized K distribution: a statistical model for weak scattering , 1987 .

[6]  J. Goodman.  Some fundamental properties of speckle , 1976 .

[7]  P. Shankar.  A general statistical model for ultrasonic backscattering from tissues , 2000 .

[8]  M.R. Boussema,et al.  Heterogeneous SAR texture characterization by means of Markov random fields , 2000, IGARSS 2000. IEEE 2000 International Geoscience and Remote Sensing Symposium. Taking the Pulse of the Planet: The Role of Remote Sensing in Managing the Environment. Proceedings (Cat. No.00CH37120).

[9]  H Ermert,et al.  Segmentation of 3D intravascular ultrasonic images based on a random field model. , 2000, Ultrasound in medicine & biology.

[10]  Peter J. W. Rayner,et al.  Unsupervised image segmentation using Markov random field models , 1997, Pattern Recognit..

[11]  P. Mohana Shankar,et al.  A general statistical model for ultrasonic backscattering from tissues , 2000, IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control.