A novel speckle suppression method for medical ultrasound images is presented. First, the logarithmic transform of the original image is analyzed into the multiscale wavelet domain. We show that the subband decompositions of ultrasound images have significantly non-Gaussian statistics that are best described by families of heavy-tailed distributions such as the alpha-stable. Then, we design a Bayesian estimator that exploits these statistics. We use the alpha-stable model to develop a blind noise-removal processor that performs a nonlinear operation on the data. Finally, we compare our technique with current state-of-the-art soft and hard thresholding methods applied on actual ultrasound medical images and we quantify the achieved performance improvement.

Synthetic aperture radar (SAR) images are inherently affected by multiplicative speckle noise, which is due to the coherent nature of the scattering phenomenon. This paper proposes a novel Bayesian-based algorithm within the framework of wavelet analysis, which reduces speckle in SAR images while preserving the structural features and textural information of the scene. First,

Speckle noise is an inherent property of medical ultrasound imaging, and it generally tends to reduce the image resolution and contrast, thereby reducing the diagnostic value of this imaging modality. As a result, the speckle reduction is considered to be a prerequisite procedure to be used, whenever ultrasound images are used for tissue characterization. Among the method, which have been proposed so far to perform this task, there exits a class of approaches that employ the multiplicative model of speckled image formation and take advantage of the logarithmical transformation in order to convert the multiplicative speckle noise to an additive noise. The present study shows conceptually and experimentally that assuming the latter to be a white Gaussian noise – as it is done in a dominant number of cases – is an oversimplified and unsafe assumption that leads to inadequate performance of the filtering methods used to suppress the noise. The study introduces a simple preprocessing procedure, which modifies the acquired radio-frequency images (without altering the anatomical information they bear), so that the noise in the domain of the logarithmic transformation becomes very close in its behavior to white Gaussian noise. It allows the filtering methods based on assuming the noise to be white and Gaussian, to perform in nearly optimal conditions. The study evaluates performances of three different non-linear filters – wavelet de-noising, total variation filtering, and anisotropic diffusion – and demonstrates that in all these cases, the “modified ” processing results in a dramatic improvement in the quality of resultant images. The test included a series of computer-simulated and in vivo experiments.

Abstract. This paper presents a 3D echocardiographic image segmentation procedure based on deformable surfaces. We rst propose to adapt ltering techniques to the cylindrical geometry of several 3D ultrasound image devices. Then we compare the e ect of di erent external forces on a surface template deformation inside volumetric echocardiographic images. An original method involving region grey-level analysis along the model normal directions is described. We relyonana priori knowledge of the cardiac left ventricle shape and on region grey-level values to perform a robust segmentation. During the deformation process the allowable surface deformation is modi ed. Finally, weshowexperimen tal results on very challenging sparse and noisy images and quantitative measurements of the left ventricle volume. 1

This paper presents a 4D (3D + time) echocardiographic image anisotropic filtering and a 3D model-based segmentation system. To improve the extraction of left ventricle boundaries, we rely on two preprocessing stages. First, we apply an anisotropic filter that reduces image noise. This 4D filter takes into account the spatial and temporal nature of echocardiographic images. Second, we adapt the usual gradient filter estimation to the cylindrical geometry of the 3D ultrasound images. The reconstruction of the endocardium takes place by deforming a deformable simplex mesh having an a priori knowledge of left ventricle shape and that is guided by a region based data attraction force. The external force formulation improves the segmentation robustness against noise and outliers. We illustrate our method by showing experimental results on very challenging sparse and noisy ultrasound images of the heart and by computing quantitative measurements of the left ventricle volume.

Segmentation remains one of the main problem of image analysis. In particular, 3D medical image segmentation is difficult due to noise, low contrast and outliers resulting from any 3D imaging technologies. However, image segmentation is a first step required for quantitative measurements, automatic diagnosis and modelling of anatomical structures in medical images. Due to the complex shape of anatomical structures and the large inter-patient shape variability, powerful modelling tools are required. In this report, we describe a segmentation tool based on deformable surfaces well suited to the shape reconstruction and modelling of anatomical structures from 3D medical images. The deformable surfaces are based on discrete meshes that can represent manifolds without any topology restrictions. These meshes rely on shape constraints at local and global scale for introducing prior knowledge on the structures to segment and for regularizing the surface deformations. Dioeerent data terms are pr...

Abstract — Speckle noise removal by means of digital image processors could improve the diagnostic potential of medical ultrasound. This paper addresses the speckle suppression issue within the framework of wavelet analysis. As a first step of our approach, the logarithm of the original image is decomposed into several scales through a multiresolution analysis employing the 2-D wavelet transform. Then, we design a maximum a posteriori (MAP) estimator, which relies on a recently introduced statistical representation for the wavelet coefficients of ultrasound images [1]. We use an alpha-stable model to develop a blind noise-removal processor that performs a non-linear operation on the data. Finally, we compare our technique to current state-of-the-art denoising methods applied on actual ultrasound images and we find it more effective, both in terms of speckle reduction and signal detail preservation.