Abstract :
Anisotropic diffusion has been widely used to reduce speckle noise from ultrasound images. However, in traditional anisotropic diffusion methods, usually four directions neighborhood are used, which have many disadvantages, such as the sensitivity to the noise, the loss of image details and false contours. In this paper, we modify the standard SRAD (speckle reducing anisotropic diffusion) algorithm in the discrete domain by considering larger neighborhoods in its computation; then we adopt a confine term to keep the closest equivalence between the filtered image and the original image. Experimental results show that, in the presence of speckle nose, this proposed method could effectively preserve edges and detailed structures while suppressing speckle noise These preliminary results indicate that the proposed speckle reduction method could improve image quality and the visibility of small structures and fine details in the medical ultrasound images.
