Denoising of Ultrasound Cervix Image Using Improved Anisotropic Diffusion Filter.

Abstract

OBJECTIVE The purpose of this study was to evaluate an improved oriented speckle reducing anisotropic diffusion (IADF) filter that suppress the speckle noise from ultrasound B-mode images and shows better result than previous filters such as anisotropic diffusion, wavelet denoising and local statistics. METHODS The clinical ultrasound images of the cervix were obtained by ATL HDI 5000 ultrasound machine from the Regional Cancer Centre, Medical College campus, Thiruvananthapuram. The standardized ways of organizing and storing the image were in the format of bmp and the dimensions of 256 x 256 with the help of an improved oriented speckle reducing anisotropic diffusion filter. For analysis, 24 ultrasound cervix images were tested and the performance measured. RESULTS This provides the quality metrics in case of maximum peak signal-to-noise ratio (PSNR) of 31 dB, structural similarity index map (SSIM) of 0.88 and edge preservation accuracy of 88%. CONCLUSIONS The IADF filter is the optimal method and it is capable of strong speckle suppression with less computational complexity.