Despeckling Ultrasound B-Mode Images Denoised by a New Unsharp Masking Method for Improved Cyst Segmentation

Abstract

A denoising technique known as new unsharp masking (UM) was recently introduced in ultrasound B-mode imaging. The main objective of this technique was to reduce clutter noise and improve the spatial resolution of ultrasound B-mode images produced with compound plane wave imaging (CPWI). Speckle noise is a granular pattern in ultrasound images that reduces the detectability of a small anechoic structure, degrading the image contrast and resolution. The UM algorithm increases speckle fluctuation which could be detrimental to post-processing such as cyst segmentation. Despeckling plays an important role in ultrasound B-mode imaging to reduce the presence of speckle. No study has been carried out on the new UM method when despeckling techniques are combined to it for cyst segmentation. In this work, to evaluate the despeckling effects on ultrasound B-mode images denoised by using new UM, 2-D median filtering with different kernel sizes were adopted. Balloon snake active contours (BSAC) based segmentation was then performed. The results show that as the despeckling filter kernel size is increasing, the convergence of the snake towards the cyst border improves. Moreover, the reduction of speckle and clutter noise facilitates the snake to achieve its minimum energy within 100 iterations.