Abstract:
The major problem of ultrasound imaging technique is inheritance of Speckle noise. Speckle is
a multiplicative noise that reduces both image contrast and detail resolution, degrades tissue
texture, reduces the visibility of small low-contrast lesions and makes continuous structures
appear discontinuous, thereby decreasing the quality and reliability of medical ultrasound. It
also limits the effective application (e.g. edge detection) of automated computer analysis. As a
result, image processing methods for restoration or reduction of speckle noise from
ultrasound images has become the predominant step in medical image processing. In this
study four de-noising technique such as mean filter, median filter, Gaussian filter and Wiener
filter have been developed for de-speckling of ultrasound kidney stone images. Then, the
techniques applied were assessed by measuring the image quality (by calculating the MSE and
PSNR of the image) and also by applying a level set image segmentation techniques to the
output images to know which types of filter produce best quality images. In terms of PSNR and
MSE results Wiener filtering eliminates more noise compared to other enhancement
techniques and outperformed mean filter, median filter and Gaussian filter by 40.54%,
49.36% and 58.5% respectively on average. On the other hand based on level set
segmentation method of the output filtered image, it was also observed that the segmentation
of the edge of kidney stone of Wiener filtered achieved low FP area ratio and best SI rates
than all other filter.