Selection of wavelet functions and thresholding parameteres in ultrasound image denoising
G. Andria, Filippo Attivissimo, Graziana Cavone, Anna Maria Lucia Lanzolla · 2013
The paper compares the filtering performance in ultrasound images wavelet denoising as function of wavelet functions and thresholding parameters. Wavelet filtering was performed by suing a new parametric thresholding based on an adaptive data-driven exponential operator. Then, the effects of wavelet mother properties on image filtering have been analyzed. Experimental tests have been carried out using ultrasound images simulated by Field II software. Different noise levels and ultrasound frequencies have been set in the image simulation to obtain a wide case of study. Image quality has been evaluated using two different metrics which take into account the classical peak signal-to-noise ratio and the edges preservation in the filtered image. Based on the results, a proper selection of both mother wavelet and threshold parameter is proposed to increase medical image quality.