Comparative analysis of despeckling filters for continuous wave Doppler images

Purpose: The problems of speckle noise, low contrast, and other artifacts in continuous wave Doppler (CWD) based echocardiographic images are identical to the issues prevailing in the B-mode images. The median filter and Gaussian low pass filter are more commonly used for reduction of noise in CWD images. The applications of state-of-the-art despeckling filters and their impact on delineation are not extensively analyzed using CW Doppler images.Methods: This paper analyses the noise reduction, edge and structure preservation characteristics of ten filters for CWD images in terms of 15 image quality metrics along with visual quality assessment. Further, three segmentation techniques were analyzed using original and noisy CWD images. Three existing delineation techniques are modified for tracing the outer boundaries of CWD spectrum. The Gaussian filter used reaction-diffusion based active contour and thresholding based segmentation techniques are replaced by despeckling filter based on anisotropic diffusion and local statistics.Results: The noise is suppressed with edges and structure well preserved on application of filters such as speckle reducing anisotropic diffusion filter and local statics (mean and variance) based filter. The number of iterations required in complete boundary tracing is reduced on replacing the Gaussian filter with despeckling filter in reaction diffusion based active contour segmentation.Conclusions: The analysis reveals that the despeckling techniques enhance the visual quality of images and play a vital role in delineation process. The reaction diffusion based active contour embedded with speckle reducing anisotropic diffusion filter results in accurate contouring of the CWD images. © 2015, Korean Society of Medical and Biological Engineering and Springer.