Speckle Reduction of Ultrasound Image via Morphological Based Edge Preserving Weighted Mean Filter

Abstract

Medical ultrasound images suffer from inherently generated speckle noise that makes difficult the radiologists to diagnose the diseases. Proper speckle reduction technique is required to improve quality of image which may help the doctor to diagnose the diseases correctly. Speckle removal is, specially, a filtering technique that reduces the amount of speckle noise. But as an effect of filtering, the edges of objects may become blur and fine details within the image may be lost. Many methods have already been proposed to achieve these two requirements. But they do not satisfy both up to the desirable level. In this paper, we propose a novel morphological operation based speckle filter which reduces the effect of speckle noise to the greater extent in one side and keeps detail information in other side. The use of morphological operators helps in extracting the structures present within the images followed by an edge preserving weighted mean filtering method that may help in despeckling of ultrasound images. The proposed technique is applied on the ultrasound simulated data as well as on the real data obtained from the ultrasound machine. The quantitative results and the output images confirm the superiority of the method when compared with some of the other contemporary speckle reduction methods.