Speckle reduction and contrast enhancement of ultrasound images using anisotropic diffusion with Jensen Shannon divergence operator

• Published in: Biomedical engineering letters Vol. 3; no. 2; pp. 87 - 94
• Main Authors: Jose, Jisha, Prahladan, Anil, Nair, Madhu S.
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01-06-2013; 대한의용생체공학회
• Subjects: Biological and Medical Physics; Biomedical Engineering and Bioengineering; Biomedicine; Biophysics; Engineering; Medical and Radiation Physics; Original Article; 의공학; Edge and contrast enhancement; Diffusion and Jensen Shannon Divergence (DJ); Ultrasound (US) imaging; Speckle reduction; Anisotropic diffusion; Fuzzy sub pixel fractional partial difference
• ISSN: 2093-9868; 2093-985X
• DOI: 10.1007/s13534-013-0091-1

Purpose
Ultrasound (US) imaging is widely used for diagnosis these days due to its advantages such as no-radiation, portability and low cost. But the two inherent drawbacks of US imaging are low contrast and speckle noise. The objective of the proposed algorithm is to simultaneously enhance the contrast and suppress the speckle noise thereby improving the visual quality of the US image.
Methods
Normalization and fuzzification are used as the preprocessing steps for the contrast enhancement and the mapping of intensity values to brightness degree of US images. A fuzzy sub-pixel fractional partial difference with Jensen Shannon divergence operator is proposed here for determining whether a pixel is noise or edge, and thereby reducing the speckle noise. The proposed method is a generalization of first, second and fourth order difference with weight information collected from the overall view of the image.
Results
Proposed method has been tested both on synthetic and real ultrasound images. The results show that the proposed method improves preservation of relevant information without compromising the quality of visual appearance. FOM and Contrast measures prove the superiority of the proposed method over conventional as well as advanced methods.
Conclusions
The experimental results demonstrate that the proposed DJ method has the advantage of maximizing speckle reduction and contrast enhancement, with great accuracy for slightly varying edges and the fine details are well preserved. The method can improve the quality of US images, and will be useful for CAD systems for cancer detection and classification based on US images.