Research On Speckle Reduction In Ophthalmology High Frequency Ultrasound Imaging

Medical ultrasound imaging is a kind of safe and non-invasive, real-efficient, low-cost way of medical imaging,and it has been widely used in clinical. With the development of ultrasound technology, ophthalmic ultrasonography has become a unique branch. Among them, the UBM (Ultrasound Biomicroscopy, UBM) for anterior segment imaging is superior to conventional ophthalmic diagnostic instruments. But there is a special kind of noise in ultrasound images, that is speckle noise, which is generated during an ultrasound image associated. Speckle noise is different from the image noise, and it belongs to the multiplicative noise, and a large part belong to the high frequency part. Speckle noise would undermine the image contrast and detail resolution, thus affecting the correct clinical judgment. Medical image is different from the general picture, denoising medical image not only needs to remove the noise, but also to keep the details and edge of the image as much as possible. For these requirements, this paper presents two speckle noise removal algorithms.First, this paper analyzes the type and characteristics of noise exists in ultrasound image and the mathematical model of speckle noise,and we compare the advantages and disadvantages of the traditional methods of suppressing speckle, exploring denoising algorithms for speckle noise. Then on the basis of Speckle Reducing Anisotropic diffusion(SRAD), we propose an improved method of SRAD.diffusion coefficient was modified to improve the robustness of the algorithm, and the number of neighbor point of spread template was increased to improve the ability of detail protection during the process of de-noising. Except a single level of image de-noising, then we proposed a de-noising algorithm based on Laplacian pyramid. The image was decomposed to Laplacian pyramid to obtain sub-band images with different frequencies, and then we select the appropriate denoising algorithm according to the different characteristics of the sub-band image, finally, the sub-band images were reconstructed to give an enhanced and de-noising image.Through experiments, the improved speckle suppression anisotropic diffusion method proposed in this paper as well as the hierarchical suppression algorithm based on Laplacian pyramid better than traditional filtering algorithm,the two algorithm can not only well remove speckle noise in ultrasound images, but also can effectively preserve image details and edge.