Research On Medical Image Denoising

In medical field,medical image processing technology is increasingly used in treatment planning and disease diagnosis.Magnetic resonance imaging(MRI),computed tomography(CT),positron emission tomography(PET),X-ray,and ultrasound are mainly used currently.Although medical imaging equipment and image acquisition equipment are very advanced,they are sometimes affected by some external factors.Noises occurring in medical image processing,may directly affect the quality of medical images and affect the quality of professional judgement.Therefore,it is very necessary to analyze and study the denoising technology in medical image processing.Denoising is the precondition for image processing and an indispensable step in image preprocessing.At present,the denoising technique for natural image preprocessing has achieved significant preformance,but there are still some problems in the research of denoising in medical image preprocessing.In order to improve quality of medical images,the effect of noise on the images is reduced.This thesis analyses and researches the denoising technology of the medical image to a deeper extent as follows:1.The thesis analyses the importance of medical image denoising for disease diagnosis and medical research,and describes the current stage of medical image denoising technology,compares the existing denoising technology methods,which provides basis for improving denoising algorithms functioning.2.In order to overcome the shortcomings of the traditional NLM algorithm,an NLM denoising algorithm based on SLIC superpixel segmentation is proposed.First,the superpixel-slicing SLIC algorithm will generate tags,and each pixel corresponds to the relatived tag.Similar superpixel tags share the same label value,which can be used to determine the location of pixels,and finally the NLM algorithm will be used to denoise.By comparing the effects of denoising through experiments,it is concluded that the algorithm makes the neighborhood search more consistent,especially at the edges.3.Based on sparse decomposition,this thesis first selected the K-SVD learning dictionary to denoise the medical image,compared to fixed dictionary.The learning dictionary contains the characteristics of information in the image,especially adaptive dictionary which can be obtained by training,so it is more adaptable to noises images,hence the denoising will achieve a better preformance.Secondly,this thesis studies the global dictionary based K-SVD and adaptive dictionary based K-SVD for the medical image processing,and compares it with other denoising methods,finally,we evaluates the denoising performance of the global dictionary based K-SVD and adaptive dictionary based K-SVD.