Ultrasound Image Restoration Algorithm Based On Natural Image Priors

Compared with other medical imaging systems,medical ultrasound imaging system has the advantage of no radiation,low cost,high sensitivity and portability.However,the ultrasound images are always of low resolution and low contrast and contain speckle noise,which greatly limits their clinical usage.In this thesis,the blind restoration algorithm proposed by Krishnan is applied to medical ultrasound images restoration.By changing the way of initializing PSF and the iterative model in Krishnan's algorithm,an ultrasound image restoration algorithm based on natural image prior is formed.Results show that the improved algorithm works better.The main works are as follows:Firstly,the research significance and objectives are introduced.Meanwhile,the research status of some restoration algorithms of the ultrasound images and regularization algorithms at home and abroad are introduced.Secondly,the some principles and the some theories,which are involved with the ultrasound imaging system,image restoration and regularization,and the image quality index,and ultrasound simulation program Field II,are introduced.The analysis shows that the gray gradient of the ultrasound image is also consistent with the heavy-tailed distribution of natural images,so it is feasible to implement the algorithm proposed by Krishnan on the ultrasound images.The model about the image degraded by the spatially-variant PSF is established,also the important parameters that affect the performance of the algorithm are analyzed and are optimized.Krishnan's algorithm is improved.Firstly,an ultrasound image is divided into the appropriate numbers of blocks by depth because of the spatial shift of PSF.Then,the way of kernel initialization in Krishnan's algorithm is changed,and the pyramid iterative way is replaced by the iterative one between the image blocks for the similarity of adjacent PSFs at different depths in the ultrasound image.Therefore,the ultrasound image restoration algorithm based on the natural image prior is obtained.Finally,we conduct the experiment on both the simulated and real ultrasound images.It is shown that the proposed algorithm is capable of improving the resolution gain of the simulated images by 38.5%,and that of the real ultrasound images by 100%,and the processing speed by 56%.