Study On Medical Image Reconstruction Algorithm Based Onadaptive Sparse Representation

Currently, the medical image processing technology has been widely used in clinical diagnosis with the rapid development of image processing technology. So it has become an important basis for medical diagnosis. But it will have some radiation to the patient during the measure data collected. How to achieve quick and accurate medical image reconstruction results with the less data is an important issue to be resolved. This paper focuses on adaptive medical image reconstruction algorithms, the specific contents are as follows:Firstly, proposing the adaptive medical image reconstruction algorithm based on dual-tree complex wavelet, for the dual-tree complex wavelet can represent image edge structure information effectively. The proposed algorithm takes into account the image global and local sparse characteristics of the image simultaneously and can achieve better image reconstruction quality by compared with traditional adaptive Compressed Sensing image reconstruction algorithm.Secondly, considering that single sparse priori can produces better reconstruction results for the particular image, an adaptive sparse representation medical image reconstruction algorithm based on two kinds of sparse constraint is proposed. The proposed algorithm combines dual-tree complex wavelet and Hessian matrix norm as global sparse priori simultaneously and takes advantage of online dictionary method to learn adaptive dictionary. Experimental results show that the proposed algorithm can reduce image reconstruction time and improve image reconstruction quality efficiently at the same time.Finally, considering that most of the images contain different image components, a novel adaptive tight frame medical image reconstruction algorithm based on image decomposition is proposed. The proposed algorithm is used to achieve image decomposition reconstruction, which combines adaptive sparse representation with dual-tree complex wavelet or total variation sparse priori respectively. The last image reconstruction result can be obtained by summing the optimal solutions of the different components of the image. Experimental results show that the proposed algorithm is feasibility and robust.