Breast MR Image 3D Super-resolution Reconstruction

Image super-resolution reconstruction is to estimate a high-resolution image with better quality from single or multiple frames low-resolution images with the application of computer technology,applying super-resolution reconstruction technology to the breast MR image can improve the MR image sequence resolution between layer and layer,and these higher-quality image sequence can assist doctors to diagnose lesions and make more precise treatment,so 3D medical image super-resolution reconstruction has wide application prospect and practical application value.In this paper,single image super-resolution methods have been studied,and the super-resolution reconstruction based on offline double dictionary learning is the key research contents,two innovation points are put forward to further improve the effect of super-resolution reconstruction: First,wavelet features are used in experiments instead of the traditional first order and second order gradient features,since wavelet features not only have the characteristics of horizontal and vertical information but includes diagonal direction information,rich information help to improve the sparse representation effect of a signal.More abundant dictionary information can be got with the use of wavelet features.Second,weighted combination method is put forward to composite high-resolution image patches,the traditional way is to take the average of all reconstruction image blocks,however,the quality of each reconstruction patch is different,the overall reconstruction quality of the image can be better by giving different weight value to each patch according to their reconstruction results.Experimental results show that the proposed super-resolution reconstruction is better than the current several classical algorithms based on the objective image quality evaluation standard of peak signal-to-noise ratio and structural similarity.The 2D super-resolution can be extended to 3D,interpolation is need when getting virtual slice between two adjacent layers,much more information can be got by applying superresolution algorithm to virtual slice.3D visualization technology is applied to illustrate the performance of 3D super-resolution reconstruction,the results are improved remarkable in subjective visual effect compared with original image.