Study on imaging reconstruction based on filtered back-projection combined with a modified shift-invariant wavelet threshold method

Abstract

In this paper, the wavelet threshold denoising method was used and combined into the FBP algorithm to reconstruct the imaging. To overcome the drawbacks of the traditional wavelet threshold functions, an improved wavelet threshold function was proposed, and to improve the reconstruction effect of image, the shift-invariant method was coupled into the improved wavelet threshold function. To verify the feasibility of the improved wavelet threshold function combined with shift-invariant method, the simulation experiments of the standard brain phantom were performed by using the FBP algorithm based on different wavelet threshold functions at the software platform of MATLAB. In addition, the peak signal-to-noise ratio (PSNR) and mean-square-error (MSE) values of different methods were computed. Experimental results show that, compared with traditional wavelet threshold functions, the image reconstruction based on the FBP of the improved wavelet threshold function combined with shift-invariant method has better reconstruction effect. Among the SL, RL, and Hann filters for the FBP of the improved wavelet threshold function combined with shift-invariant method, the PSNR value based on RL filter was increased 38.7%, and the MSE value was increased 82%. Keywords: image reconstruction, filtered back-projection algorithm, wavelet threshold denoising.