Denoising Algorithm Combining Cone Beam CT Projection Data with Reconstructed Image

Abstract

In order to obtain high-quality cone-beam CT images and improve the accuracy of medical diagnosis, this paper proposes a denoising algorithm combining Cone Beam CT projection data and reconstructed images. Firstly, the noisy projection data are decomposed by wavelet transform, and the corresponding decomposition coefficients are obtained. Threshold processing of its high-frequency coefficients, the low-frequency part is subjected to Wiener filtering. The denoised coefficients are separately reconstructed by FDK to obtain four sets of 3D data. Then, the inverse wavelet transform and interpolation are performed to get the reconstructed 3D image. Finally, the reconstructed cone beam CT image is repaired by an anisotropic diffusion algorithm, which can suppress artifacts. The number of iterations during anisotropic diffusion is automatically selected based on the PSNR value of the image. The experimental results show that both PSNR and image quality are satisfactory, and the denoising algorithm in this paper saves about half of the time compared with the traditional CBCT denoising algorithm.