Research On CBCT Image Denoising Algorithm

With the development of smart healthcare technology,many medical industries are moving towards intelligence,and meantime medical imaging technology get a rapid development leveraging smart healthcare technology.Cone-Beam Computed Tomography(CBCT)has received widespread attention for CBCT has the advantages of strong real-time performance and high sensitivity.Besides that,CBCT can obtain a more satisfactory reconstructed image with less radiation dose.There are two main applications of CBCT images in clinical medicine.One is for the diagnosis of diseased tissues:a more accurately treatment plans can be formulated using obtained three-dimensional images;the second is for real-time tracking of radiation therapy to achieve the goal of precise treatment.Compared with traditional CT,CBCT has many advantages.However,the contrast of the image in the soft tissue is lower with artifacts,CBCT images are particularly easy to be noise pollution for that CBCT uses less radiation dose than traditional CT,these factors seriously reduces the quality of CBCT images.Increasing the X-ray radiation dose can give a satisfactory image,but the patient will absorb more radiation and damage healthy cell structures.Therefore,it is of great research value to improve the quality of CBCT images without increasing the radiation dose.This article first analyzes traditional medical images and their existing noise,and then focuses on CBCT images.Through the analysis of CBCT imaging principles,we understand that CBCT images can be de-noised from the projection domain and the reconstruction domain.In this paper,different denoising methods are used to study different characteristics of the two domains.The main work and innovations of this article are as follows:1.Denoising in the reconstruction domain.By analyzing the partial differential equation denoising model,a partial differential equation denoising model suitable for CBCT images is proposed The model uses Log operator to construct a new edge detector.An adaptive diffusion function is designed by the new edge detector,and at the same time the edge energy retention index is proposed according to the fuzzy coefficient,so that the partial differential equation can be adaptively terminated when iteratively denoising.Experiments show that the improved denoising model is effective in CBCT images.2.Denoising combined with projection domain and reconstruction domain.Aiming at the solving the disadvantages of massive data in the projection domain and long time for denoising,a denoising algorithm combining projection domain and reconstruction domain is proposed for denoise in the projection domain of the CBCT image.This algorithm is based on wavelet transform to achieve rapid reconstruction.First,the image in the projection domain is decomposed by wavelet,the decomposed image is Wiener filtered in the low frequency part,and gets a threshold processing in the high frequency part.The processed high and low frequency data are then separately processed FDK reconstruction,followed by inverse wavelet transformation and interpolation of the reconstructed high and low frequency data to get the reconstructed CBCT image.Because of the existence of artifacts in obtained CBCT image,the processing of partial differential equations is needed in the reconstruction domain.Compared with the traditional projection domain denoising algorithm,the proposed algorithm saves nearly half of the time as well as gets a better quality obtained image.