The Denoising Algorithm Of Low-Dose CT Images Based On Bayesian Estimation

CT(Computed Tomography) has become an important means of clinical diagnosis and treatment. In order to reduce the damage of X-ray radiation dose on the human body, the low-dose CT scanning has gradually become a hot topic in clinical application. However, due to the reduction of radiation dose, low-dose CT images contain a large number of quantum noise, and the noise seriously affects the image quality. Besides, Due to the quantum noise relating to pixel values of the CT images,the traditional method of removing Gaussian noise is not good for the low-dose CT images. Therefore, we mainly study the algorithm of removing quantum noise in low-dose CT images in this paper. The main work of the paper is as follows:The estimation accuracy of noise variance determines the de-noising effect. To solve the poor accuracy of the traditional Median Absolute Deviation(MAD) method,an improved algorithm is proposed for noise variance estimation. Firstly, the noisy image is decomposed into the low-frequency sub-band coefficients and multi-directional high-frequency sub-band coefficients based on the Non-subsampled shearlet transform(NSST). Secondly, based on the high-frequency sub-band coefficients, the value of the noise variance is estimated using the MAD method.Thirdly, we choose some variance candidates in the neighborhood of the estimated value, and calculate the Residual Autocorrelation Power(RAP) of every variance candidate based on the Bayesian maximum a posteriori estimation(MAP) method.Finally, the accuracy of the noise variance estimation is improved using the residual autocorrelation power. A range of experiments demonstrate that the accuracy of the proposed noise variance estimation is much higher than the traditional MAD method.At the same time, the proposed algorithm is used to denoise the same CT image with different noise intensity and the different CT images with the same Gaussian noise intensity. The experimental results show that proposed algorithm not only is effective,but also has strong adaptive ability to remove noise.The low-dose CT(LDCT) scanning is an effective way to reduce the X-ray radiation dose. In this paper, due to quantum noise caused by the reduction of X-rayradiation dose leads to degradation of low-dose images quality, we propose a quality improvement algorithm for low-dose CT images based on the improved noise variance method. Firstly, LDCT image is transformed using the Anscombe transform,and the quantum noise is transformed into noise which approximately obeys Gaussian distribution. Secondly, the transformed image is decomposed into low-frequency coefficient sub-bands and multi-directional high-frequency coefficient sub-bands based on shearlet transform. Then, for high-frequency coefficient sub-bands of the low signal noise ratio, we utilize the improved noise variance estimation method,which is combined with Bayesian maximum posterior probability method to obtain the more accurate non-noise high-frequency coefficients. Finally, the reconstructed image is obtained using the shearlet inverse transform and Anscombe inverse transform. A series of experiments are performed on the conventional dose CT images added Poisson noise and the actual low-dose CT images. Quantitative evaluation and visual effects show that the proposed algorithm not only can effectively remove the noise, but also can preserve the edge information and texture information of the image.Because the two-dimensional image is not enough intuitive, which will cause that the work efficiency of doctors is not high or even lead to misdiagnosis. In this paper, the proposed algorithm is used to denoise a series of low-dose chest CT images,then we make three dimensional reconstruction. The experimental results show that the reconstructed image using the proposed algorithm not only improves the efficiency of doctors, but also improve the accuracy of clinical diagnosis.