Study On The Noise Reduction Algorithms For The Narrow-bins Of Spectral CT

In recent years,the spectral CT can get the projection data sets by using photon counting detector to detect objects at the same time under the different energy channels.Because different density materials have different attenuation coefficient values under certain energy,the material and pixel similarity can be calculated directly through different energy characteristics.Small differences in the density of pixels in multiple energies result in a very large difference in the mean square deviation,which is conducive to highlighting the details in the image contrast or the difference hour.Therefore,the development of CT can provide the possibility for the realization of the new function of traditional CT,which has a great advantage in practical application.The spectral CT needs to determine the energy channel,so the number of photons detected in the channel is less than the total number,which makes the noise influence increase.Therefore,the study of noise reduction algorithm is the necessary way for the development of energy spectrum CT.Because the reconstruction of the image under the single energy channel of the spectral CT is similar to that of traditional CT,we can introduce the algorithm with good denoising effect in traditional CT to the reconstruction of the spectral CT.In this paper,we study the sinogram noise reduction algorithm and the algorithm based on a priori information of reference spectral image in spectral CT.(1)The noise of the narrower spectrum channel of spectral CT have certain statistical properties,and the projection data after system calibration and the logarithmic transformation obeying the non-stationary gaussian distribution.There is a nonlinear relationship between the mean and variance of the projection.you can use the model,by means of different scanning Angle projection data using nonparametric regression method to fitting for noiseless projection data.Classical statistical iterative algorithm can be used first to deal with noise projection,in view of the projection of pixels between anisotropic smoothing parameter to optimize the original MAP algorithm,comparing with the existing SB algorithm,get all kinds of algorithm of spectrum CT projection domain data noise reduction effect,narrow spectrum shows that the algorithm for the applicability of the energy spectrum CT.(2)The spectral CT images have a thorough knowledge of the structural similarity between channels,which can be used to constraint reconstruction algorithm to realize the image denoising.In this paper,by using the same structure of image in different energy channels to design the regularization,and combining with the existing NLTV item which has better effect of image noise to constraint reconstruction algorithm.We present a new reconstruction algorithm based on the weighted NLTV term with the structural priori information,the simulation results show that the algorithm is better than the existing algorithm in the effect in denoising.