Research On Three-term Conjugate Gradient Method And Parameter Optimization In X-ray Luminescence Computed Tomography

X-ray luminescent computed tomography(XLCT)is an optical molecular imaging technique with high resolution and sensitivity.It has the advantages of low cost,fast imaging speed and high imaging accuracy.XLCT uses X-ray to irradiate organisms containing nanophosphors,then the emitted light spreads to the surface of organisms through scattering and absorption of tissues,and the distribution of light sources in the body is reconstructed through the prior information of the surface.XLCT has an important application value in the fields of early tumor detection,pharmaceutical research and lesion tracing and tissue observation.However,because the reconstruction problem is a large sparse matrix solution problem,fast and robust reconstruction algorithm is the focus of current XLCT research.Regularization method is often used in XLCT reconstruction,in which regularization algorithm and parameters are important factors affecting the reconstruction results.In this paper,a three-term conjugate gradient algorithm is proposed to accelerate the reconstruction of inverse problems on the framework of incomplete variables,and a regularization parameter selection strategy based on incomplete variables is proposed to solve the parameter selection problem in regularization method.The main research works of this paper are as follows:(1)In the iterative sparse regularization reconstruction algorithm,a large number of matrix calculations are carried out in the iterative process,resulting in a large number of iterations and time.To improve the efficiency of reconstruction,a three-term Conjugate Gradient(TTCG)algorithm was proposed.Combined with the incomplete variables reconstruction framework,the proposed algorithm adds useful truncation information in the descending direction of restart,reduces the redundant calculation starting from negative gradient and speeds up reconstruction by using the third direction information.Simulation and real experiments show that compared with the original incomplete variables truncated conjugate gradient method,the reconstruction method based on three-term conjugate gradient algorithm can increases solution speed by over 20% while maintaining the same high-precision results,and the solution speed of in vivo experiment is increased by about50%.(2)As regularization parameters in regularization methods seriously affect the final results,most regularization methods need to set parameters empirically,which limits the stability of reconstruction algorithms.By introducing Karush-Kuhn-Tucker equivalent conditional residuals,we propose a regularization parameter selection strategy based on incomplete variables.The regularization parameters are selected at the equilibrium point where the residuals change from gentle to oscillating and combined with TTCG algorithm for reconstruction.Numerical simulation and real experiments show that the regularization parameters selected by the strategy proposed in this paper can control the reconstruction accuracy of XLCT under 1mm in different conditions,and the DICE coefficient of the reconstruction is higher than that of the L-curve and U-curve selection strategies.