Research Of CT Image Reconstruction Algorithm

Computed tomography(CT) image reconstruction is a mathematical process that uses a specific mathematical algorithm to reconstruct the CT image according to the obtained projection data derived from X-ray CT scans of the patients or targets. At present CT image reconstruction is confronted with some new challenges. One of them is the complexity of the collection of a geometrical image data, and the high computational complexity of the reconstruction algorithm. Another one is that there is noise in the low-dose reconstructed image,and it's difficult to obtain a lower dose, higher quality scanned image. How to improve and design effective optimization method is the key to solve the problem.Aiming at some deficiencies in CT reconstruction algorithm, this thesis will conduct deep research to the CT reconstruction algorithm with the purpose of perfecting the basic CT reconstruction algorithm and expanding the practical applications of CT reconstruction technology. In summary, the works in this thesis can be categorized into the following two aspects.To reduce the reconstruction time and accelerate the convergence speed, the article adopts a model-based image reconstruction, which is a powerful technique for solving ill-posed inverse problems. Compared with traditional method, it can provide better estimates image from noisy measurements and from incomplete data. Based on the optimized gradient techniques and OS-LALM methods, order subsets linearized Augmented Lagrangian optimized gradient methods(OS-LALM-OGM) is proposed for CT image reconstruction.Experimental results show that the proposed algorithm significantly accelerates the convergence of X-ray CT image reconstruction, and improves the quality of reconstructed image, moreover, reduce OS artifacts when using many subsets.Reducing the radiation dose to the patient and obtaining a high precision quality of reconstructed image, and meeting the real-time request in clinical, thethesis uses matrix reconstruction technology as a kind of efficient data analysis tool. The reconstruction matrix approach supposes that the candidate restoring matrix should have low rank. Thus, this paper proposes a new CT image reconstruction algorithm, the reconstruction process has two steps: firstly, the low rank weighted nuclear matrix norm minimization(WNNM) is applied to image denoising, then using a low-rank decomposition of matrix to update CT images. Experimental results show that the proposed method has strong ability to keep the details of the CT images, and the characteristics of low-rank matrix can simplify the calculation process, and reduce the complexity of the algorithm.The proposed approach improves the accuracy of the reconstructed image, and ultimately achieves a high quality image.