High Quality Reconstruction Based On Ordered Subsets Penalized Weighted Least-squares For Low-dose Computed Tomography

With the rapid development of the radiography technology, there are many CT equipment increase the dose of X-Ray to gain more clear medical image information. However, with the development of the radiography hygiene and the public self-protect-raising, more and more people take notice in the problem of the X-Ray dose. They require that to decrease the X-Ray dose in not to affect the diagnosis as far as possible. Summons the low dose CT equipment immediately the birth. Low dose CT may apply to the clinical in many aspects. For example the high-risk group lung cancer sieves examines, the CT enhancement scanning, reexamination of the lung tubercle pathological change, the follow-up visit about the lung cancer after treatment and the chemotherapy, infants and the child lungs CT inspection. How to effectively reduce X-ray dosage became one of primary missions at present medical CT research. Because of the quantum noise influence in the low dose CT scanning, the projection data of the reconstruction image degenerated. It is unable to achieve the request of the clinical practice in the traditional method that is used widespread in analysis reconstruction algorithm (FBP). Thanks to the method that based on statistics iterative compares with the traditional FBP method in processing the low dose CT has the remarkable merit, the correlation reconstruction algorithm obtains the extensive research.To solve the noise problem of image reconstruction in the low-dose CT. It investigates ordered subsets to the classical method of penalized weighted least-squares (OS-PWLS) to address this problem. Low dose CT information can be processed effectively improved by the OS-PWLS algorithm under less sample information. Because of the introduction of the Ordered Subsets, our proposed method enhances the speed effectively, and keeps the high quality image and perfect spatial resolution in the low-dose CT reconstruction.In the first chapter made the detailed summary to the history and actuality of CT image reconstruction, as well as the image algorithm.In Chapter2 main analyse and compare with each algorithm rationale, characteristic and application scope in CT image reconstruction at present. But obtains the reconstructed image includes serious noise and the resolution is low. It's unable to achieve the request for image reconstruction in the low dose CT.The main emphasis is on Chapter 3, Due to the method that based on statistics iterative compares with the traditional FBP method in processing the low dose CT has the remarkable merit, but the convergence rate is inferior to analyze algorithm. So this article proposes the new iterative algorithm, that base on the penalized weighted least-squares that is the most advanced in the world, introduces the ordered subsets technic, we can obtain quite satisfactory result on the speed and effect.Chapter 4 introduce the low dose CT practice situation in clinical, may solve many difficult problems on the clinical, specially advantage in many disease that need general surveys and sieves examines, Enables more diseases to be able to discover earlier and treats earlier.Chapter 5 briefly summarizes the major contribution of the current work and provides some suggestions for the future research.