Noise Reduction For Low-dose CT Image Based On Particle Filtering

With the rapid development of the medical imaging technology, there are many CTequipments to gain more clearly medical image information by increasing the dose ofX-Ray. However, with the development of the radiography hygiene and the publicself-protect-raising, more and more people take notice in the problem of the X-Ray dose.How to effectively reduce X-ray dosage is becoming one of primary missions at presentmedical CT research. Because of the quantum noise influence in the low dose CTscanning，the projection data of the reconstruction image degenerated．It is unable toachieve the request of the clinical practice in the traditional method that is usedwidespread in analysis of reconstruction algorithm (such as FBP: Filtered BackProjection).To improve the quality of low-dose CT image is essentially a problem of imagedenoising, if this problem can be solved with software algorithms, we have not toreplace the CT hardware, under this premise that low-dose CT or ultra-low-dose CT canbe more widely used. Many research groups at home and abroad are trying to improvemethods used in enhancing quality of low-dose CT image; under this background thesubject in this context is to study the issues of low-dose CT image denoising.In this study, to obtain the statistical properties of the sinogram data, experimentswith physical phantoms were first performed to acquire low-dose projections repeatedlyat a fixed angle．Statistical analysis of the data revealed that the noise of the sinogramdata could be regarded as approximately Gaussian distributed with a nonlinearsignal-dependent variance，and non-stationary noise model was established based onabove findings．Considering the feature and the distribution of the noise, this paperbased on statistical properties of low dose sinogram data, studies on the particle filtering(PF) denoising algorithm. The paper proposes two methods for the filtering of imagenoise, namely GPF (General PF) and PDPF (Projection Data be denoised with PF).Combination of noise, this basic particle filter algorithm is improved, specifically forlow-dose CT sinogram data to filter out the noise. After statistically-based restoration,sinograms were reconstructed using the classical FBP method for diagnosis as well asfurther analysis and processing. Finally, this paper discusses the convergence of particlefilter, using three methods of quantitative analysis to evaluate this algorithm. Since theproposed framework for noise reduction is algorithm based software，it could be useddirectly in current CT equipment and has a potential to be used in a broad clinicalpractice．...