Image Restoration For Low-dose Computed Tomography

X-ray computed tomography (CT) has experienced tremendous growth in recently 30 years and has achieved substantial enhancement both in efficiency and precise. With the emergence of multi-slice CT and dual-source CT, some difficult technical, such as high-quality imaging of the coronary arteries and major organ perfusion, have been applied in clinical. However, as the new CT equipment used higher X-ray dose than conventional ones, making it more and more concerned about potential harm to humans from high X-ray dose. So, minimizing x-ray exposure to the patients has been one of the major efforts in the CT fields.Recently, with the rapid development of hardware, a number of new algorithms and ideas appears from the field of image denoising and restoration, and this is provided new opportunities to improve the quality of low-dose CT images. In this article, we will review the field of image denoising and make a sense of the current non-local, multi-scale algorithm, then apply the new ideas to the field of low-dose CT restoration.The work we do on the restoration algorithms of Low-dose CT image are as following:(1) A novel approach is proposed to improve low dose CT image quality by a non-local weighs prior which is acquired from previous normal-dose scan. First, we register the low-dose and previous normal-dose scan CT images and filter the low dose image to suppress noise. Then, the weight matrix of the registered normal-dose image is calculated by the non-local means weighting formula, and the low-dose image is recovered using this weight matrix as a prior. Simulated phantom and perfusion data experiment results demonstrate a highly quality improvement in terms of suppressing noise-induced streak artifacts and preserving resolution of the low dose image. The proposed method is especially effective for clinical short-term repeated CT scan, such as perfusion and radiation therapy.(2) A novel approach is proposed based on low-dose CT projection restoration. In this paper, First, projection data is transformed to a Gaussian distribution from a Poisson distribution using nonlinear Anscombe transform. Then, the transformed data is filtered by an efficient BM3D algorithm based. Last, the reconstruction result is achieved by inverse Anscombe transform and filtered back projection (FBP) method. Because the variance of the Gaussian noise is known, the proposed scheme can be implemented without any manual parameters. Simulated and clinical low-dose CT data experiment results demonstrate that a highly quality CT image can be reconstructed with noise-induced streak artifacts effectively suppression.