A Three-Dimensional Denoising Method for Low-Dose Computed Tomography

Low-dose Computed Tomography (CT) becomes an increasingly significant research object in recent years result from the harm of radiation used in X-ray CT to health of patients. However, lowering the radiation dose will lead to increase of noise and reduction of image quality. In order to obtain high-quality low-dose CT images, a method based on block-matching three-dimension (3D) and generalized relative quality assessment is proposed in this paper. By grouping non-local similar image patches into a 3D stack, attenuating the noise in the stack and inversed stack, a denoised image can be acquired. In attenuating noise stage, a threshold should be selected to get the best performance; therefore, an optimizing scheme is used to select the best threshold. The proposed method performs well in inhibiting the noise in CT images and low-dose CT images. Experimental results demonstrate that the proposed method can obtain high-quality low-dose CT images in terms of both subjective visual quality and peak signal-to-noise ratio.