In abdomen computed tomography (CT), repeated radiation exposures are often inevitable for cancer patients who receive surgery or radiotherapy guided by CT images. Low-dose scans should thus be considered in order to avoid the harm of accumulative x-ray radiation. This work is aimed at improving abdomen tumor CT images from low-dose scans by using a fast dictionary learning (DL) based processing. Stemming from sparse representation theory, the proposed patch-based DL approach allows effective suppression of both mottled noise and streak artifacts. The experiments carried out on clinical data show that the proposed method brings encouraging improvements in abdomen low-dose CT images with tumors.

The x-ray exposure to patients has become a major concern in Computed Tomography (CT) and minimizing the radiation exposure has been one of the major efforts in CT field. Due to the plenty high-attenuation tissues in human chest, under low dose scan protocols, thoracic low-dose CT (LDCT) images tend to be severely degraded by excessive mottled noise and non-stationary streak artifacts. Their removal is rather a challenging task because the streak artifacts with directional prominence are often hard to be well discriminated from the attenuation information of normal tissues. This paper describes a two-step processing scheme called "Artifact Suppressed Large-scale Nonlocal Means" (AS-LNLM) for suppressing both noise and artifacts in thoracic LDCT images. Specific scale and direction properties were exploited to discriminate the noise and artifacts from image structures. Parallel implementation has been introduced to speed up the whole processing by more than 100 times. Phantom and patient CT images were both acquired for evaluation purpose. Comparative qualitative and quantitative analyses were both performed that allows concluding on the efficacy of our method in improving thoracic LDCT data.

International audienceMoments and moment invariants have become a powerful tool in pattern recognition and image analysis. Conventional methods to deal with color images are based on RGB decomposition or graying, which may lose some significant color information. In this paper, by using the algebra of quaternions, we introduce the quaternion Zernike moments (QZMs) to deal with the color images in a holistic manner. It is shown that the QZMs can be obtained from the conventional Zernike moments of each channel. We also provide the theoretical framework to construct a set of combined invariants with respect to rotation, scaling and translation (RST) transformation. Experimental results are provided to illustrate the efficiency of the proposed descriptors