Translator Disclaimer
30 March 2016 Noise reduction in computed tomography using a multiplicative continuous-time image reconstruction method
Author Affiliations +
In clinical X-ray computed tomography (CT), filtered back-projection as a transform method and iterative reconstruction such as the maximum-likelihood expectation-maximization (ML-EM) method are known methods to reconstruct tomographic images. As the other reconstruction method, we have presented a continuous-time image reconstruction (CIR) system described by a nonlinear dynamical system, based on the idea of continuous methods for solving tomographic inverse problems. Recently, we have also proposed a multiplicative CIR system described by differential equations based on the minimization of a weighted Kullback–Leibler divergence. We prove theoretically that the divergence measure decreases along the solution to the CIR system, for consistent inverse problems. In consideration of the noisy nature of projections in clinical CT, the inverse problem belongs to the category of ill-posed problems. The performance of a noise-reduction scheme for a new (previously developed) CIR system was investigated by means of numerical experiments using a circular phantom image. Compared to the conventional CIR and the ML-EM methods, the proposed CIR method has an advantage on noisy projection with lower signal-to-noise ratios in terms of the divergence measure on the actual image under the same common measure observed via the projection data. The results lead to the conclusion that the multiplicative CIR method is more effective and robust for noise reduction in CT compared to the ML-EM as well as conventional CIR methods.
© (2016) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Yusaku Yamaguchi, Takeshi Kojima, and Tetsuya Yoshinaga "Noise reduction in computed tomography using a multiplicative continuous-time image reconstruction method", Proc. SPIE 9783, Medical Imaging 2016: Physics of Medical Imaging, 97834T (30 March 2016);

Back to Top