Authors: Kavkirat Kaur; Shailendra Tiwari
Addresses: Department of Computer Science and Engineering, Thapar Institute of Engineering and Technology, Patiala-147004, India ' School of Computer Science and Statistics (SCSS), Trinity College Dublin (TCD), Ireland
Abstract: There is a high demand for a framework that can eliminate the Gaussian noise and provide high-quality images in the case of low-dose radiation in computed tomography (CT). This paper presents a new statistical image reconstruction (SIR) algorithm for this purpose. The proposed framework uses a combination of data fidelity and regularisation terms. The data fidelity term is derived using maximum likelihood expectation maximisation (MLEM) algorithm. After that, a complex diffusion (CD) filter is applied as a regularisation term into the proposed framework that minimises the noise while preserving the edges along with the fine structure information in the reconstructed image. The proposed model has been evaluated on both simulated and real standard thorax phantoms. The results are compared with the other standard methods and it is observed that the proposed model has many desirable advantages such as better noise robustness, less computational cost, enhanced denoising effect.
Keywords: computed tomography; CT; noise reduction; maximum likelihood expectation maximisation; MLEM; complex diffusion; CD; Gaussian noise.
International Journal of Biomedical Engineering and Technology, 2021 Vol.35 No.1, pp.37 - 69
Received: 19 Oct 2017
Accepted: 12 Jan 2018
Published online: 25 Jan 2021 *