X-ray computed tomography (CT) is now a widely used imaging modality for numerous medical purposes. Due to the biological risk of x-ray radiation, developing CT methodologies minimizing x-ray exposure to patients while achieving the clinical tasks has been a major concern. However, reducing the radiation dose as in low-dose CT techniques results in images that are often degraded by noise and artifacts. We present here two methods for denoising of low-dose CT images. In the first method, a noisy image is decomposed into three frequency bands namely low-band, middle-band and high-band such that the noise component mainly is presented in the middle and high bands. Then, by exploiting the fact that a small patch of the noisy image can be approximated by a linear combination of several elements in a given dictionary of noise-free image patches, generated from noise-free CT images taken at nearly the same position with the noisy image, noise on these two bands is effectively eliminated. In the second method, the noisy image is denoised patch-wise in which each noisy patch is estimated by a sparse representation using a dictionary of patches built from noise-free example images, which are similar to the noisy image. Experimental results conducted on both synthetic and real noise data demonstrated the efficiency of the proposed methods.
Speaker: Mr. Nguyen Thanh Trung, AVITECH
Time: 15:30, Tuesday, February 26, 2019
Venue: E3-707, 144 Xuan Thuy, Cau Giay, Hanoi
Nguyen Thanh Trung received his B.Eng.degree in Electronics and Telecommunicationsfrom the Faculty of Technology, Vietnam Na-tional University, Hanoi, Vietnam in 2003, M.S.degree in electronics engineering from Uni-versity of Engineering and Technology, Viet-nam National University, Hanoi, Vietnam in 2012. He is now a Phd student at VNU-UET and a lecturer at the Faculty of Electronics and Communication, University of Information and Communication Technology, Thainguyen University, Vietnam. His research interests include biomedicalsignal and image processing,sparse coding, machine learning.