“…To overcome this issue, ultra-low-dose CT (ULDCT) has been studied intensively [ 4 , 5 , 6 ]. Because images obtained by LDCT or ULDCT are severely affected by noise (e.g., streak artifact) and differentiation between normal/abnormal findings on the noisy CT images is difficult, image-processing techniques have been utilized to improve image quality of LDCT or ULDCT [ 4 , 5 , 6 , 7 , 8 , 9 , 10 ]. While many methods have been proposed in previous studies, image-processing techniques of LDCT or ULDCT can be roughly divided into two categories: raw-data-based techniques [ 4 , 5 , 6 ] and post-processing techniques [ 7 , 8 , 9 , 10 ].…”