TSAR: A two-stage approach to motion artifact reduction in OCTA images

Open Webpage

Benteng Ma, Xiaomeng Li, Xu Lin, Xiaoyu Bai, Dongping Shao, Chubin Ou, Lin An, Jia Qin, Kwang-Ting Cheng

Published: 2026, Last Modified: 24 Jan 2026Pattern Recognit. 2026EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•We propose a framework TSAR that can simultaneously remove displacement artifacts, duplicated scanning artifacts, and white line artifacts in OCTA images.•We develop a novel transformer, HFormer, which can capture channel-wise, axial-wise, and intra-axial relationships among OCTA images, effectively addressing directional misalignment and discontinuity.•We design a conditional diffusion-based RCDM that enhances white line artifact removal in OCTA images by leveraging contextual information beyond the artifacts.