中国光学（中英文）2024，Vol.17Issue(4)：982-994,13.DOI:10.37188/CO.EN-2023-0024
基于窄带成像及放大内镜的脂质分割方法
Lipid segmentation method based on magnification endoscopy with narrow-band imaging
摘要
Abstract
Magnification endoscopy with narrow-band imaging(ME-NBI)has been widely used for cancer diagnosis.However,some microstructures are rendered invisible by a white opaque substance(WOS)com-posed mainly of lipids.In such lesions,the morphological structure of lipids becomes another marker of tu-mor grade.We propose a lipid segmentation method.First,the lipid image enhancement algorithm and the specular reflection correction algorithm are introduced.Then,in the framework of the active contour model,the proposed segmentation method extracts local information from modified hue value and global informa-tion from intensity value and adaptively obtains the weight factor to segment the lipid region based on the ini-tial contour.This method's effectiveness is verified by a phantom experiment,which shows that it attained higher than 90% in several key measures:pixel accuracy,sensitivity,and Dice coefficient.The proposed method can accurately reflect the shape of lipids to provide available information for doctors.关键词
窄带光成像/脂质/分割/活动轮廓模型/仿体Key words
narrow-band imaging/lipids/segmentation/active contour model/phantom分类
计算机与自动化引用本文复制引用
武治晟,邹鸿博,朱文武,齐伟明,王立强,袁波,杨青,徐晓蓉,严蕙蕙..基于窄带成像及放大内镜的脂质分割方法[J].中国光学（中英文）,2024,17(4):982-994,13.基金项目
国家自然科学基金(No.T2293751) （No.T2293751）
国家重点研发计划(No.2021YFC2400103)Supported by the National Natural Science Foundation of China(No.T2293751) （No.2021YFC2400103）
the National Key Research and Development Program of China(No.2021YFC2400103) （No.2021YFC2400103）
