基于图像投影网络的眼底图像分割技术的研究与应用  

Research and Application of Fundus Image Segmentation Technology Based on Image Projection Networks

作　　者：陈志哲 谢翔宇 刘申奥 彭磊 CHEN Zhizhe;XIE Xiangyu;LIU Shenao;PENG Lei(College of Medical Information and Artificial Intelligence Shandong First Medical University(Shandong Academy of Medical Sciences),Taian 271016,China)

机构地区：[1]山东第一医科大学(山东省医学科学院)医学信息与人工智能学院,山东泰安271016

出　　处：《智能物联技术》2025年第1期125-128,共4页Technology of Io T& AI

基　　金：山东省大学生创新创业训练计划项目(S202410439029);山东省本科高校教学改革研究项目(Z2022141)。

摘　　要：为了解决传统视网膜图像分析中依赖视网膜层分割和传统投影图所带来的操作复杂和准确性不足等问题,基于图像投影网络的眼底图像分割技术,在光学相干断层扫描血管造影(Optical Coherence Tomography Angiography,OCTA)数据集上实现无须进行视网膜层分割和传统投影图的视网膜指标量化。图像投影网络的核心组件是投影学习模块,通过组合多个投影学习模块能够有效地处理三维OCTA数据,并生成精确的二维分割结果。实验表明图像投影网络在视网膜血管分割和中央凹无血管区分割任务上均展现出卓越性能,突显了其在复杂视网膜疾病处理中的巨大潜力。To address the challenges of traditional retinal image analysis,including operational complexity and insufficient accuracy caused by reliance on retinal layer segmentation and conventional projections,this study proposes a fundus image segmentation technique based on Image Projection Network.This method quantifies retinal metrics from the Optical Coherence Tomography Angiography(OCTA)dataset without the need for retinal layer segmentation or traditional projection images.The core component of the Image Projection Network is the Projection Learning Module.By stacking multiple such modules,the network efficiently processes three-dimensional OCTA data and generates precise two-dimensional segmentation results.Experiments have shown that Image Projection Network exhibit excellent performance in both retinal vessel segmentation and central fovea avascular segmentation tasks.This highlights its enormous potential in the treatment of complex retinal diseases.

关 键 词：光学相干断层扫描血管造影(OCTA) 视网膜血管 中央凹无血管区 图像分割 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]