机器学习和图像识别在黄斑裂孔手术预后预测中的应用与进展
The application and progress of machine learning and image recognition in predicting the prognosis of macular hiatus surgery
程甜甜 1韩若安 1陈有信1
作者信息
- 1. 中国医学科学院北京协和医院眼科 中国医学科学院北京协和医学院,北京 100730
- 折叠
摘要
黄斑裂孔(MH)是发生于黄斑区的神经上皮缺失,可导致严重的中心视力下降.根据MH的病因和病变分期不同,临床主要采用玻璃体切割手术、内界膜剥离术和眼内气体填充,以达到解剖和功能恢复.裂孔直径、黄斑裂孔指数、裂孔形成因子、黄斑闭合指数等术前光学相干断层扫描(OCT)参数以及患者基线特征与MH手术预后密切相关,基于机器学习和图像识别的人工智能模型在预测手术预后结果方面体现出了潜力.本文旨在总结影响MH手术预后的重要参数,分析现有机器学习和图像识别在MH手术预测中的应用现状和可能的改进方向,为相关研究和临床应用提供依据和新的思路.
Abstract
Macular hiatus(MH)is a neuroepithelial deficiency that occurs in the macular area and can lead to severe central visual impairment.According to the different causes and stages of MH,clinical methods mainly include vitrectomy,inner limiting membrane dissection,and intraocular gas filling to achieve anatomical and functional recovery.The preoperative optical coherence tomography(OCT)parameters such as hole diameter,macular hole index,hole formation factor,and macular closure index,as well as patient baseline characteristics,are closely related to the prognosis of MH surgery.Artificial intelligence models based on machine learning and image recognition have shown potential in predicting surgical outcomes.This article aims to summarize the important parameters that affect the prognosis of MH surgery,analyze the current application status and possible improvement directions of machine learning and image recognition in MH surgery prediction,and provide a basis and new ideas for related research and clinical applications.
关键词
玻璃体视网膜手术/黄斑裂孔/光学相干断层扫描/图像识别Key words
Vitreoretinal surgery/Macular hole/Optical coherence tomography/Image recognition引用本文复制引用
基金项目
国家自然科学基金(82271112)
广东省重点领域研发计划(2021B0101420005)
出版年
2024
NETL
NSTL
万方数据