Scholarlink

Background <p>Retinal microvascular parameters are potential biomarkers for systemic vascular health. This study aimed to assess the interocular consistency of these parameters in patients with coronary atherosclerotic disease(CAD) using AI-based quantitative analysis.</p> Methods <p>We analysed 862 colour fundus photographs from 431 patients. Retinal vascular parameters—including fractal dimension (FD), vessel density (VD), and vessel diameters—were automatically measured using a ResNet 101-UNet deep learning model. Interocular symmetry was compared across groups stratified by Gensini score, CAD presence, and lesion location. Correlation was assessed using Pearson and intraclass correlation coefficients (ICC).</p> Results <p>Except for minor curvature variations in the overall population, no statistically significant interocular differences were found across any subgroups (Gensini scores, CAD vs. Non-CAD, or vessel location; all <i>P</i> &gt; 0.05). All variables demonstrated good interocular consistency, with Pearson coefficients ranging from 0.387 to 0.833 and ICCs from 0.384 to 0.824. While linear regression suggested minor associations between gender and interocular differences in VD and mean branch angle, generalised linear mixed models—adjusting for age, gender, and clinical history—confirmed no significant interocular variance in fundus metrics across all subgroups (<i>P</i> &gt; 0.05).</p> Conclusion <p>Retinal vascular parameters exhibit high interocular symmetry in individuals regardless of their CAD status. These findings suggest that single-eye imaging may potentially represent retinal vascular status in cardiovascular risk assessment. This approach could help streamline screening protocols and provides a rationale for the standardised use of fundus examination as a non-invasive auxiliary tool for CAD evaluation.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Comparative analysis of quantitative fundus results of the left and right eyes and screening for coronary atherosclerotic disease: a binocular study

  • 期刊论文

  • 发表日期：

    2026年6月29日

查看全文
  • Zhenyan Wu,
  • Xue Jiang,
  • Yu Xin,
  • Xinying Guo,
  • Jie Zhang,
  • WenYi Zhang,
  • Jian Liu,
  • Yiran Wang,
  • Saiguang Ling,
  • Zibing Jin,
  • Caixia Guo

摘要

Background

Retinal microvascular parameters are potential biomarkers for systemic vascular health. This study aimed to assess the interocular consistency of these parameters in patients with coronary atherosclerotic disease(CAD) using AI-based quantitative analysis.

Methods

We analysed 862 colour fundus photographs from 431 patients. Retinal vascular parameters—including fractal dimension (FD), vessel density (VD), and vessel diameters—were automatically measured using a ResNet 101-UNet deep learning model. Interocular symmetry was compared across groups stratified by Gensini score, CAD presence, and lesion location. Correlation was assessed using Pearson and intraclass correlation coefficients (ICC).

Results

Except for minor curvature variations in the overall population, no statistically significant interocular differences were found across any subgroups (Gensini scores, CAD vs. Non-CAD, or vessel location; all P > 0.05). All variables demonstrated good interocular consistency, with Pearson coefficients ranging from 0.387 to 0.833 and ICCs from 0.384 to 0.824. While linear regression suggested minor associations between gender and interocular differences in VD and mean branch angle, generalised linear mixed models—adjusting for age, gender, and clinical history—confirmed no significant interocular variance in fundus metrics across all subgroups (P > 0.05).

Conclusion

Retinal vascular parameters exhibit high interocular symmetry in individuals regardless of their CAD status. These findings suggest that single-eye imaging may potentially represent retinal vascular status in cardiovascular risk assessment. This approach could help streamline screening protocols and provides a rationale for the standardised use of fundus examination as a non-invasive auxiliary tool for CAD evaluation.