文献速递眼底病与葡萄膜炎

深度学习系统在澳大利亚原住民中检测糖尿病视网膜病变的验证

Validation of a deep learning system for the detection of diabetic retinopathy in Indigenous Australians.
PubMed 文献 · Br J Ophthalmol · 2024-01-29

这项回顾性外部验证研究在澳大利亚珀斯一家原住民初级保健机构,用864例糖尿病患者的1682张连续单视野黄斑中心视网膜照片,比较深度学习系统(DLS)与视网膜专科医生检测糖尿病视网膜病变(DR)的性能。DLS检测轻于轻度以上DR、视力威胁性DR和全因可转诊DR的敏感度均优于视网膜专科医生,特异度略有降低。

要点

  • DLS检测轻于轻度以上DR的敏感度98.0% vs 视网膜专科医生87.1%,特异度95.1% vs 97.0%。
  • DLS检测视力威胁性DR的敏感度96.2% vs 84.4%,特异度95.8% vs 97.8%。
  • DLS检测全因可转诊DR的敏感度93.7% vs 74.4%,特异度91.7% vs 96.3%。

核心结论

  • DLS检测DR的敏感度优于视网膜专科医生,特异度相似,显示其有潜力支持澳大利亚原住民DR筛查。

研究设计与样本

  • 回顾性外部验证研究,比较DLS与视网膜专科医生检测轻于轻度以上DR、视力威胁性DR和全因可转诊DR的性能。
  • 验证集为澳大利亚珀斯一家原住民初级保健机构的864例糖尿病患者(平均年龄54.9岁,52.4%为女性)的1682张连续单视野黄斑中心视网膜照片。
  • 以三人专科医生小组裁决作为参考标准。

主要发现

  • 检测轻于轻度以上DR:DLS敏感度98.0%(95%CI 96.5至99.4)vs 视网膜专科医生87.1%(95%CI 83.6至90.6),McNemar检验p<0.001;DLS特异度95.1%(95%CI 93.6至96.4)vs 97.0%(95%CI 95.9至98.0),p=0.006。
  • 检测视力威胁性DR:DLS敏感度96.2%(95%CI 93.4至98.6)vs 视网膜专科医生84.4%(95%CI 79.7至89.2),p<0.001;DLS特异度95.8%(95%CI 94.6至96.9)vs 97.8%(95%CI 96.9至98.6),p=0.002。
  • 检测全因可转诊DR:DLS敏感度93.7%(95%CI 91.8至95.5)vs 视网膜专科医生74.4%(95%CI 71.1至77.5),p<0.001;DLS特异度91.7%(95%CI 90.0至93.3)vs 96.3%(95%CI 95.2至97.4),p<0.001。

临床怎么用

  • 作者认为DLS有潜力支持澳大利亚原住民这一服务不足且糖尿病眼病负担高的人群的DR筛查。

局限性

  • 原文未报告局限性。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

BACKGROUND/AIMS: Deep learning systems (DLSs) for diabetic retinopathy (DR) detection show promising results but can underperform in racial and ethnic minority groups, therefore external validation within these populations is critical for health equity. This study evaluates the performance of a DLS for DR detection among Indigenous Australians, an understudied ethnic group who suffer disproportionately from DR-related blindness. METHODS: We performed a retrospective external validation study comparing the performance of a DLS against a retinal specialist for the detection of more-than-mild DR (mtmDR), vision-threatening DR (vtDR) and all-cause referable DR. The validation set consisted of 1682 consecutive, single-field, macula-centred retinal photographs from 864 patients with diabetes (mean age 54.9 years, 52.4% women) at an Indigenous primary care service in Perth, Australia. Three-person adjudication by a panel of specialists served as the reference standard. RESULTS: For mtmDR detection, sensitivity of the DLS was superior to the retina specialist (98.0% (95% CI, 96.5 to 99.4) vs 87.1% (95% CI, 83.6 to 90.6), McNemar's test p<0.001) with a small reduction in specificity (95.1% (95% CI, 93.6 to 96.4) vs 97.0% (95% CI, 95.9 to 98.0), p=0.006). For vtDR, the DLS's sensitivity was again superior to the human grader (96.2% (95% CI, 93.4 to 98.6) vs 84.4% (95% CI, 79.7 to 89.2), p<0.001) with a slight drop in specificity (95.8% (95% CI, 94.6 to 96.9) vs 97.8% (95% CI, 96.9 to 98.6), p=0.002). For all-cause referable DR, there was a substantial increase in sensitivity (93.7% (95% CI, 91.8 to 95.5) vs 74.4% (95% CI, 71.1 to 77.5), p<0.001) and a smaller reduction in specificity (91.7% (95% CI, 90.0 to 93.3) vs 96.3% (95% CI, 95.2 to 97.4), p<0.001). CONCLUSION: The DLS showed improved sensitivity and similar specificity compared with a retina specialist for DR detection. This demonstrates its potential to support DR screening among Indigenous Australians, an underserved population with a high burden of diabetic eye disease.

来源:Br J Ophthalmol,查看原文。版权归原作者所有。

引用

  1. Chia MA, Hersch F, Sayres R, et al. Validation of a deep learning system for the detection of diabetic retinopathy in Indigenous Australians. Br J Ophthalmol. 2024. PMID: 36746615. doi:10.1136/bjo-2022-322237

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。