叶一,冯伟,丁耀东,陈青,张阳,林立,马彤,王斌,常献刚,戈宗元,王晓怡,蔡龙军,曾勇

• 1:首都医科大学附属北京安贞医院冠心病中心
• 2:北京鹰瞳科技发展股份有限公司
• 3:北京智精灵科技有限公司

摘要(Abstract)：

目的 开发基于眼底视网膜图像的深度学习模型，提高冠心病（CHD）患者轻度认知障碍（MCI）的检出率，实现早期干预并改善预后。方法 本研究为单中心横断面研究，回顾性纳入2021年11月至2022年12月首都医科大学附属北京安贞医院由冠状动脉造影（至少1支冠状动脉血管狭窄≥50%）诊断为CHD的患者。整个数据集被按照8﹕2的比例随机分为训练组和测试组进行模型开发，之后采用时间验证的方法纳入同中心2023年1月至2023年4月的患者数据对模型进行验证。MCI诊断标准为迷你精神状态检查（MMSE）<27分或蒙特利尔认知评估（MoCA）<26分。研究使用4种不同的卷积神经网络（CNN）架构对眼底图像进行训练，并通过模型集成建立了MCI检测的综合视觉模型。采用受试者工作曲线（ROC）的曲线下面积（AUC）、敏感度和特异度评估人工智能模型的性能。结果 纳入3 368例CHD患者，收集5 880张符合条件的眼底图像。基于MMSE量表结果为算法标签，其中男性2 898例，MCI患者527例。深度学习模型在测试组的AUC为0.733(95%CI 0.688～0.778)，采用敏感度和特异度之和最大的工作点，算法在测试组的敏感度为0.577(95%CI 0.528～0.625)，特异度为0.758(95%CI 0.714～0.802)，对应验证组的AUC为0.710(95%CI0.601～0.818)。基于MoCA量表结果为算法标签，其中男性2437例，MCI患者1626例。深度学习模型在测试组的AUC为0.702(95%CI 0.671～0.733)。选取敏感度和特异度之和最大的工作点，算法的敏感度为0.749(95%CI 0.719～0.778)，特异度为0.561(95%CI 0.527～0.595)，对应验证组的AUC为0.674(95%CI0.622～0.726)。结论 基于眼底图像的深度学习算法模型具有较好的诊断性能，有可能作为一种无创、便捷和快速筛查CHD人群中的MCI的新方法。

关键词(KeyWords)： 轻度认知障碍;冠心病;眼底图像;深度学习

Abstract:

Keywords:

基金项目(Foundation): 中华医学会心血管病学分会（CSC）临床研究专项基金项目（CSCF2023A03）;; 首都卫生发展科研专项项目[首发（2024-1-2061）];; 北京市医管中心项目[扬帆计划（YGLX202324）]

作者(Author): 叶一,冯伟,丁耀东,陈青,张阳,林立,马彤,王斌,常献刚,戈宗元,王晓怡,蔡龙军,曾勇

参考文献(References)：

• [1]Anderson ND.State of the science on mild cognitive impairment(MCI)[J].CNS Spectr,2019,24(1):78-87.DOI:10.1017/S1092852918001347.
• [2]Burks HB,des Bordes JKA,Chadha R,et al.Quality of life assessment in older adults with dementia:a systematic review[J].Dement Geriatr Cogn Disord,2021,50(2):103-110.DOI:10.1159/000515317.
• [3]张瑞涛,田振宇,曾勇,等.一种新型血管内冲击波球囊用于严重冠状动脉钙化病变预处理的有效性和安全性研究[J].中国介入心脏病学杂志,2025,33(2):61-70.DOI:10.3969/j.issn.1004-8812.2025.02.001.
• [4]Arntzen KA,Schirmer H,Wilsgaard T,et al.Impact of cardiovascular risk factors on cognitive function:the Troms?study[J].Eur J Neurol,2011,18(5):737-743.DOI:10.1111/j.1468-1331.2010.03263.x.
• [5]Liang X,Huang Y,Han X.Associations between coronary heart disease and risk of cognitive impairment:a meta-analysis[J].Brain Behav,2021,11(5):e02108.DOI:10.1002/brb3.2108.
• [6]Leng X,Espeland MA,Manson JE,et al.Cognitive function and changes in cognitive function as predictors of incident cardiovascular disease:the women's health initiative memory study[J].J Gerontol A Biol Sci Med Sci,2018,73(6):779-785.DOI:10.1093/gerona/glx138.
• [7]Zhuang L,Yang Y,Gao J.Cognitive assessment tools for mild cognitive impairment screening[J].J Neurol,2021,268(5):1615-1622.DOI:10.1007/s00415-019-09506-7.
• [8]Nasreddine ZS,Phillips NA,Bédirian V,et al.The Montreal Cognitive Assessment,MoCA:a brief screening tool for mild cognitive impairment[J].J Am Geriatr Soc,2005,53(4):695-699.DOI:10.1111/j.1532-5415.2005.53221.x.
• [9]Li F,Liu Z,Chen H,et al.Automatic detection of diabetic retinopathy in retinal fundus photographs based on deep learning algorithm[J].Transl Vis Sci Technol,2019,8(6):4.DOI:10.1167/tvst.8.6.4.
• [10]Cheung CY,Xu D,Cheng CY,et al.A deep-learning system for the assessment of cardiovascular disease risk via the measurement of retinal-vessel calibre[J].Nat Biomed Eng,2021,5(6):498-508.DOI:10.1038/s41551-020-00626-4.
• [11]Hua R,Xiong J,Li G,et al.Development and validationof a deep learning algorithm based on fundus photographs for estimating the CAIDE dementia risk score[J].Age Ageing,2022,51(12):afac282.DOI:10.1093/ageing/afac282.
• [12]Shi XH,Ju L,Dong L,et al.Deep learning models for the screening of cognitive impairment using multimodal fundus images[J].Ophthalmol Retina,2024,8(7):666-677.DOI:10.1016/j.oret.2024.01.019.
• [13]McGeechan K,Liew G,Macaskill P,et al.Risk prediction of coronary heart disease based on retinal vascular caliber(from the Atherosclerosis Risk In Communities[ARIC]Study)[J].Am J Cardiol,2008,102(1):58-63.DOI:10.1016/j.amjcard.2008.02.094.
• [14]Arevalo-Rodriguez I,Smailagic N,Roqué-Figuls M,et al.Mini-Mental State Examination(MMSE)for the early detection of dementia in people with mild cognitive impairment(MCI)[J].Cochrane Database Syst Rev,2021,7(7):CD010783.DOI:10.1002/14651858.CD010783.pub3.
• [15]He K,Zhang X,Ren S,et al.Deep residual learning for image recognition[J].IEEE,2016.DOI:10.1109/CVPR.2016.90.
• [16]Collins GS,Dhiman P,Ma J,et al.Evaluation o f clinical prediction models(part 1):from development to external validation[J].BMJ,2024,384:e074819.DOI:10.1136/bmj-2023-074819.
• [17]Kwak K,Giovanello KS,Bozoki A,et al.Subtyping of mild cognitive impairment using a deep learning model based on brain atrophy patterns[J].Cell Rep Med,2021,2(12):100467.DOI:10.1016/j.xcrm.2021.100467.
• [18]Basaia S,Agosta F,Wagner L,et al.Automated classification of Alzheimer's disease and mild cognitive impairment using a single MRI and deep neural networks[J].Neuroimage Clin,2019,21:101645.DOI:10.1016/j.nicl.2018.101645.
• [19]Kashani AH,Asanad S,Chan JW,et al.Past,present and future role of retinal imaging in neurodegenerative disease[J].Prog Retin Eye Res,2021,83:100938.DOI:10.1016/j.preteyeres.2020.100938.
• [20] Zhang K,Liu X,Xu J,et al.Deep-learning models for the detection and incidence prediction of chronic kidney disease and type 2 diabetes from retinal fundus images[J].Nat Biomed Eng,2021,5(6):533-545.DOI:10.1038/s41551-021-00745-6.
• [21]Poplin R,Varadarajan AV,Blumer K,et al.Prediction of cardiovascular risk factors from retinal fundus photographs via deep learning[J].Nat Biomed Eng,2018,2(3):158-164.DOI:10.1038/s41551-018-0195-0.
• [22]Cheung CY,Ran AR,Wang S,et al.A deep learning model for detection of Alzheimer's disease based on retinal photographs:a retrospective,multicentre case-control study[J].Lancet Digit Health,2022,4(11):e806-e815.DOI:10.1016/S2589-7500(22)00169-8.
• [23]Safiri S,Karamzad N,Singh K,et al.Burden of ischemic heart disease and its attributable risk factors in 204 countries and territories,1990-2019[J].Eur J Prev Cardiol,2022,29(2):420-431.DOI:10.1093/eurjpc/zwab213.
• [24]张阿莲,范例,卓杨,等.冠状动脉临界病变患者光学相干断层成像检出分层斑块的临床意义[J].中国介入心脏病学杂志,2025,33(3):155-162.DOI:10.3969/j.issn.1004-8812.2025.03.006.
• [25]Yusuf S,Islam S,Chow CK,et al.Use of secondary prevention drugs for cardiovascular disease in the community in highincome,middle-income,and low-income countries(the PURE Study):a prospective epidemiological survey[J].Lancet,2011,378(9798):1231-1243.DOI:10.1016/S0140-6736(11)61215-4.
• [26]Fihn SD,Gardin JM,Abrams J,et al.2012 ACCF/AHA/ACP/AATS/PCNA/SCAI/STS guideline for the diagnosis and management of patients with stable ischemic heart disease:executive summary:a report of the American College of Cardiology Foundation/American Heart Association task force on practice guidelines,and the American College of Physicians,American Association for Thoracic Surgery,Preventive Cardiovascular Nurses Association,Society for Cardiovascular Angiography and Interventions,and Society of Thoracic Surgeons[J].Circulation,2012,126(25):3097-3137.DOI:10.1161/CIR.0b013e3182776f83.
• [27]Shioji K,Izuhara M,Mitsuoka H,et al.Achievement rates of Japan Atherosclerosis Society Guidelines 2007LDL-cholesterol goals with rosuvastatin or atorvastatin in patients who had not achieved their goal with atorvastatin[J].Cardiovasc Ther,2014,32(3):97-104.DOI:10.1111/1755-5922.12066.
• [28] Jortveit J,Halvorsen S,Kaldal A,et al.Unsatisfactory risk factor control and high rate of new cardiovascular events in patients with myocardial infarction and prior coronary artery disease[J].BMC Cardiovasc Disord,2019,19(1):71.DOI:10.1186/s12872-019-1062-y.
• [29]Newman MF,Kirchner JL,Phillips-Bute B,et al.Longitudinal assessment of neuro cognitive function after coronary-artery bypass surgery[J].N Engl J Med,2001,344(6):395-402.DOI:10.1056/NEJM200102083440601.