您的位置: 首页 > 2021年3月 第36卷 第3期 > 文字全文
2023年7月 第38卷 第7期11
目录

基于眼底彩照的冠心病智能分类系统

Intelligent classification system of coronary heart disease based on fundus color photographs

来源期刊: 眼科学报 | 2021年3月 第36卷 第3期 188-191 发布时间: 收稿时间:2023/5/6 11:55:47 阅读量:3567
作者:
关键词:
冠心病眼底彩照人工智能
coronary heart disease fundus color photograph artificial intelligence
DOI:
10.3978/j.issn.1000-4432.2021.03.17
收稿时间:
 
修订日期:
 
接收日期:
 
目的:探索基于眼底彩照和人工智能构建冠心病智能诊断系统的可行性。方法:于2013—2014年收集广东省人民医院530例患者共2117张眼底彩照,其中冠心病217例共909张眼底彩照。根据患者有无冠心病的情况进行标记,使用Inception-V3深度卷积神经网络训练人工智能模型,随后使用验证数据判断模型的准确率。计算深度卷积网络模型的准确性、一致率、敏感性、特异性和受试者工作特性曲线下面积(area under the curve,AUC)。结果:在2117张眼底彩照中,1903张用于模型训练,214张用于模型的性能评估。在测试集中,该算法的准确性为98.1%,一致率为98.6%,敏感性为100.0%,特异性为96.7%,AUC为0.988(95%CI:0.974~1.000)。结论:眼底彩照联合人工智能技术可精准判定冠心病,该模型具备较高的敏感性和特异性,但须进一步增加样本量,使用大样本量数据验证该模型,排除过拟合的可能性。
Objective: To explore the feasibility of developing a deep learning algorithm for detecting coronary heart diseases based on fundus color photography and artificial intelligence (AI). Methods: A total of 2 117 fundus  color photographs were taken from 530 patients in Guangdong Provincial People’s Hospital from 2013 to 2014,including 909 fundus color photographs from 217 patients with coronary heart disease (CHD). According to whether the patient had coronary heart disease or not, the Inception-V3 depth convolution neural network was used to train the deep learning model, and then the validation data were used to judge the accuracy of the model. The accuracy, consistency rate, sensitivity and specificity of the deep convolution network model and the area under the working characteristic curve (AUC) were calculated. Results: Among the 2 117 fundus color photographs, 1 903 were used for model training, and 214 were used to test the accuracy of the model. In the test dataset, the accuracy of the algorithm was 98.1%, the consistency rate was 98.6%, the sensitivity was 100.0%, and the specificity was 96.7%. The AUC was 0.988 (95% CI, 0.974–1.000). Conclusion: The combination of fundus color photography and artificial intelligence can achieve the accurate diagnosis of the coronary heart disease, and the model has high sensitivity and specificity. However, future studies are warranted to validate our model and exclude the possibility of over-fitting.
冠心病是全球致死率最高的疾病[1],且大多数心脏疾病都属于冠心病[2]。因此,对于如何尽早预测和判断冠心病的高危人群极为重要。准确的预测模型对于降低冠心病高危人群的患病风险具有重大价值,但在临床中仍缺乏较为全面和精准的模型。之前有学者[3-8]分析冠心病患者的临床数据,得出池化队列方程、弗雷明汉和系统性冠状动脉风险评估方程等心血管疾病患病风险的计算公式。这些模型虽然取得了重大突破,但准确性仍待提高,并且模型包含很多有创的抽血检查,如血糖和血脂等,使这些模型难以大规模推广。
视网膜的图像可以通过极为简单、快捷、无创的方法取得。近期有研究[9-12]表明:视网膜情况可能反映心脏的状态,其主要体现在血管的粗细程度、血管弯曲度及微血管改变。然而,目前的研究多局限在相关性研究,缺乏利用眼部特征的冠心病诊断模型构建研究。人工智能技术可智能识别医学影像信息,近年来智能图片识别已取得突破性成果,将该技术应用于医学疾病的判定也是研究热点之一[13-14]
因此,本研究利用人工智能技术构建了一个基于眼底彩照的冠心病的智能诊断系统,检测模型性能,以探究眼底彩照和人工智能在诊断或预测冠心病中的应用价值。

1 材料与方法

1.1 数据收集

本研究遵循赫尔辛基宣言,并获得中山眼科中心伦理委员会的批准。由于本研究回顾性分析的完全匿名化的眼底彩照,故医学伦理委员会批准免知情同意签署。收集2013—2014年来自广东省人民医院和中山眼科中心530名患者共2 117张眼底彩照,其中冠心病217例共909张眼底彩照,包括107例心绞痛患者和110例心肌梗死患者。
扩瞳后收集每名受试者双眼共4张眼底彩照像(Cannon CR-2相机),每只眼分别收集1张以视盘为中心和1张以黄斑为中心的眼底彩照。眼底彩照的排除标准:1)屈光介质混浊,眼底彩照超过50%的区域不能清晰成像;2)有眼部外伤病史;3)有其他眼部疾病,如糖尿病视网膜病变、老年黄斑变性、近视视网膜病变、或青光眼等情况;4 )因身体或精神原因不能配合检查。

1.2 图像处理与模型构建

根据临床诊断将眼底彩照划分为有、无冠心病,每1张经过标注的眼底彩照先预处理,将像素值初始化为0~1,大小调整为299×299,随后图像在水平方向平移0~3°或随机旋转90°、180°或270°进行数据扩增。将数据按照9:1划分为训练集和验证集,采用Inception-V3的模型进行训练。

1.3 统计学处理

采用STATA 14.0软件分析数据。计量资料采用S-W检验检测其是否呈正态分布,如符合正态性,以均数±标准差(x±s)表示,两组间均数差异采用独立样本t检验;如不符合正态性,则使用秩和检验。计数资料采用卡方检验进行统计分析。以准确率、一致性、敏感性、特异性和受试工作特性曲线下面积(area under the curve,AUC)评价模型的性能。P<0.05为差异有统计学意义。

2 结果

受试者年龄(64.9±9.2)岁;男420例,女120例;糖尿病患者87例,无糖尿病者443例。按9:1划分训练集和验证集,1 903张眼底彩照用于模型构建即训练集,214张用于模型测试,测试集和验证集的临床特征见表1。在两个数据集中,有无冠心病受试者的年龄和性别差异均无统计学意义(P>0.05)。
在测试集中,该模型的准确性为98.1%,一致率为98.6%,敏感性为100.0%,特异性为96.7%,AUC为0.988(95%CI:0.974~1.000,图1)。

表1 训练集和验证集受试者的临床特征
Table 1 Clinical characteristics of participants in training and validation dataset

20230506113518_0198.png
20230506113622_2074.png

图1 人工智能模型判定冠心病的ROC曲线
Figure 1 ROC curve of artificial intelligence in diagnosis of coronary heart disease

3 讨论

冠心病的常规检查手段包括心电图、血清生物标志物、冠状动脉计算机化X线体层照相术(computed tomography,CT)血管成像、冠状动脉造影等。冠状动脉CT常被用来做冠心病的无创性筛查,但在医疗资源不足的地区或社区筛查常因缺乏昂贵的CT检查设备而无法进行。冠状动脉造影是诊断冠心病的金标准,但却有创,且需要专业的心内科医生进行操作,更重要的是0.2%~0.9%的患者会出现造影并发症,如心律失常、假性动脉瘤、动脉瘘、心肌梗死和造影剂过敏等,因而难以推广到大范围的筛查和可疑病例的筛选中。本研究为诊断和预测冠心病提供了一种新思路,即采用简单、便捷、无创、费用低廉的眼底彩照作为切入点,与前沿的人工智能技术相结合,实现冠心病的诊断或预测。
本研究报道的冠心病智能预测模型采用深度学习网络框架,并取得了良好的表型。近年来,也有少量使用眼底彩照判定或预测心脑血管疾病相关指标的智能模型,但与本研究观察的冠心病仍存在差异[15-16]。谷歌使用284 355例患者的眼底彩照数据构建智能评估心脑血管危险因素及心脑血管重大事件模型,其中重大心脏事件判定模型的AUC仅为0.70[15]。由于在该数据集中大多数为健康人群,有重大心脏事件的患者仅有631例,这也是造成该模型准确率较低的原因之一。此外,Son等[16]使用20130例患者的眼底彩照智能评估冠状动脉钙化评分,AUC达到82.3%~83.2%。虽然本模型已获得较高的准确率,但需要未来用大样本量的数据进一步验证该模型,排除过拟合的可能。
临床上,眼底彩照常规用于眼底疾病诊治,而并非检测心血管疾病。视网膜眼底图像反应了视网膜血管的概况,而视网膜的血管可能反应心血管的负荷状态,当心脏负荷较高时,会产生高血压类似的血管改变,如视网膜血管僵硬增加和小静脉改变,这些细微改变在眼底图像中肉眼常常是难以识别的。但深度学习算法可在像素级别上发现并捕捉这些微小改变从而用于心血管疾病的诊断与预测。
综上所述,本研究发现通过眼底彩照和人工智能技术结合构建的智能系统用于冠心病的辅助诊断完全可行,特别是在我国心血管疾病患者群体庞大,医疗资源分布不均和基层专科医生的匮乏的环境下,利用眼底彩照和人工智能技术诊断或预测冠心病具有广阔的应用前景。
1、World Health Organization. The top 10 causes of death. 2020[EB/OL]. https://www.who.int/news-room/fact-sheets/detail/the-top-10- causes-of-death. 2021-01-22.World Health Organization. The top 10 causes of death. 2020[EB/OL]. https://www.who.int/news-room/fact-sheets/detail/the-top-10- causes-of-death. 2021-01-22.
2、Drugs & Diseases C. Risk factors for coronary artery disease. 2016[EB/OL]. https://www.healthline.com/health/coronary-artery-disease/riskfactors. 2020-12-29.Drugs & Diseases C. Risk factors for coronary artery disease. 2016[EB/OL]. https://www.healthline.com/health/coronary-artery-disease/riskfactors. 2020-12-29.
3、Stone NJ, Robinson JG, Lichtenstein AH, et al. 2013 ACC/ AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines[ J]. J Am Coll Cardiol, 2014, 63(25 Pt B): 2889-2934.Stone NJ, Robinson JG, Lichtenstein AH, et al. 2013 ACC/ AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines[ J]. J Am Coll Cardiol, 2014, 63(25 Pt B): 2889-2934.
4、Wilson PW, D'Agostino RB, Levy D, et al. Prediction of coronary heart disease using risk factor categories[ J]. Circulation, 1998, 97(18): 1837-1847.Wilson PW, D'Agostino RB, Levy D, et al. Prediction of coronary heart disease using risk factor categories[ J]. Circulation, 1998, 97(18): 1837-1847.
5、National Cholesterol Education Program (NCEP) expert panel on detection, evaluation, and treatment of high blood cholesterol in adults (Adult Treatment Panel III). Third report of the National Cholesterol Education Program (NCEP) expert panel on detection, evaluation, and treatment of high blood cholesterol in adults (Adult Treatment Panel III) final report[ J]. Circulation, 2002, 106(25): 3143-3421.National Cholesterol Education Program (NCEP) expert panel on detection, evaluation, and treatment of high blood cholesterol in adults (Adult Treatment Panel III). Third report of the National Cholesterol Education Program (NCEP) expert panel on detection, evaluation, and treatment of high blood cholesterol in adults (Adult Treatment Panel III) final report[ J]. Circulation, 2002, 106(25): 3143-3421.
6、D'Agostino RB Sr, Vasan RS, Pencina MJ, et al. General cardiovascular risk profile for use in primary care: the Framingham Heart Study[ J]. Circulation, 2008, 117(6): 743-753.D'Agostino RB Sr, Vasan RS, Pencina MJ, et al. General cardiovascular risk profile for use in primary care: the Framingham Heart Study[ J]. Circulation, 2008, 117(6): 743-753.
7、Conroy RM, Py?r?l? K, Fitzgerald AP, et al. Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project[ J]. Eur Heart J, 2003, 24(11): 987-1003.Conroy RM, Py?r?l? K, Fitzgerald AP, et al. Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project[ J]. Eur Heart J, 2003, 24(11): 987-1003.
8、Graham I, Atar D, Borch-Johnsen K, et al. European guidelines on cardiovascular disease prevention in clinical practice: executive summary: Fourth Joint Task Force of the European Society of Cardiology and Other Societies on Cardiovascular Disease Prevention in Clinical Practice (Constituted by representatives of nine societies and by invited experts)[ J]. Eur Heart J, 2007, 28(19): 2375-2414.Graham I, Atar D, Borch-Johnsen K, et al. European guidelines on cardiovascular disease prevention in clinical practice: executive summary: Fourth Joint Task Force of the European Society of Cardiology and Other Societies on Cardiovascular Disease Prevention in Clinical Practice (Constituted by representatives of nine societies and by invited experts)[ J]. Eur Heart J, 2007, 28(19): 2375-2414.
9、Seidelmann SB, Claggett B, Bravo PE, et al. Retinal vessel calibers in predicting long-term cardiovascular outcomes: the atherosclerosis risk in communities study[ J]. Circulation, 2016, 134(18): 1328-1338.Seidelmann SB, Claggett B, Bravo PE, et al. Retinal vessel calibers in predicting long-term cardiovascular outcomes: the atherosclerosis risk in communities study[ J]. Circulation, 2016, 134(18): 1328-1338.
10、Wong TY, Kamineni A, Klein R, et al. Quantitative retinal venular caliber and risk of cardiovascular disease in older persons: the cardiovascular health study[ J]. Arch Intern Med, 2006, 166(21): 2388-2394.Wong TY, Kamineni A, Klein R, et al. Quantitative retinal venular caliber and risk of cardiovascular disease in older persons: the cardiovascular health study[ J]. Arch Intern Med, 2006, 166(21): 2388-2394.
11、McGeechan K, Liew G, Macaskill P, et al. Meta-analysis: retinal vessel caliber and risk for coronary heart disease[ J]. Ann Intern Med, 2009, 151(6): 404-413.McGeechan K, Liew G, Macaskill P, et al. Meta-analysis: retinal vessel caliber and risk for coronary heart disease[ J]. Ann Intern Med, 2009, 151(6): 404-413.
12、Liew G, Wong TY, Mitchell P, et al. Retinopathy predicts coronary heart disease mortality[ J]. Heart, 2009, 95(5): 391-394.Liew G, Wong TY, Mitchell P, et al. Retinopathy predicts coronary heart disease mortality[ J]. Heart, 2009, 95(5): 391-394.
13、吴乐正. “眼与人工智能”迎来时代挑战的新台阶[ J]. 眼科学 报, 2021, 36(1): 1-3.
WU LZ. “Eye and artificial intelligence” ushers in a new step of challenge of the times[ J]. Yan Ke Xue Bao, 2021, 36(1): 1-3.
吴乐正. “眼与人工智能”迎来时代挑战的新台阶[ J]. 眼科学 报, 2021, 36(1): 1-3.
WU LZ. “Eye and artificial intelligence” ushers in a new step of challenge of the times[ J]. Yan Ke Xue Bao, 2021, 36(1): 1-3.
14、吴晓航, 刘力学, 陈睛晶, 等. 白内障人工智能辅助诊断系统在社 区筛查中的应用效果[ J]. 眼科学报, 2021, 36(1): 4-9.
WU XH, LIU LX, CHEN JJ, et al. Application of artificial intelligence-assisted diagnostic system for community-based cataract screening[ J]. Yan Ke Xue Bao, 2021, 36(1): 4-9.
吴晓航, 刘力学, 陈睛晶, 等. 白内障人工智能辅助诊断系统在社 区筛查中的应用效果[ J]. 眼科学报, 2021, 36(1): 4-9.
WU XH, LIU LX, CHEN JJ, et al. Application of artificial intelligence-assisted diagnostic system for community-based cataract screening[ J]. Yan Ke Xue Bao, 2021, 36(1): 4-9.
15、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.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.
16、Son J, Shin JY, Chun EJ, et al. Predicting high coronary artery calcium score from retinal fundus images with deep learning algorithms[ J]. Transl Vis Sci Technol, 2020, 9(6): 28.Son J, Shin JY, Chun EJ, et al. Predicting high coronary artery calcium score from retinal fundus images with deep learning algorithms[ J]. Transl Vis Sci Technol, 2020, 9(6): 28.
1、国家重点研发计划(2018YFC0116500),国家自然科学基金(81420108008),广东省科技计划项目(2013B20400003)。
This work was supported by the National Key R&D Program (2018YFC0116500), National Natural Science Foundation (81420108008), Science and Technology Planning Project of Guangdong Province (2013B20400003), China.()
上一篇
下一篇
其他期刊
  • 眼科学报

    主管:中华人民共和国教育部
    主办:中山大学
    承办:中山大学中山眼科中心
    主编:林浩添
    主管:中华人民共和国教育部
    主办:中山大学
    浏览
  • Eye Science

    主管:中华人民共和国教育部
    主办:中山大学
    承办:中山大学中山眼科中心
    主编:林浩添
    主管:中华人民共和国教育部
    主办:中山大学
    浏览
推荐阅读
出版者信息
目录