切换至 "中华医学电子期刊资源库"

中华诊断学电子杂志 ›› 2024, Vol. 12 ›› Issue (01) : 1 -8. doi: 10.3877/cma.j.issn.2095-655X.2024.01.001

机器学习

基于冠状动脉CT血管成像的影像组学列线图鉴别诊断易损斑块的价值
王林源1, 熊鑫1, 杨坤1, 邓勇志1,()   
  1. 1. 030024 太原,山西医科大学附属心血管病医院,山西省心血管病医院(研究所),山西省心血管病临床医学研究中心心脏大血管外科
  • 收稿日期:2023-07-07 出版日期:2024-02-26
  • 通信作者: 邓勇志
  • 基金资助:
    山西省医学重点科研项目重大科技攻关专项(2021XM04)

The value of radiomics nomogram based on coronary CT angiography in differential diagnosis of vulnerable plaques

Linyuan Wang1, Xin Xiong1, Kun Yang1, Yongzhi Deng1,()   

  1. 1. Department of Cardiovascular Surgery, the Affiliated Cardiovascular Hospital of Shanxi Medical University, Shanxi Cardiovascular Hospital (Institute), Shanxi Cardiovascular Disease Clinical Medical Research Center, Taiyuan 030024, China
  • Received:2023-07-07 Published:2024-02-26
  • Corresponding author: Yongzhi Deng
引用本文:

王林源, 熊鑫, 杨坤, 邓勇志. 基于冠状动脉CT血管成像的影像组学列线图鉴别诊断易损斑块的价值[J]. 中华诊断学电子杂志, 2024, 12(01): 1-8.

Linyuan Wang, Xin Xiong, Kun Yang, Yongzhi Deng. The value of radiomics nomogram based on coronary CT angiography in differential diagnosis of vulnerable plaques[J]. Chinese Journal of Diagnostics(Electronic Edition), 2024, 12(01): 1-8.

目的

探讨基于冠状动脉CT血管成像(CCTA)的影像组学列线图鉴别诊断冠状动脉疾病(CAD)患者易损斑块的价值。

方法

回顾性分析2018年1月至2022年12月在山西医科大学附属心血管病医院接受CCTA检查的93例CAD患者资料,共95个非钙化斑块,其中包括43个易损斑块,52个稳定斑块。以8∶2的比例随机将数据划分为训练集(n=76)及测试集(n=19)。应用Mann-Whitney U检验、相关系数法和最小绝对收缩选择算子(LASSO)回归选择最优的影像组学特征,构建影像组学模型并计算影像组学评分(Radscore)。同时,根据临床资料建立临床模型。结合临床因素和Radscore,构建影像组学列线图。采用受试者工作特征(ROC)曲线、校准曲线及决策曲线分析评估模型的性能。

结果

最终筛选出13个具有非零系数的最优特征。结合Radscore和临床因素的列线图模型在训练集和测试集中的AUC分别为1.000,0.922,敏感度为1.000,1.000,特异度为1.000,0.800,准确度为1.000,0.895。校准曲线显示列线图预测结果与实际结果之间具有良好的一致性。决策曲线分析表明,列线图比临床模型更具有临床应用价值。

结论

结合临床因素和影像组学特征的影像组学列线图能够准确、客观地鉴别诊断易损斑块,具有良好的诊断性能,可用于指导临床决策。

Objective

To investigate the value of radiomics nomogram based on coronary CT angiography (CCTA) in differential diagnosis of vulnerable plaques in patients with coronary artery disease (CAD).

Methods

From January 2018 to December 2022, 93 CAD patients who had undergone CCTA scans at the Affiliated Cardiovascular Hospital of Shanxi Medical University with complete data were enrolled. There were 95 non-calcified plaques, including 43 vulnerable plaques and 52 stable plaques. In an 8: 2 ratio, the lesions were randomly separated into a training set (n=76) and a testing set (n=19). To choose radiomics features appropriate for a CCTA-image-based radiomics signature, the Mann-Whitney U test, the correlation coefficient approach, and least absolute shrinkage and selection operator (LASSO) were used, and a radiomics score (Radscore) was created. Using clinical data and CCTA findings, a clinical model was constructed. Subsequently, the separate clinical parameters and Radscore were integrated to create a radiomics nomogram. The receiver operating characteristics (ROC) curve, calibration curve, and decision curve analysis were used to evaluate the performance of the radiomics signature, clinical model, and nomogram.

Results

Finally, 13 best features with non-zero coefficients were selected. In the training and test sets, the AUC of the nomogram model combined with Radscore and clinical variables were 1.000 and 0.922, sensitivity were 1.000 and 1.000, specificity were 1.000 and 0.800, and accuracy were 1.000 and 0.895, respectively. The calibration curve demonstrated good agreement between expected and actual results. The decision curve analysis revealed that the nomogram had greater clinical application value compared to the clinical model.

Conclusion

Combined with clinical factors and imaging features, the radiomics nomogram can accurately and objectively identify vulnerable plaques, and has good diagnostic performance, which can be used to guide clinical decision-making.

图1 55岁男性冠状动脉疾病患者左冠状动脉前降支斑块勾画示意图注:a图为冠状动脉CT血管成像原始图像;b图为手动分割斑块(箭头所示)
表1 训练集和测试集CAD患者临床特征和CCTA征象分析结果
项目 训练集(n=76) 统计量 P 测试集(n=19) 统计量 P
稳定斑块(n=42) 易损斑块(n=34) 稳定斑块(n=10) 易损斑块(n=9)
年龄(岁,±s) 55.83±9.73 57.03±11.30 t=0.496 0.622 58.00±9.17 50.33±13.65 t=-1.452 0.165
男[例(%)] 27(64.29) 31(91.18) χ2=7.517 0.006 6(60.00) 7(77.78)   0.628*
BMI[kg/m2M(Q1Q3),±s] 24.92(22.77,26.31) 24.66(22.27,27.52) Z=0.136 0.892 24.27±3.60 26.29±4.38 t=-1.098 0.287
高血压病[例(%)] 23(54.76) 25(73.53) χ2=2.844 0.092 4(40.00) 5(55.56)   0.656*
糖尿病[例(%)] 6(14.29) 8(23.53) χ2=1.068 0.301 2(20.00) 3(33.33)   0.628*
高脂血症[例(%)] 26(61.90) 29(85.29) χ2=5.140 0.023 6(60.00) 7(77.78)   0.628*
吸烟[例(%)] 16(38.10) 15(44.12) χ2=0.282 0.595 4(40.00) 5(55.56)   0.656*
总胆固醇[mmol/L,M(Q1Q3),±s] 4.10(3.53,4.75) 4.03(3.41,4.79) Z=0.016 0.987 4.84±1.17 5.12±1.92 t=-0.387 0.704
总甘油三酯[mmol/L,M(Q1Q3)] 1.49(0.97,2.09) 1.53(0.99,2.09) Z=0.199 0.843 1.46(0.985,3.22) 1.96(1.64,3.28) Z=-1.061 0.288
HDL-C[mmol/L,M(Q1Q3),±s] 1.00(0.87,1.16) 0.97(0.74,1.13) Z=1.102 0.270 1.18±0.37 1.02±0.26 t=-1.081 0.295
LDL-C[mmol/L,M(Q1Q3),±s] 2.61(1.93,2.87) 2.37(1.87,2.83) Z=1.008 0.313 2.77±0.62 3.05±1.24 t=0.628 0.539
传统斑块特征                
长度[mm,M(Q1Q3),±s] 13.72(12.18,16.60) 14.83(12.03,17.72) Z=0.757 0.449 15.04±1.86 16.87±2.68 t=1.738 0.100
体积[mm3M(Q1Q3)] 68.60(54.30,88.19) 93.83(60.27,127.26) Z=2.659 0.008 61.66(33.06,80.61) 91.97(69.60,166.79) Z=-1.960 0.050
斑块位置[例(%)]       0.297*       0.025*
前降支 36(85.71) 25(73.53)     8(80.00) 5(55.56)    
右冠状动脉 6(14.29) 8(23.53)     0 4(44.44)    
回旋支 0 1(2.94)     2(20.00) 0    
高危斑块特征[例(%)]                
低密度斑块 23(54.76) 34(100.00) χ2=20.510 <0.001 4(40.00) 9(100.00)   0.011*
正性重构 6(14.29) 27(79.41) χ2=32.439 <0.001 0 5(55.56)   0.011*
"餐巾环"征 1(2.38) 14(41.18) χ2=17.850 <0.001 0 4(44.44)   0.033*
图2 LASSO回归分析通过10折交叉验证的CAD患者基于CCTA的影像组学特征筛选过程图注:a图为基于CCTA提取的1 906个影像组学特征进行统计分析后的P值结果,以进行特征的初步筛选;firstorder为一阶统计特征;gldm为灰度依赖矩阵特征;glszm为灰度区域大小矩阵特征;shape为形状特征;b图为随参数λ变化的均方误差;c图为随参数λ变化的特征系数分布;d图为最终筛选的13个最优特征及其系数。a~d图通过10折交叉验证,不断调整超参数λ,以获取最优特征组合。LASSO为最小绝对收缩选择算子;CAD为冠状动脉疾病;CCTA为冠状动脉CT血管成像
表2 不同影像组学模型在训练集和测试集中鉴别诊断易损斑块的效能
图3 不同影像组学模型在测试集中鉴别诊断易损斑块的ROC曲线注:ROC为受试者工作特征;AUC为ROC曲线下面积
表3 临床模型、影像组学模型和列线图模型鉴别诊断易损斑块的ROC曲线结果
图4 基于影像组学评分与临床因素建立的列线图
图5 列线图、影像组学模型及临床模型鉴别诊断易损斑块的ROC曲线注:a图为训练集;b图为测试集;ROC为受试者工作特征;AUC为ROC曲线下面积
图6 列线图、影像组学及临床模型鉴别诊断易损斑块的校准曲线注:a图为训练集;b图为测试集
图7 列线图、影像组学及临床模型鉴别诊断易损斑块的决策曲线注:a图为训练集;b图为测试集
[1]
Bhatt DL, Lopes RD, Harrington RA.Diagnosis and treatment of acute coronary syndromes:a review[J].JAMA2022327(7):662-675.DOI:10.1001/jama.2022.0358.
[2]
Khan MH, Rochlani Y, Yandrapalli S, et al.Vulnerable plaque:a review of current concepts in pathophysiology and imaging[J].Cardiol Rev202028(1):3-9.DOI:10.1097/CRD.0000000000000238.
[3]
Grodecki K, Cadet S, Staruch AD, et al.Noncalcified plaque burden quantified from coronary computed tomography angiography improves prediction of side branch occlusion after main vessel stenting in bifurcation lesions:results from the CT-PRECISION registry[J].Clin Res Cardiol2021110(1):114-123.DOI:10.1007/s00392-020-01658-1.
[4]
Greenland P, Blaha MJ, Budoff MJ, et al.Coronary calcium score and cardiovascular risk[J].J Am Coll Cardiol201872(4):434-447.DOI:10.1016/j.jacc.2018.05.027.
[5]
van Veelen A, van der Sangen N, Delewi R, et al. Detection of vulnerable coronary plaques using invasive and non-invasive imaging modalities[J].J Clin Med202211(5):1361.DOI:10.3390/jcm11051361.
[6]
Masuda T, Nakaura T, Funama Y, et al.Deep learning with convolutional neural network for estimation of the characterisation of coronary plaques:validation using IB-IVUS[J].Radiography (Lond)202228(3):661-662.DOI:10.1016/j.radi.2022.05.002.
[7]
Lee JM, Choi KH, Koo BK, et al.Prognostic implications of plaque characteristics and stenosis severity in patients with coronary artery disease[J].J Am Coll Cardiol201973(19):2413-2424.DOI:10.1016/j.jacc.2019.02.060.
[8]
Motoyama S, Ito H, Sarai M, et al.Plaque characterization by coronary computed tomography angiography and the likelihood of acute coronary events in mid-term follow-up[J].J Am Coll Cardiol201566(4):337-346.DOI:10.1016/j.jacc.2015.05.069.
[9]
Kiriᶊli HA, Schaap M, Metz CT, et al.Standardized evaluation framework for evaluating coronary artery stenosis detection,stenosis quantification and lumen segmentation algorithms in computed tomography angiography[J].Med Image Anal201317(8):859-876.DOI:10.1016/j.media.2013.05.007.
[10]
Cury RC, Abbara S, Achenbach S, et al.CAD-RADS(TM) Coronary Artery Disease-Reporting and Data System.An expert consensus document of the Society of Cardiovascular Computed Tomography (SCCT),the American College of Radiology (ACR) and the North American Society for Cardiovascular Imaging (NASCI).Endorsed by the American College of Cardiology[J].J Cardiovasc Comput Tomogr201610(4):269-281.DOI:10.1016/j.jcct.2016.04.005.
[11]
Xu P, Xue Y, Schoepf UJ, et al.Radiomics:the next frontier of cardiac computed tomography[J].Circ Cardiovasc Imaging202114(3):e011747.DOI:10.1161/CIRCIMAGING.120.011747.
[12]
Chen Q, Pan T, Yin X, et al.CT texture analysis of vulnerable plaques on optical coherence tomography[J].Eur J Radiol2021(136):109551.DOI:10.1016/j.ejrad.2021.109551.
[13]
Jiang XY, Shao ZQ, Chai YT, et al.Non-contrast CT-based radiomic signature of pericoronary adipose tissue for screening non-calcified plaque[J].Phys Med Biol202267(10) DOI:10.1088/1361-6560/ac69a7.
[14]
Shaw LJ, Blankstein R, Bax JJ, et al.Society of Cardiovascular Computed Tomography/North American Society of Cardiovascular Imaging-expert consensus document on coronary CT imaging of atherosclerotic plaque[J].J Cardiovasc Comput Tomogr202115(2):93-109.DOI:10.1016/j.jcct.2020.11.002.
[15]
Roth GA, Mensah GA, Johnson CO, et al.Global burden of cardiovascular diseases and risk factors,1990-2019:update from the GBD 2019 study[J].J Am Coll Cardiol202076(25):2982-3021.DOI:10.1016/j.jacc.2020.11.010.
[16]
Khan MA, Hashim MJ, Mustafa H, et al.Global epidemiology of ischemic heart disease:results from the global burden of disease study[J].Cureus202012(7):e9349.DOI:10.7759/cureus.9349.
[17]
Rothwell PM, Eliasziw M, Gutnikov SA, et al.Analysis of pooled data from the randomised controlled trials of endarterectomy for symptomatic carotid stenosis[J].Lancet2003361(9352):107-116.DOI:10.1016/s0140-6736(03)12228-3.
[18]
Hoffmann U, Ferencik M, Udelson JE, et al.Prognostic value of noninvasive cardiovascular testing in patients with stable chest pain:insights from the PROMISE trial (prospective multicenter imaging study for evaluation of chest pain)[J].Circulation2017135(24):2320-2332.DOI:10.1161/CIRCULATIONAHA.116.024360.
[19]
Bittencourt MS, Hulten E, Ghoshhajra B, et al.Prognostic value of nonobstructive and obstructive coronary artery disease detected by coronary computed tomography angiography to identify cardiovascular events[J].Circ Cardiovasc Imaging20147(2):282-291.DOI:10.1161/CIRCIMAGING.113.001047.
[20]
Achenbach S, Ropers D, Hoffmann U, et al.Assessment of coronary remodeling in stenotic and nonstenotic coronary atherosclerotic lesions by multidetector spiral computed tomography[J].J Am Coll Cardiol200443(5):842-847.DOI:10.1016/j.jacc.2003.09.053.
[21]
Bittner DO, Mayrhofer T, Puchner SB, et al.Coronary computed tomography angiography-specific definitions of high-risk plaque features improve detection of acute coronary syndrome[J].Circ Cardiovasc Imaging201811(8):e007657.DOI:10.1161/CIRCIMAGING.118.007657.
[22]
Theofilis P, Sagris M, Antonopoulos AS, et al.Non-invasive modalities in the assessment of vulnerable coronary atherosclerotic plaques[J].Tomography20228(4):1742-1758.DOI:10.3390/tomography8040147.
[23]
Knuuti J, Wijns W, Saraste A, et al.2019 ESC guidelines for the diagnosis and management of chronic coronary syndromes[J].Eur Heart J202041(3):407-477.DOI:10.1093/eurheartj/ehz425.
[24]
Maroules CD, Hamilton-Craig C, Branch K, et al.Coronary Artery Disease Reporting and Data System (CAD-RADS(TM)):inter-observer agreement for assessment categories and modifiers[J].J Cardiovasc Comput Tomogr201812(2):125-130.DOI:10.1016/j.jcct.2017.11.014.
[25]
Kolossváry M, Kellermayer M, Merkely B, et al.Cardiac computed tomography radiomics:a comprehensive review on radiomic techniques[J].J Thorac Imaging201833(1):26-34.DOI:10.1097/RTI.0000000000000268.
[26]
Kolossváry M, Park J, Bang JI, et al.Identification of invasive and radionuclide imaging markers of coronary plaque vulnerability using radiomic analysis of coronary computed tomography angiography[J].Eur Heart J Cardiovasc Imaging201920(11):1250-1258.DOI:10.1093/ehjci/jez033.
[27]
Kolossváry M, Karády J, Szilveszter B, et al.Radiomic features are superior to conventional quantitative computed tomographic metrics to identify coronary plaques with napkin-ring sign[J].Circ Cardiovasc Imaging201710(12):e006843.DOI:10.1161/CIRCIMAGING.117.006843.
[28]
Lin A, Kolossváry M, Cadet S, et al.Radiomics-based precision phenotyping identifies unstable coronary plaques from computed tomography angiography[J].JACC Cardiovasc Imaging202215(5):859-871.DOI:10.1016/j.jcmg.2021.11.016.
[29]
Chen Q, Pan T, Wang YN, et al.A coronary CT angiography radiomics model to identify vulnerable plaque and predict cardiovascular events[J].Radiology2023307(2):e221693.DOI:10.1148/radiol.221693.
[30]
Ebersberger U, Bauer MJ, Straube F, et al.Gender differences in epicardial adipose tissue and plaque composition by coronary CT angiography:association with cardiovascular outcome[J].Diagnostics(Basel)202313(4):624.DOI:10.3390/diagnostics13040624.
[31]
Andreini D, Magnoni M, Conte E, et al.Coronary plaque features on CTA can identify patients at increased risk of cardiovascular events[J].JACC Cardiovasc Imaging202013(8):1704-1717.DOI:10.1016/j.jcmg.2019.06.019.
[32]
Ridker PM, Bhatt DL, Pradhan AD, et al.Inflammation and cholesterol as predictors of cardiovascular events among patients receiving statin therapy:a collaborative analysis of three randomised trials[J].Lancet2023401(10384):1293-1301.DOI:10.1016/S0140-6736(23)00215-5.
[1] 张茜, 陈佳慧, 高雪萌, 赵傲雪, 黄瑛. 基于高帧频超声造影的影像组学特征鉴别诊断甲状腺结节良恶性的价值[J]. 中华医学超声杂志(电子版), 2023, 20(09): 895-903.
[2] 孙帼, 谢迎东, 徐超丽, 杨斌. 超声联合临床特征的列线图模型预测甲状腺乳头状癌淋巴结转移的价值[J]. 中华医学超声杂志(电子版), 2023, 20(07): 734-742.
[3] 唐玮, 何融泉, 黄素宁. 深度学习在乳腺癌影像诊疗和预后预测中的应用[J]. 中华乳腺病杂志(电子版), 2023, 17(06): 323-328.
[4] 吴佩琪. 乳腺癌瘤周影像组学研究进展[J]. 中华乳腺病杂志(电子版), 2023, 17(05): 301-304.
[5] 张秋彬, 张楠, 林清婷, 徐军, 朱华栋, 姜辉. 急性胰腺炎合并急性肾损伤患者的预后评估[J]. 中华危重症医学杂志(电子版), 2023, 16(05): 382-389.
[6] 郑鹏, 吴赛萍, 谢秀璋, 史庆丰. 术前预测感染性肾结石列线图模型的构建及验证[J]. 中华实验和临床感染病杂志(电子版), 2023, 17(05): 299-306.
[7] 王晓梅, 刘冰, 马丽琼, 卢祖静, 苗建军. 基于LASSO-Cox回归分析的非轻症急性胰腺炎死亡风险列线图预测模型的建立和临床应用效果分析[J]. 中华普通外科学文献(电子版), 2024, 18(01): 44-50.
[8] 甄子铂, 刘金虎. 基于列线图模型探究静脉全身麻醉腹腔镜胆囊切除术患者术后肠道功能紊乱的影响因素[J]. 中华普外科手术学杂志(电子版), 2024, 18(01): 61-65.
[9] 唐旭, 韩冰, 刘威, 陈茹星. 结直肠癌根治术后隐匿性肝转移危险因素分析及预测模型构建[J]. 中华普外科手术学杂志(电子版), 2024, 18(01): 16-20.
[10] 王万川, 麦广智, 张晓槟, 冯家豪. 腹腔镜下LAR术治疗低位直肠癌保肛术中钉仓数量与术后LARS的关系[J]. 中华普外科手术学杂志(电子版), 2023, 17(05): 491-496.
[11] 徐慧新, 刘波, 唐立钧. 体外冲击波治疗>1 cm输尿管上段结石失败的预测模型建立[J]. 中华腔镜泌尿外科杂志(电子版), 2023, 17(05): 506-511.
[12] 孙振, 谭天华, 郑洋洋, 李喆, 宋京海. 基于术前纤维蛋白原与白蛋白比值构建肝癌微血管侵犯的预测模型[J]. 中华肝脏外科手术学电子杂志, 2024, 13(01): 27-32.
[13] 潘晓帆, 徐勤义, 陆瑨, 王丹, 刘路路, 董万利. 颅内动脉瘤破裂介入术后并发脑疝的风险因素分析[J]. 中华脑科疾病与康复杂志(电子版), 2024, 14(01): 37-44.
[14] 黄利军, 熊志勇, 李丹凤. 开发和验证预测脑肿瘤术后消化道出血的列线图:单中心研究[J]. 中华脑科疾病与康复杂志(电子版), 2023, 13(06): 321-326.
[15] 薛文欣, 常婉英, 肖瑶, 张盼盼, 党晓智, 宋宏萍. 乳腺恶性非肿块型病变列线图预测模型的构建及外部验证[J]. 中华临床医师杂志(电子版), 2023, 17(10): 1045-1050.
阅读次数
全文


摘要