人工智能在心血管超声应用中的现状和前景(2)

3.2 心肌运动和应变测量。Kusunose等[7]研究发现与传统二维超声心动图相比，利用深度卷积神经网络可更好的检测区域壁运动异常并区分冠状动脉梗死区域。Cikes等[8]利用复杂超声心动图数据（整个心动周期的左室容积和变形数据，而不是单个数据点）和临床参数的ML算法识别心衰并对心脏再同步化治疗的反应进行评估，证实通过整合临床参数和全心周期成像数据，无监督的ML可以为表型异质性心力衰竭队列提供一个有临床意义的分类，并可能有助于优化特定治疗的反应率。另有研究证实[9-10]，ML算法有助于区分缩窄性心包炎、限制性心肌病以及肥厚性心肌的重塑。Zhang等[11]采用AI软件和手工勾画对左室心肌的纵向应变进行了比较研究。发现AI自动测量的心肌全局纵向应变与手动应变变化最小（绝对值为1.4%～1.6%）。

3.3 心脏瓣膜评估。有学者[12]采用AI软件对二尖瓣几何形状进行测量，测量参数包括二尖瓣环面积、瓣环高度和宽度、瓣叶连合间距、前后叶长度等。发现相对于常规超声心动图，所有评估的成像参数均获得了更好的观察者间一致性，而且所花费的时间明显较少。Prihadi等[13]研究证实，经食管超声心动图AI软件能够精确地对主动脉瓣结构以及冠状动脉开口进行测量和定位，且与多层螺旋CT的测量结果具有良好的相关性。

4 展 望

在海量医学信息和影像数字化日益积累的现代医学时代，AI和ML为疾病诊断和风险预测等问题提供了新的解决方案。通过AI对超声心动图数据进行预测、建模和精确分析，可以帮助超声医师快速、准确地处理大量心脏超声影像学数据，既有利于应对当前医疗信息数量的急剧增长，又有利于提高处理数据信息的能力。未来，针对AI的研究应关注超声图像数据特征定义及其提取方法的标准化，以确保可推广性和可再现性，促进AI向更加个性化的医疗模式转变。此外，AI系统与远程医疗等软件的集成，将使智能心脏超声诊断系统渗透到资源消耗负担最繁重的地区，提高经济效益。

[1]Oikonomou EK,Siddique M,Antoniades intelligence in medical imaging:A radiomic guide to precision phenotyping of cardiovascular disease[J].Cardiovasc Res,2020,Feb 24;cvaa021.

[2]Dey D,Slomka PJ,Leeson P,et Intelligence in Cardio-vascular Imaging:JACC State-of-the-Art Review[J].J Am Coll Cardiol,2019,73(11):1317-1335.

[3]Tran BX,Latkin CA,Vu GT,et al.The Current Research Landscape of the Application of Artificial Intelligence in Managing Cerebrovascular and Heart Diseases:A Bibliometric and Content Analysis[J].Int J Environ Res Public Health,2019,16(15):2699.

[4]Gandhi S,Mosleh W,Shen J,et ,machine learning,and artificial intelligence in echocardiography:A brave new world[J].Echocardiography,2018,35(9):1402-1418.

[5]Alsharqi M,Woodward WJ,Mumith JA,et intelligence and echocardiography[J].Echo Res Pract,2018,5(4):R115-R125.

[6]Xu B,KocyigitD,Grimm R,et of artificial intelligence in multimodality cardiovascular imaging:A state-of-theart review[J].Prog Cardiovasc Dis,2020,19;S0033-0620(20)-8.

[7]Kusunose K,Abe T,Haga A,et al.A Deep Learning Approach for Assessment of Regional Wall Motion Abnormality From Echocardiographic Images[J].JACC Cardiovasc Imaging,2020,13(2 Pt 1):374-381.

[8]Cikes M,Sanchez-Martinez S,Claggett B,et learningbased phenogrouping in heart failure to identify responders to cardiac resynchronization therapy[J].Eur J Heart Fail,2019,21(1):74-85.

[9]Narula S,Shameer K,Salem Omar AM,et -Learning Algorithms to Automate Morphological and Functional Assessments in 2D Echocardiography[J].J Am Coll Cardiol,2016,68(21):2287-2295.

[10]Sengupta PP,Huang YM,Bansal M,et machine-learning algorithm for cardiac imaging:a pilot study for differentiating constrictive pericarditis from restrictive cardiomyopathy[J].Circ Cardiovasc Imaging 2016,9(6):e004330.

[11]Zhang J,Gajjala S,Agrawal P,et al.Fully automated echocardiogram interpretation in clinical practice[J].Circulation,2018,138(16):1623-1635.

[12]Kagiyama N,Toki M,Hara M,et and accuracy of novel automated mitral valve quantification:three-dimensional transesophageal echocardiographic study[J].Echocardiography,2016,33(5):756-63.

[13]Prihadi EA,van Rosendael PJ,Vollema EM,et ,accuracy,and reproducibility of aortic annular and root sizing for transcatheter aortic valve replacement using novel automated threedimensional echocardiographic software:comparison with multi-detector row computed tomography[J].J Am Soc Echocardiogr,2018,31(4):

文章来源：《临床心血管病杂志》 网址: http://www.lcxxgbzz.cn/qikandaodu/2021/0512/751.html