Toward bedside computation of myocardial infarction risk using noninvasive analysis of patients living with coronary artery disease.
Toward bedside computation of myocardial infarction risk using noninvasive analysis of patients living with coronary artery disease.
复制标题
使用冠状动脉疾病患者的无创分析来床边计算心肌梗死风险。
DOI：
10.1152/ajpheart.00628.2022
复制
发表时间：
2023
期刊：
影响因子：
--
通讯作者：
Kekenes-Huskey,PeterMichael
中科院分区：
文献类型：
--
作者：
Kekenes-Huskey,PeterMichael
心肌梗死（MI）是冠状动脉疾病中常见的一种。形成狭窄病变的炎性动脉斑块与疾病进展有关。如果狭窄早期发现，生活方式的改变，包括改善饮食、锻炼、体重管理和戒烟，可以减少心脏事件。然而，外科手术，包括冠状动脉搭桥手术，支架植入术和血管成形术，可能是必要的。在心肌梗死发生前识别24条潜在的阻塞性狭窄动脉至关重要。25例病变动脉解剖检查和常规冠状动脉造影[1]为26项用于此目的的有创技术。也可以使用压敏导丝测量动脉病变前后的压力变化;这产生低血流储备分数（FFR），[2]是冠状动脉狭窄的指标[3]。29经皮冠状动脉介入治疗可改善患者结局并减少30起后续事件，但在严重病变和远端动脉血管系统中的治疗效用有限[1]。此外，这些侵入性方法仅提供了32个粗略的度量，例如血管狭窄，几乎没有关于动脉粥样硬化程度的细节，并且容易受到运动伪影的影响[1]。34在“基于深度学习的冠状动脉狭窄阻力预测”中，35 Sun等人开发了一种非侵入性协议，用于根据应用于患者特定冠状动脉数据的计算分类器预测FFR和狭窄风险36。37这需要a）开发从患者来源的冠状动脉数据获知38的3D理想化几何形状的大集合，B）模拟以估计39理想化几何形状的FFR，以及c）基于模拟的FFR数据训练分类器以索引40患者来源的冠状动脉数据（参见图1）。41 42 Sun等人的研究利用已发表的患者血管数据，通过非侵入性技术（计算机断层扫描血管造影术）43预测动脉几何结构中的狭窄。该技术通过X射线创建患者45血管系统的3D重建，可以识别与心血管疾病密切相关的狭窄[2]。计算机断层扫描血管造影提供了47个丰富的数据，包括血管长度和宽度，可与分析48个模型一起使用，以可靠地预测FFR。一个这样的模型，伯努利方程，量化了由于狭窄区域处的对流和收缩50、扩散和从病变血管到51个非狭窄区域的扩张的“压降”而导致的病变内的49个能量损失[3]。52
16 17 18 A myocardial infarction (MI) is a common occurrence in coronary artery 19 disease. Inflammatory arterial plaques that form stenotic lesions are implicated 20 in the disease progression. If stenosis is caught early, lifestyle changes including 21 improved diet, exercise, weight management, and smoking cessation can reduce 22 cardiac events. Surgical interventions, including coronary artery bypass surgery, 23 stents, and angioplasty, may however be necessary. It is critical to identify 24 potentially obstructive, stenotic arteries before an MI occurs. Inspection of 25 pathological artery anatomy and conventional coronary angiography [1] are 26 invasive techniques for this purpose. The pressure change before and after an 27 arterial lesion can also be measured using a pressure-sensitive guidewire; this 28 yields low fractional flow reserve (FFR),[2] an indicator of coronary stenosis [3]. 29 Percutaneous coronary interventions can improve patient outcomes and reduce 30 subsequent events, but have limited therapeutic utility in severe lesions and 31 distal arterial vasculature [1]. Further, these invasive approaches only provide 32 coarse metrics such as vessel narrowing, with little detail about the degree of 33 atherosclerosis, and can be susceptible to motion artifacts [1]. 34 In ‘Deep Learning-based Prediction of Coronary Artery Stenosis Resistance’, 35 Sun et al developed a non-invasive protocol to predict FFR and stenotic risk 36 from computational classifiers applied to patient-specific coronary artery data. 37 This entailed a) developing a large set of 3D idealized geometries informed 38 from patient-derived coronary artery data, b) simulations to estimate FFR for 39 idealized geometries, and c) training classifiers based on simulated FFR data to 40 index patient-derived coronary artery data (see Fig. 1). 41 42The Sun et al study leverages published patient vascular data in order to 43 predict stenosis in arterial geometries from a non-invasive technique: Computerized 44 tomography angiography. This technique creates 3D reconstructions of patient 45 vasculature from X-rays that can identify stenosis that strongly associate with 46 cardiovascular disease [2]. Computerized tomography angiography provides 47 ample data including vessel length and width that can be used with analytic 48 models to reliably predict FFR. One such model, the Bernoulli equation, quantifies 49 energy losses within a lesion due to ‘pressure drops’ from convection and constriction 50 at the stenotic region, diffusion, and expansion from the lesioned vessel to 51 non-stenotic regions [3]. 52