Department of Vascular and Endovascular Surgery, Saint-Joseph Hospital, 26 boulevard de Louvain, 13008, Marseilles, France.
Department of Vascular and Endovascular Surgery, Saint-Joseph Paris Hospital Group, Paris, France.
Semin Vasc Surg. 2024 Sep;37(3):306-313. doi: 10.1053/j.semvascsurg.2024.07.002. Epub 2024 Jul 17.
Current planning of aortic and peripheral endovascular procedures is based largely on manual measurements performed from the 3-dimensional reconstruction of preoperative computed tomography scans. Assessment of device behavior inside patient anatomy is often difficult, and available tools, such as 3-dimensional-printed models, have several limitations. Digital twin (DT) technology has been used successfully in automotive and aerospace industries and applied recently to endovascular aortic aneurysm repair. Artificial intelligence allows the treatment of large amounts of data, and its use in medicine is increasing rapidly. The aim of this review was to present the current status of DTs combined with artificial intelligence for planning endovascular procedures. Patient-specific DTs of the aorta are generated from preoperative computed tomography and integrate aorta mechanical properties using finite element analysis. The same methodology is used to generate 3-dimensional models of aortic stent-grafts and simulate their deployment. Post processing of DT models is then performed to generate multiple parameters related to stent-graft oversizing and apposition. Machine learning algorithms allow parameters to be computed into a synthetic index to predict Type 1A endoleak risk. Other planning and sizing applications include custom-made fenestrated and branched stent-grafts for complex aneurysms. DT technology is also being investigated for planning peripheral endovascular procedures, such as carotid artery stenting. DT provides detailed information on endovascular device behavior. Analysis of DT-derived parameters with machine learning algorithms may improve accuracy in predicting complications, such as Type 1A endoleaks.
目前，主动脉和外周血管腔内手术的规划主要基于从术前计算机断层扫描的三维重建中进行的手动测量。评估器械在患者解剖结构内的行为通常很困难，并且可用的工具（如 3D 打印模型）存在许多限制。数字孪生（DT）技术已成功应用于汽车和航空航天工业，并于最近应用于血管内主动脉瘤修复。人工智能允许处理大量数据，其在医学中的应用正在迅速增加。本综述旨在介绍结合人工智能进行血管内手术规划的 DT 的最新现状。从术前计算机断层扫描生成特定于患者的主动脉 DT，并使用有限元分析整合主动脉机械性能。同样的方法用于生成主动脉支架移植物的三维模型并模拟其部署。然后对 DT 模型进行后处理，以生成与支架移植物过度扩张和贴合相关的多个参数。机器学习算法允许将参数计算为综合指数，以预测 1A 型内漏风险。其他规划和尺寸调整应用包括用于复杂动脉瘤的定制开窗和分支支架移植物。DT 技术也正在研究用于规划外周血管腔内手术，如颈动脉支架置入术。DT 提供了有关血管内器械行为的详细信息。使用机器学习算法分析 DT 衍生参数可能会提高预测并发症（如 1A 型内漏）的准确性。
