Automatic segmentation of in-vivo intra-coronary optical coherence tomography images to assess stent strut apposition and coverage
Automatic segmentation of in-vivo intra-coronary optical coherence tomography images to assess stent strut apposition and coverage
复制标题
DOI：
10.1007/s10554-011-9824-3
复制
发表时间：
2012-02-01
影响因子：
2.1
通讯作者：
D'hooge, J.
中科院分区：
文献类型：
--
作者：
Ughi, G. J.;Adriaenssens, T.;D'hooge, J.
冠状动脉内支架置入术是目前治疗冠状动脉粥样硬化性疾病的标准方法。这项技术的广泛应用促进了该领域的密集研究活动，这些设备的设计不断改进，旨在减少再狭窄(支架节段再次狭窄)和血栓形成(血栓形成导致的突然闭塞)的问题。最近，一种新的、基于光的冠状动脉内成像方式--光学相干断层扫描(OCT)被开发并引入临床实践。由于其非常高的轴向分辨率(10-15微米)，它允许在体内评估支架支架的对位和新生内膜覆盖(治疗节段愈合的标志)。因此，它为正确部署支架、不同类型支架在体内的行为以及新型支架(例如药物洗脱支架)对血管壁愈合的影响提供了有价值的信息。然而，当前OCT方法的主要缺点是对这些图像的分析需要大量的--目前是人工--的后处理。在这篇手稿中，提出了一种算法，允许全自动分析支架支架在冠状动脉中的对位和覆盖。通过分析A型线的强度分布，自动检测和分割血管管腔和支架支柱。根据这些数据，可以自动测量对位和覆盖范围。该算法通过两个有经验的操作员的手动评估作为参考进行了验证。自动测量与人工测量之间有较高的皮尔逊相关系数(R=0.96~0.97)，而Bland-Altman分析显示无显著偏差，具有良好的一致性限度。因此，该算法提供了一个健壮和快速的工具，在体内OCT回撤中自动估计支架支柱的对位和覆盖。这对于将这项技术集成到临床常规中以及对更大临床试验的数据集进行分析将是重要的。
The implantation of intracoronary stents is currently the standard approach for the treatment of coronary atherosclerotic disease. The widespread adoption of this technology has boosted an intensive research activity in this domain, with continuous improvements in the design of these devices, aiming at reducing problems of restenosis (re-narrowing of the stented segment) and thrombosis (sudden occlusion due to thrombus formation). Recently, a new, light-based intracoronary imaging modality, optical coherence tomography (OCT), was developed and introduced into clinical practice. Due to its very high axial resolution (10-15 mu m), it allows for in vivo evaluation of both stent strut apposition and neointima coverage (a marker of healing of the treated segment). As such, it provides valuable information on proper stent deployment, on the behaviour of different stent types in-vivo and on the effect of new types of stents (e.g. drug-eluting stents) on vessel wall healing. However, the major drawback of the current OCT methodology is that analysis of these images requires a tremendous amount of-currently manual-post-processing. In this manuscript, an algorithm is presented that allows for fully automated analysis of stent strut apposition and coverage in coronary arteries. The vessel lumen and stent struts are automatically detected and segmented through analysis of the intensity profiles of the A-lines. From these data, apposition and coverage can then be measured automatically. The algorithm was validated using manual assessments by two experienced operators as a reference. High Pearson's correlation coefficients were found (R = 0.96-0.97) between the automated and manual measurements while Bland-Altman analysis showed no significant bias with good limits of agreement. As such, it was shown that the presented algorithm provides a robust and fast tool to automatically estimate apposition and coverage of stent struts in in-vivo OCT pullbacks. This will be important for the integration of this technology in clinical routine and for the analysis of datasets of larger clinical trials.