经皮冠状动脉介入手术术中导航方法研究

发布时间：2018-07-26 07:26

【摘要】：目前,心血管疾病的患病率和死亡率呈现逐年持续上升的趋势,经皮冠状动脉介入手术是目前针对心血管疾病的普适的治疗手段。该方法具有手术创伤小、手术时间短、手术安全性高等优势。基于X造影图像的术中导航是冠状动脉介入手术重要组成环节。目前主要采用主从式血管介入手术机器人完成手术器械的递送,降低X射线对医生的辐射。然而,非实时性和低准确度的视觉反馈信息极大地制约了微创介入手术机器人在临床中的应用。本文主要立足于该临床需求,基于实验室已研制的冠状动脉介入手术机器人平台开展基于X造影图像的术中导航研究,为手术机器人提供实时准确的视觉反馈,从而有效降低手术风险,提高手术的精度。结合国家自然科学基金项目“机器人辅助血管介入手术中器械跟踪与控制方法研究”,本文主要的研究工作包括:冠状动脉介入手术术中导航中X造影图像的处理、基于X造影图像对介入器械检测和血管轮廓检测等的相关算法研究。具体的研究内容和贡献包括:首先,针对传统视觉反馈中导丝检测存在的低准确率、非实时性的问题,本文采用了基于LBP特征的级联Ada Boost分类器对导丝进行检测,该算法实现了导丝的自动检测,有效的提升了视觉反馈的时效性,由于受到X造影图像中类似导丝的骨骼轮廓的干扰,该方法的检测准确率较低,不能满足临床的需求;其次,针对基于机器学习中导丝检测准确率较低的问题,本文采用基于深度学习的方法对导丝进行检测,该算法在导丝检测的实时性和准确性方面取得了满意的结果;最后,血管轮廓检测也是术中导航中另一个重要的环节,针对血管轮廓检测,本文首先采用了一种多尺度的图像增强算法对X造影图像进行增强,基于增强后的图像采用经典的图割法对血管轮廓进行检测。多尺度的增强算法有效的提升了血管轮廓检测的准确性。本文通过对X造影头像进行图像处理,手术器械中导丝和血管轮廓进行检测,获取其位置信息,为医生提供实时准确的视觉反馈信息。
[Abstract]:At present, the prevalence and mortality rate of cardiovascular disease is increasing year by year. Percutaneous coronary intervention is a universal treatment for cardiovascular disease. This method has advantages of less trauma, shorter operation time and higher safety. Intraoperative navigation based on X-ray images is an important component of coronary intervention. At present, the master-slave vascular interventional robot is used to deliver the surgical instruments to reduce the radiation to doctors. However, non-real-time and low-accuracy visual feedback greatly restricted the clinical application of minimally invasive interventional robot. In this paper, based on the clinical requirement, the research of intraoperative navigation based on X-ray images is carried out on the platform of coronary intervention robot developed in laboratory, which provides real-time and accurate visual feedback for the robot. In order to effectively reduce the risk of surgery, improve the accuracy of surgery. Combined with the project of National Natural Science Foundation of China, "instrument tracking and control method in robot-assisted vascular interventional surgery", the main research work in this paper includes: X ray image processing in navigation during coronary artery interventional surgery. Based on X-ray image, the relevant algorithms of interventional instrument detection and vascular contour detection are studied. The specific research contents and contributions include: firstly, aiming at the problems of low accuracy and non-real-time in the traditional visual feedback guided wire detection, this paper uses cascaded Ada Boost classifier based on LBP features to detect the guide wire. The algorithm realizes the automatic detection of guide wire, and effectively improves the timeliness of visual feedback. Due to the interference of bone contour similar to guide wire in X-ray image, the detection accuracy of this method is low, which can not meet the needs of clinical. In order to solve the problem of low accuracy of guide wire detection in machine learning, the method based on depth learning is used to detect guide wire in this paper. The algorithm has achieved satisfactory results in real-time and accuracy of wire guide detection. Vascular contour detection is another important link in intraoperative navigation. In this paper, a multi-scale image enhancement algorithm is used to enhance X-ray image. Based on the enhanced image, the classical image cutting method is used to detect the blood vessel contour. Multi-scale enhancement algorithm effectively improves the accuracy of vascular contour detection. In this paper, the image processing of X ray image and the detection of guiding wire and vascular contour in surgical instruments are carried out to obtain the position information and to provide the doctors with real-time and accurate visual feedback information.
【学位授予单位】：哈尔滨理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R541.4;TP18

【相似文献】