基于三维超声的肝脏微创治疗智能导航系统关键技术研究

项目名称： 基于三维超声的肝脏微创治疗智能导航系统关键技术研究

项目编号： No.61501141

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 无线电电子学、电信技术

项目作者： 赵悦

作者单位： 哈尔滨工业大学

项目金额： 20万元

中文摘要： 微创介入式治疗创伤小、恢复快、出血少，是近年来为医生和患者广泛接受的一种新型治疗方式，因此，如何准确有效地使用计算机手段辅助医师完成微创检查及治疗的可视化导航过程成为了一个新的国际研究热点。三维超声图像可以反映组织内部的空间关系，立体感强，接近于实际情况，是未来导航系统的主要引导工具。本项目将使用临床条件下的三维超声数据，拟解决肝脏肿瘤微创治疗中肿瘤定位难、运动模型建立难以及创伤风险规避难的问题，提出实用的解决方法，为计算机辅助的智能导航治疗系统提供更加安全有效的技术手段。本项目的主要内容包括：1) 基于深度学习思想，研究肝脏肿瘤自动识别与定位算法；2) 结合超声图像的光学和纹理信息，研究三维运动估计算法并建立位移预测模型；3) 基于盖然论，探讨医用探针最优路径规划算法。本项目是医学影像领域的一个国际发展前沿，同时在临床应用方面也具有重要的实际应用价值。

中文关键词： 医学图像处理；三维超声成像；深度学习；运动估计；路径规划

英文摘要： Minimally invasive interventional treatment has many advantages as small trauma, quick recovery and little bleeding, and it is a new type of treatment which is widely accepted by doctors and patients in recent years. Therefore, how to accurately and effectively help the physicians with the minimally invasive tools to achieve the target region using computer assistant methods has become a new international research hot spot. Three-dimensional ultrasound image scan reflects the spatial relationship within the soft tissues, has stronger sense of stereo perception and is more close to the actual situations. So it can be the main guide tool of the future navigation systems. This project intends to solve the existing problems of hepatic minimally invasive interventional therapy, such as tumor localization, motion model construction and maximum injury limitation, using the data volumes under the clinical conditions, and proposes the practical solutions. It will provide more safe and effective technical methods of the intelligent computer aided navigation systems for the clinical medicine. The main content of this project includes: 1) research of the automatic recognition and localization algorithm of the hepatic tumor using the deep learning theory; 2) based on the optical and textural information of the ultrasound images, study the 3D motion estimation algorithm, and build the displacement prediction model; 3) based on probabilistic theory, explore the optimal path planning algorithm of the biopsy needles. This project is not only an international frontier development on the field of 3D ultrasound imaging, but also has very important practical value in the clinical applications.

英文关键词： medical image processing;3D ultrasound imaging;deep learning;motion estimation;path planning