Real-time Non-Rigid 3D Reconstruction and Registration for Laparoscopic-guided Minimally Invasive Liver Surgery

项目概要/摘要 肝脏变形导致微创肝脏手术（MILS）期间肿瘤定位困难。的 该提案的目标是通过补偿肝脏来开发一种有效的 MILS 手术导航工具 变形并将术前数据映射到患者的解剖结构。具体来说，我们将开发一种非刚性的 同时定位和建图（SLAM）方法来估计肝脏表面的变形 立体腹腔镜视频。我们将开发机器学习方法来检测地标并执行非 严格登记。该算法将在GPU上实现以实现实时性。初步数据有 论证了可行性。在R00阶段，我们将主要针对临床需求，开发新的 提供术中指导的方法。该项目将极大提高MILS的肿瘤切除精度。 该奖项的候选人周浩印博士是布里格姆大学外科规划实验室 (SPL) 的博士后 妇女医院 (BWH) 和哈佛医学院 (HMS)。周博士拥有丰富的经验和专业知识 计算机视觉、机器学习及其在医学中的应用。 BWH 是基础、 人类疾病的临床和转化研究，并建立了多个研究项目 促进年轻研究者的工作和职业生涯发展。国家图像引导中心 治疗和高级多模态图像引导操作 (AMIGO) 套件将极大地支持这项研究。 周博士的长期研究目标是开发和应用先进的计算机视觉和机器学习 提高对疾病的理解、诊断、治疗和预防的技术，以提供更好的医疗保健。 他的长期职业目标是成为一名工作在医学图像前沿的独立研究者 处理和图像引导治疗。为了实现这些目标，周博士计划接受更多的教育和 以下四个方面的培训： (1) 在医院开展转化研究的关键培训 外科医生和放射科医生的环境，(2) 外科技术开发的知识 指导，(3) 机器学习及其在医学中的应用培训，以及 (4) 写作资助培训 独立申请并寻求资金。周博士将参加哈佛大学精选的正规课程， 哈佛 Catalyst、麻省理工学院 CSAIL 和斯坦福大学课程。他将每周参加 BWH、HMS 和 MIT 举办的研讨会。他 每年还将参加一到两次学术会议来讨论他的工作并会见该领域的专家。 拥有一支强大的导师团队，包括一名主要导师、三名共同导师和两名合作者 为该奖项的K99阶段举办，将为研究和职业生涯提供坚实的支持 周博士基于其在不同研究领域的成熟专业知识而获得发展。威廉·M.教授 Wells III（主要导师）是医学图像处理领域的教授。 Jayender Jagadeesan 教授（联合导师） 是手术机器人和手术导航领域的助理教授。博士。 Ali Tavakkoli 和 Jiping Wang（共同 导师）是经验丰富的外科医生。所有导师和合作者均来自 BWH、HMS。

Project Summary/Abstract Liver deformation leads to difficulties in tumor localization during minimally invasive liver surgery (MILS). The goal of this proposal is to develop an efficient surgical navigation tool for MILS by compensating for liver deformation and mapping preoperative data to the patient’s anatomy. Specifically, we will develop a non-rigid simultaneously localization and mapping (SLAM) approach to estimate the deformation of liver surface from stereo laparoscopy videos. We will develop machine-learning methods to detect landmarks and perform non- rigid registration. The algorithms will be implemented on a GPU to achieve real-time. Preliminary data has demonstrated the feasibility. During the R00 phase, we will mainly address the clinical needs and develop novel ways to provide intraoperative guidance. This project will greatly improve the tumor resection accuracy in MILS. The candidate for this award Dr. Haoyin Zhou is a postdoc at Surgical Planning Laboratory (SPL), Brigham and Women’s Hospital (BWH) and Harvard Medical School (HMS). Dr. Zhou has extensive experience and expertise in computer vision, machine learning and their applications in medicine. BWH is an international leader in basic, clinical and translational research on human diseases, and has established multiple research programs to promote the work and professional career development of young investigators. National Center for Image Guided Therapy, and Advanced Multi-modality Image Guided Operating (AMIGO) suite will greatly support this research. Dr. Zhou’s long-term research goal is to develop and apply advanced computer vision and machine learning technologies to improve understanding, diagnosis, treatment, and prevention of diseases for better health care. His long-term career goal is to become an independent investigator working at the frontier of medical image processing and image-guided therapy. To achieve these goals, Dr. Zhou plans to receive more education and training in the following four areas: (1) Critical training in conducting translational research in the hospital environment with surgeons and radiologists, (2) knowledge in the development of technologies for surgical guidance, (3) training in machine learning and its applications in medicine, and (4) training on writing grant applications independently and seeking funding. Dr. Zhou will participate in formal courses selected from Harvard, Harvard Catalyst, MIT CSAIL and Stanford Courses. He will attend weekly seminars at BWH, HMS and MIT. He will also attend one or two academic conferences per year to discuss his work and meet with experts in the field. A strong mentoring team, including one primary mentor, three co-mentors, and two collaborators, has been organized for the K99 phase of this award, which will provide solid support on both research and career development to Dr. Zhou based on their well-established expertise in diverse research fields. Prof. William M. Wells III (primary mentor) is a professor in medical image processing. Prof. Jayender Jagadeesan (co-mentor) is an assistant professor in surgical robotics and surgical navigation. Drs. Ali Tavakkoli and Jiping Wang (co- mentors) are experienced surgeons. All mentors and collaborators are from BWH, HMS.