时间：2013年5月23日（星期四）10:30am-12:00am
地 点：东11楼二楼会议室
演讲人: Dongshan Fu, Ph.D. 付东山教授
邀请人：丁明跃教授
演讲题目: Patient Alignment and Dynamic Target Tracking in the CyberKnife® Robotic Radiosurgery System
演讲人简介：付东山，浙江金华人，国家“计划”专家，江苏省首批产业教授；曾任美国Accuray公司影像研发部资深科学家、主管，现任江苏瑞尔医疗科技有限公司总经理。拥有美国发明专利二十余项；发表国际期刊论文十余篇，任Medical Physics和AAPM审稿人。
演讲内容简介：
The CyberKnife Robotic Radiosurgery System is able to detect, track and correct for patient movement during patient setup and treatment delivery. An imaging system, which consists of two diagnostic X-ray generators mounted to the ceiling and two amorphous silicon detectors sitting on the ground, generates two orthogonal high-resolution X-ray images, and is used to detect the patient and internal target (tumor) location. A treatment couch then aligns the patient automatically to establish the precise initial spatial location of the patient. In treatment delivery, the imaging system repeats periodically the X-ray image acquisition and target location detection. As a result, a robot can precisely point the radiation beam to the target by moving a compact linear accelerator that is attached on the robot. For the tumor that moves with respiration, in addition to periodic tumor location detection, an optical tracking system performs a real time measurement of external skin makers. A mathematic correlation model between the internal target location and the external marker movement is initially built before beam delivery and regularly updated during treatment to predict the tumor respiratory motion in real time. Therefore, the robot can continuously align the radiation beam to the moving target.
Internal target location detection is achieved by 2D-3D registration of intra-treatment X-ray images to the pre-operative CT. Digitally reconstructed radiographs (DRRs) are generated from the CT prior to treatment. Similarity measures compare the X-ray images and the DRRs to determine the patient position and orientation change between the time of CT scanning and the time of treatment. Tracking is accomplished by periodically detecting the tumor position in the entire course of treatment delivery. Different tracking methods are used for different clinical applications, including skull tracking for intracranial radiosurgery, fiducial-less spine tracking for spinal treatment, direct lung tumor tracking for non-invasive lung tumor treatment, and fiducial tracking for most of soft tissue tumor treatment. Experimental measurements on anthropomorphic phantoms showed that the accuracy (RMS* error) is less than 0.3 mm for each translation and less than 0.3o for each rotation. Retrospective analysis of clinical image data showed that the tracking error is less than 1 mm in fiducial, skull and spine tracking. Experiments also indicated that the targeting error is less then 1.5 mm for moving target, using the combination of optical marker tracking with either fiducial or fiducial-less lung tumor tracking.
* RMS – root mean square.
