| 研究実績の概要 |
The purpose of this research is to develop a markerless tracking system for image-guided radiation therapy. The tracking system is capable of tracking the respiration-induced tumor motion automatically in real-time during radiation delivery, and will be able to provide conformable tumor motion information and to allow the treatment device to deliver high-dose conformable radiation to the moving target accurately.
According to the research plan, the research achievements in the FY2018 are summarized as follows. (1) Several thorax phantom-based experiments have been conducted at Tohoku university hospital and sets of kilo-voltage (kV) images data and megavoltage (MV) image data have been acquired for evaluation of tumor tracking system. (3) Based on our previous study, a key-point based tracking method has been published in a prime journal (IF 2017: 2.7). (4) In order to improve the tracking accuracy, a hidden Markov model-based method is proposed to extract the tumor from the radiographic image sequences. The preliminarily experimental results demonstrated the effectiveness of the proposed method. (5) Several deep learning-based methods have been investigated for tumor segmentation and tracking in kV and MV images.
|
| 次年度使用額が生じた理由 |
According to the research plan, a graphics processing unit (GPU)-based tracking software framework should be developed and evaluated in FY2018. For this task, a high-performance computer or workstation with GPGUP were investigated. In September 2018, Nvidia launched a new RTX graphics cards, GeForce RTX 2080 Ti, which was the considered for our development. However, after the RTX 2080 Ti launching, we found that many owners of Nvidia's RTX reported a dying issue of this product. Nvidia also confirmed that the problem is due to "limited test escapes" after the problems were reported.
Therefore, we plan to wait the company solve this hardware issue, and re-investigate a new GPGPU in the FY2019.
|
