Name
Title | ||
|---|---|---|
Lung tumour motion tracking using a weighted optical flow algorithm with linear accelerator gantry rotation |
Abstract | ||
|---|---|---|
A weighted optical flow algorithm can be used to track tumour motion in real-time in order to optimize radio-therapy of lung cancer. Lung tumour motion due to realistic breathing patterns was simulated using a 3D printed tumour moved by an actuator controlled by LabView. Portal images were acquired on a 6MV linear accelerator using an Electronic Portal Imaging Device (EPID) at 2.5 frames/sec as the simulated tumour was moved. From these sequential images, an optical flow algorithm was used to calculate the tumour velocity and position. Tracking the position of the tumour helps maneuver the radiation beam to follow this movement and to optimize the dose to the tumour while minimizing dose to normal tissues. The optical flow algorithm has been previously used to track motion with a static treatment gantry and was able to detect tumour motion with an average accuracy of 0.5 mm. This work evaluates the ability of this method to track motion when the radiation beam, mounted in an isocentric gantry, is also rotating. Gantry rotation can introduce perceived relative motion on the EPID. This relative motion is dependent on the distance of the tumour to the iso-centre along the X-axis and Z-axis. By incorporating the necessary image transformations to account for the gantry rotation, the tumour motion was tracked with an average positional accuracy of 0.5 mm, similar to the case of a static gantry, and well within other uncertainties associated with this treatment. We conclude that gantry rotation does not affect the ability of the optical flow algorithms to track tumour motion during radiation therapy and can be used to minimize the dose to surrounding normal tissues while maximizing the dose to the tumour. |
Year | DOI | Venue |
|---|---|---|
2017 | 10.1109/ISSPIT.2017.8388680 | 2017 IEEE International Symposium on Signal Processing and Information Technology (ISSPIT) |
Keywords | Field | DocType |
motion tracking,EPID,optical flow,relative motion | Computer vision,Computer science,Relative motion,Algorithm,Breathing patterns,Artificial intelligence,Linear particle accelerator,Optical imaging,Optical flow,Match moving,Actuator,Radiation beam | Conference |
ISSN | ISBN | Citations |
2162-7843 | 978-1-5386-4663-2 | 0 |
PageRank | References | Authors |
0.34 | 0 | 4 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Gabrielle Fontaine | 1 | 0 | 0.34 |
| Bilal A. Ahmed | 2 | 61 | 17.20 |
| Troy Peng T. Teo | 3 | 0 | 0.34 |
| Stephen Pistorius | 4 | 22 | 4.75 |