Purpose: To estimate 3D prostate motion in real-time during irradiation from 2D prostate positions acquired from a kV imager on a standard linear accelerator utilizing a Kalman Filter(KF) framework. The advantage of this novel method is threefold: (1) eliminating the need of an initial learning period, therefore reducing patient imaging dose, (2) more robust against measurement noise and (3) more computationally efficient.
Methods: A KF framework was implemented to estimate 3D motion from 2D projection measurements in real-time during prostate cancer treatments. The noise covariance matrix was adaptively estimated from the last 10 measurements. This method did not require an initial learning period as the KF process distribution was initialized using a population covariance matrix. This method was evaluated using a ground-truth motion dataset of 17 prostate cancer patients (536 trajectories) measured with electromagnetic transponders. 3D motion was projected onto a rotating imager (SID=180cm) (pixel size=0.388mm) and rotation speed of 2Â°/s and 6Â°/s to simulate VMAT treatments. Gantry-varying additive random noise (â‰¤5mm) was added to ground-truth measurements to simulate segmentation error and image quality degradation due to the patientâ€™s hip bones. For comparison, motion was also estimated using the clinically implemented Gaussian PDF method initialized with 600 projections and updated with every projection.
Results: Without noise, the 3D Root-mean-square-errors (3D RMSEs) of motion estimated by the KF method was 0.4Â±0.1mm and 0.3Â±0.2mm for 2Â°/s and 6Â°/s gantry rotation, respectively. With noise, 3D RMSEs of KF estimated motion was 1.1Â±0.1mm for both slow and fast gantry rotation scenarios. In comparison, using a Gaussian PDF method, with noise, 3D RMSEs was 2Â±0.1mm for both gantry rotation scenarios.
Conclusion: This work presents a fast and accurate framework for real-time 2D-3D prostate motion tracking on a standard-equipped linear accelerator. This method has sub-mm accuracy and precision and is highly robust against measurement noise.
Funding Support, Disclosures, and Conflict of Interest: D T Nguyen and C C Shieh acknowledge funding from Australian NHMRC and New South Wales Cancer Institute Early Career Fellowships. P Keall acknowledge funding from Australian NHMRC Senior Principal Research Fellowship. R O'Brien acknowledge funding from NSW Cancer Institute Fellowship.