Abstract Book

ESTRO 37

S517

Results The mean error remained less than 1 mm for translations and less than 1º of rotation for both scenarios, with the all markers visible case having slightly smaller errors. The table below summarises the mean±standard deviation of errors for 6DoF motion estimated with 6D-IDC. Figure 1 shows the distributions of errors for both investigated scenarios, the whiskers represent 99.9% of the data. Figure 2 shows a patient’s trace in which 6D-IDC is as accurate with one marker (Fig. 2-IA) as with all markers (Fig.2-IB) and another trace in which 6D-IDC with one marker is less accurate (Fig.2-IIA) than with all markers (Fig.2-IIB). Update with 3 markers Update with 1 marker Left-Right (LR) (mm) -0.03±0.32 -0.07±0.38 Superior-Inferior (SI) (mm) -0.01±0.13 0.12±0.83 Anterior-Posterior (AP) (mm) 0.03±0.52 -0.1±0.60 Rotation around LR(º) 0.07±1.18 0.02±1.29 Rotation around SI(º) 0.07±1.0 0.15±1.13 Rotation around AP(º) 0.06±1.32 0.07±1.46

Conclusion We developed a patient-specific TDD for H&N RT. The TDD significantly decreased the radiation dose to the tongue compared to SMP, which may potentially reduce RT-related toxicity in oral tongue. PO-0948 Use of an interdimensional correlation framework to estimate real-time intrafraction 6DoF motions D.T. Nguyen 1 , J. Bertholet 2 , J.H. Kim 1 , R. O'Brien 1 , J.T. Booth 3 , P.R. Poulsen 2 , P.J. Keall 1 1 The University of Sydney, Radiation Physics Laboratory, Sydney, Australia 2 Aarhus University Hospital, Department of Oncology, Aarhus, Denmark 3 Royal North Shore Hospital, Northern Sydney Cancer Centre, Sydney, Australia Purpose or Objective Increasing evidence suggests that intrafraction tumour motion monitoring needs to include both 3D translations and rotations. Currently, methods to estimate the 3D rotation first require the translational position of at least three target points. We present the first method for directly estimating real-time motion with six-degrees-of- freedom (6DoF) from the target’s projection on a single rotating x-ray imager. This method exploits the inherent correlation between each DoF, and is hence named 6-DoF Inter-Dimensional Correlation (6D-IDC). Once the inferring 2D-6DoF model is built, it allows the estimation of 6DoF motion using less than 3 projected points, which is a distinct advantage of this method. Material and Methods The accuracy of the 6D-IDC method was evaluated in silico with 81 liver tumour motion traces from 19 patients with three implanted markers. The 3D position of each marker was projected onto a gantry-mounted imager with an imaging rate of 11Hz. After an initial 110° gantry rotation (200 images), a correlation model between the superior-inferior translation and the five other DoFs was built using a non- linear least squares method. Two real-time scenarios were investigated: (1) all markers visible and (2) only one marker visible. The latter case represents the clinical scenario where one or two markers are obscured by intervening anatomy, such as projections through vertebral bodies. In the all markers visible simulations, the correlation model was updated after each subsequent frame to estimate 6DoF motion in real-time. In the one marker visible simulations, the 6DoF motion was estimated based on the single marker’s projection using the initial correlation model.

Figure 1.

Figure 2. Conclusion

Direct real-time 6DoF target motion estimate on from projected marker positions on a 2D imager was developed and investigated. The algorithm was found to have sub- mm and sub-degree accuracy when applied to a large clinical liver cancer motion database. The estimation error is slightly higher if only one marker projection is visible compared with when all three markers are visible.