SU-E-J-160: Motion Analysis and 3D Plane Fit Through the Tumor Center of Mass Positions in the 4DCT Data Set

Purpose: To model GTV’s center of mass (COM) motion planes based on 4DCT lung data sets and establish possible correlation with the GTV’s size and/or location as an aid in defining adequate GTV margins based on the most prominent motion amplitude in the 3D space. Methods: For the modeling of the COM motion throughout the respiratory phases, eleven lung cancer patients were used and the least square fitting was applied. The least square fitting generally involves either data smoothening or apparent trend identification. Quadratic minimization, singular value decomposition, orthogonal projections (orthogonal regression), are ways to find a least square solution. The linear least squares method was initially used for the planar fitting of the 3D point data. Evaluation of the goodness of fit was performed through the root mean square deviations (root mean square error, RMSE) and normalized RMSE (NRMSE) values. Results: The summed square of residuals of the fit method varied from 0.02 to 0.26 cm and the coefficient of determination from 0.04 to 0.89, while RMSE ranged from 0.05 to 0.13 and NRMSE from 0.11 to 0.29. The most prominent motion was in the transverse and sagittal planes. A correlation was found between the best fit motion plane and the GTV size and/or location. In the lower GTVs, when maximum SI was 1.5, AP and RL motions were 0.87 and 1.0 cm, respectively. This Result confirms the need for margins larger than 0.5 cm for these GTVs. Conclusion: The lower RMSE and NRMSE values showed that for each patient a good planar fit can be achieved for the COM motion path. Knowing the GTV size and location, a good estimate of the motion plane in the 3D space can be made. The required dosimetric margins can be easily calculated knowing only one motion extent and motion plane angle.