Program Information
Intra-Treatment Tumor Position Estimation: Feasibility and Achievable Accuracy
D Zhou*, J Liang , S Chen , D Yan , William Beaumont Hospital, Royal Oak, MI
Presentations
TU-H-605-7 (Tuesday, August 1, 2017) 4:30 PM - 6:00 PM Room: 605
Purpose: To investigate the feasibility and accuracy of using intra-treatment diaphragm motion to estimate lung tumor motion.
Methods: Eighty-one paired 4D-CBCT images obtained at daily pre- and post-treatment from 19 lung cancer SBRT patients were used in the study. For each treatment fraction, the geometric relationship between diaphragm and tumor in S-I direction manifested on the pre-treatment images were obtained. Rigid registration was performed between the end-exhalation phase and each of the other 9 phases; diaphragm was registered based on its apex and tumor was registered based on a 3D-shaped ROI. Two estimation models for intra-treatment tumor position tracking were constructed using the linear and quadratic regression between diaphragm and tumor positions manifested on the pre-treatment images. Applying the estimation model on the diaphragm motion obtained from the post-treatment 4D-CBCT image, the tumor motion positions were estimated, and compared with the actual tumor positions measured from post-treatment 4D-CBCT. Geometric relationships between diaphragm and tumor were used to test the estimation accuracy. Receiver-operating-characteristic (ROC) analysis was used to perform the statistic test.
Results: Both the linear and quadratic models provide similar estimation accuracy. 95% of daily intra-treatment tumor motion positions can be estimated within 3mm; 83% within 2mm and 54% within 1mm. Linear regression residual determined from pre-treatment image shows the largest prediction power for the intra-treatment tumor position estimation discrepancy < 3mm with the AUC = 0.92 (p < .05).
Conclusion: Utilizing the relationship between diaphragm and tumor positions manifested on the pre-treatment 4D CBCT image, intra-treatment tumor motion position can be estimated using the intra-treatment diaphragm motion positions with the accuracy < 3mm for 95% of confidence. The estimation accuracy can be predicted using the regression residual determined from the pre-treatment 4D CBCT image.
Funding Support, Disclosures, and Conflict of Interest: Supported by Elekta Research Grant
Contact Email: