Click here to

Session: Multi-Disciplinary General ePoster Viewing [Return to Session]

Deformation-Based Internal Margin Design for Radiotherapy Planning of Lung Cancer

B Spieler1*, B Rich1, L Young2, M Monterroso1, Y Yang3, F Yang1, (1) Department Of Radiation Oncology, University Of Miami, Miami, FL, (2) Department Of Radiation Oncology, University of Washington, Seattle, WA, (3) Department of Radiation Oncology, The First Affiliated Hospital of University of Science and Technology of China, Hefei, China


PO-GePV-M-100 (Sunday, 7/10/2022)   [Eastern Time (GMT-4)]

ePoster Forums

Purpose: In curative-intent radiotherapy (RT) for non-small cell lung cancer (NSCLC), many post-RT failures occur in or adjacent to the planning target volume, suggesting the possibility of suboptimal targeting. Current practices resort to 4D-FBCT in conjunction with deformable image registration (DIR)-based algorithms to propagate a manually contoured reference GTV from a single respiratory phase to all other phases for internal target definition. The goal of this study was to examine and assess the margin of error involved within this process.

Methods: The study cohort included 20 patients with locally advanced NSCLC who underwent chemoradiation. 4D-FBCTs were obtained as respiration-correlated CTs, using 10 breathing phases (0-90%, phase-based binning) and audio-visual feedback to decrease respiratory abnormalities. Expert radiation oncologists delineated the target on all 10 phases of each patient’s 4D scan. For each of the manually contoured volumes, DIR-based target propagation was performed. Maximum symmetric surface distance (MSSD) was used to characterize the surface differences between a given auto-propagated internal GTV (AP-IGTV) and its corresponding ground truth internal GTV (GT-IGTV), i.e., the union of manually delineated volumes on all 10 phases. The dependency of MSSD on the deformation index (DI), defined as the volume ratio between a given AP IGTV and the volume used for its propagation, was further interrogated by Spearman’s correlation.

Results: MSSD between AP-IGTVs and their respective GT-IGTVs ranged from 0.13 to 1.77 cm with a median (interquartile range) of .74 (0.45) cm. DI varied from 1.03 to 1.99 for all AP_IGTV propagations, with a median (interquartile range) of 1.36 (0.27). The Spearman’s rank correlation coefficient between MSSD and DI was statistically significant at 0.48 (p<0.0001).

Conclusion: The observed significant correlation between MSSD and DI suggests that the amount of deformation involved in target propagation may be of value in predicting maximum margin of error for DIR-based IGTV generation.


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email