Encrypted login | home

Program Information

Impact of Optimal Weighting for Functional Lung Avoidance Using CT Ventilation

no image available
T Kanai

T Kanai1,2*, N Kadoya2 , Y Nakajima2,3 , Y Miyasaka2 , Y Ieko2 , T Kajikawa2 , K Ito2 , T Yamamoto2 , S Dobashi2 , K Takeda2 , K Nemoto1 , K Jingu2 , (1) Yamagata University Faculty of Medicine, (2) Tohoku University Graduate School of Medicine, (3) Tokyo Metropolitan Cancer and Infectious Diseases Center Komagome Hospital,


SU-I-GPD-J-58 (Sunday, July 30, 2017) 3:00 PM - 6:00 PM Room: Exhibit Hall

Purpose: CT ventilation is increasingly used as a functional avoidance in thoracic radiation therapy. Treatment plans can be optimized by reducing functionally weighted lung dose. However, effective weighting method for reducing radiation pneumonitis is still unclear. The aim of this study is to investigate an optimal weighting method to realize effective planning optimization.

Methods: 54 patients treated with conventional or hypo-fractionated radiotherapy were analyzed. CT ventilation images were obtained from 4D-CT images using intensity-based deformable registration and Hounsfield unit-based calculation. Ventilation values of every pixel were converted into percentile. 19 different threshold values (5-95, every 5 percentile) were determined. Functionally-weighted mean lung dose (fMLD) and V20 (fV20) were calculated by weighting the dose voxel only with higher ventilation value than these thresholds. The predictive accuracy of CTCAE grade≧2 radiation pneumonitis was estimated by area under the curve (AUC) of receiver operating characteristic analysis. Statistical significance was tested by Student’s t-test.

Results: A combination of fV20 and 15 percentile threshold showed the best predictive accuracy (AUC: 0.631, p=0.03). 15 percentile threshold also showed good result with fMLD (AUC: 0.619, p=0.08). These AUC values were higher than those with conventional DVH parameters (0.607 with V20 and 0.595 with MLD). On the contrary, 95 percentile showed the lowest AUC values (fV20: 0.518, p=0.38, fMLD: 0.526, p=0.32).

Conclusion: Around 15 percentile was supposed to be optimal threshold to reduce radiation pneumonitis. Functional weighting improved the predictive accuracy of MLD and V20. However, extremely high threshold, such as 90 percentile, may be ineffective for reducing radiation pneumonitis.

Contact Email: