Click here to


Are you sure ?

Yes, do it No, cancel

Predicting Post-SBRT Pulmonary Function Using 4DCT-Derived Ventilation Imaging

N Myziuk1*, G Sakthivel2 , L Foster2 , D Solis3 , I Sala 1, E Castillo4 , T Guerrero1,2,4 , (1) William Beaumont Hospital, Dept. of Radiation Oncology, Royal Oak, MI (2) Oakland University William Beaumont School of Medicine, Rochester, MI, (3) Mary Bird Perkins Cancer Center, Baton Rouge, LA (4) Beaumont Health Research Institute, Royal Oak, MI (6)


(Sunday, 7/14/2019)  

Room: ePoster Forums

Purpose: Radiation Therapy Oncology Group (RTOG) 0236 protocol reported a 3-year survival of 90.6% for stage-I lung cancer in surgically inoperable patients when stereotactic 0% for conventional radiation therapy (RT). The RTOG report also noted grade III or IV respiratory toxicity in 27.3% and 3.6% of the patients, respectively. To date, unlike pulmonary risk assessment (PRA) used to predict residual lung function following thoracic surgery, there is no similar method to predict pulmonary function following radiotherapy. The proposed method aims to provide radiation oncologists with an accurate tool to pre-emptively assess post-SBRT lung function.

Methods: Helical 4DCT images, matching volumetric dose distribution (RTDOSE), and pulmonary function tests (PFT) results were retrospectively examined for a cohort of N=50, stage-I lung cancer patients. Pre-treatment inhale and exhale phase images were segmented to identify lung parenchyma. Ventilation images were calculated from the inhale/exhale pairs, via a density change-based algorithm, utilizing a deformable image registration (DIR) recovered spatial transformation. DIR was used to register the RTDOSE and ventilation images, which were superimposed to calculate their product sum. Least squares regression was used to model post-SBRT PFT (FEV1) scores as a function of dose delivered to functional lung tissue. Correlation between predicted and actual PFT score was examined using the Pearson correlation coefficient.

Results: The Pearson coefficients between predicted PFT and actual PFT scores at 6 weeks, 3 months, 6 months, 9 months and 12 months were 0.92, 0.91, 0.92, 0.95 and 0.79, respectively. The (associated) empirically derived model parameter(s) used to predict post-SBRT PFT scores were λ=0.0124(0.06), -0.0018(0.05), -0.0271(0.10), -0.0096(0.03) and 0.0003(0.03).

Conclusion: These results suggest that model based prediction of post-SBRT lung toxicity is possible with a high degree of accuracy for surgically inoperable patients undergoing SBRT.


Ventilation/perfusion, Lung, Dose


TH- response assessment : CT imaging-based

Contact Email