Program Information
Comparison of Hyperpolarized 3-He MRI and CT Ventilation From a Cohort of Patients On the FLAIR Double-Blind, Randomized Trial
D Hoover1*, T Yamamoto2 , K Sheikh3 , D Capaldi4 , G Parraga5 , D Palma6 , A Louie7 , M Daly8 , S Kabus9 , M Bal10 , B Yaremko11 , (1) London Regional Cancer Program, London, Ontario, (2) UC Davis School of Medicine, Sacramento, CA, (3) Robarts Research Institute, London, ON, (4) Robarts Research Institute, London, Ontario, (5) Robarts Research Institute, London, ON, (6) London Regional Cancer Program, London, ON, (7) London Regional Cancer Program, London, ON, (8) University of California Davis, Sacramento, CA, (9) Philips Research, Hamburg, ,(10) Philips, Best, ,(11) London Regional Cancer Program, London, ON
Presentations
SU-E-605-3 (Sunday, July 30, 2017) 1:00 PM - 1:55 PM Room: 605
Purpose: CT ventilation imaging (V-CT) is an emerging modality that enables lung functional avoidance in radiotherapy. The purpose of this study was to compare hyperpolarized 3-He MRI lung ventilation (V-MRI) with V-CT in lung cancer patients.
Methods: In this secondary analysis of the FLAIR double-blind randomized clinical trial (NCT02002052) of functional avoidance in locally advanced lung cancer, 10 of the 32 patients with pre-treatment 3-He MRI and 4DCT who had the largest ventilation deficits were reviewed. Using a previously published k-means clustering technique, V-MRI maps were segmented into regions of ventilation defect, hypo-ventilation, and normal ventilation. V-CT was computed with deformable image registration and image analysis using Jacobian and HU methods. The average V-CT within each of the three MRI-defined ventilation regions was calculated. The diagnostic performance of V-CT in predicting MRI-defined normal lung ventilation was assessed using individual patient ROC and area under the curve (AUC) calculations. The Youden Index was used to determine a single optimal V-CT threshold for all patients, from which Dice coefficients were calculated for the V-MRI and V-CT normally ventilated lung structures.
Results: The mean ± SD relative volume of the MRI-defined defect/hypo/normally ventilated lung was 24±10%/19±3%/57±10%, respectively. The mean percentile ventilation in the defect/hypo/normal regions for the HU and Jacobian methods were statistically different for all patients individually (P<0.001); averaging over all 10 patients the mean ± SD was respectively 36±7/44±5/56±5 (P<0.001) and 39±8/44±6/ 55±4 (P<0.001). The median [min,max] AUC for HU and Jacobian methods was 0.65 [0.52,0.84] and 0.60 [0.51,0.86], respectively. An optimal percentile corresponding to 3-He MRI normal lung ventilation for HU/Jacobian methods was found to be 37%/37% with corresponding Dice coefficients 0.56±0.07/0.54±0.07, respectively.
Conclusion: The correlation between V-CT and V-MRI suggests physiological significance of CT ventilation imaging. The clinical significance of these findings needs to be confirmed.
Funding Support, Disclosures, and Conflict of Interest: This study is funded by a Grant-in Aid from the Ontario Thoracic Society/Canadian Lung Association (D.H., B.Y.), along with funding from the Ontario Institute for Cancer Research (D.A.P.). S.K. and M.B. are employees of Philips Medical Systems and were not involved in data collection or analysis.
Contact Email: