252 - Comparison of CT Ventilation Imaging and Hyperpolarized <sup>3</sup>He MRI in Patients with Lung Cancer
Presenter(s)
S. Ito1,2, Y. Nakajima1,2, Y. Fujita1, D. P. I. Capaldi3,4, K. Sheikh3,4, G. Parraga4, S. Kabus5, D. A. Palma6, B. Yaremko6, D. Hoover3,6, and T. Yamamoto7; 1Department of Radiological Sciences, Komazawa University, Tokyo, Japan, 2Department of Radiation Oncology, Tokyo Metropolitan Cancer and Infectious Diseases Center, Komagome Hospital, Tokyo, Japan, 3Department of Medical Biophysics, Western University, London, ON, Canada, 4Imaging Research Laboratories, Robarts Research Institute, Western University, London, ON, Canada, 5Department of Medical Image Processing & Analytics, Philips Research, Hamburg, Germany, 6Department of Radiation Oncology, Western University, London, ON, Canada, 7Department of Radiation Oncology, University of California Davis School of Medicine, Sacramento, CA
Purpose/Objective(s): Functional avoidance radiotherapy (RT) may reduce pneumonitis by sparing highly ventilated lung regions (HVL). Computed tomography ventilation imaging (CTVI) has great potential for clinical translation and has been validated in numerous studies. However, previous studies focused on image comparisons, and little is known about dosimetric implications of “good” or “poor” image correlations or agreements in RT. This study assessed the spatial agreement of HVL and correlations and agreements of HVL dose-function metrics of RT plans between CTVI and hyperpolarized 3He magnetic resonance imaging (MRI) as a reference.
Materials/Methods: This study included 27 patients with non-small-cell lung cancer receiving planned 60 Gy in 30 fractions in the Functional Lung Avoidance for Individualized RT (FLAIR) randomized phase 2 trial (NCT02002052), 16 patients randomized to the standard arm and 11 to the avoidance arm. All patients underwent 4-dimensional (4D) CT and 3He MRI. 4D CT images were used to generate ventilation images using the Jacobian- (CTJac) and Hounsfield Unit-based (CTHU) methods. HVL was segmented by k-means clustering for both modalities. The spatial agreement between CTVI- and ³He MRI-based HVLs was assessed using the dice similarity coefficient (DSC). Pearson correlation coefficients were calculated for the HVL dose-function metrics, including V20Gy and mean lung dose (MLD), using the delivered RT plans. Bland–Altman analysis was used to assess the mean differences and 95% limits of agreement. To assess the association between the HVL dose-function metrics and the change from baseline in quality of life (QOL) at 3 months after RT, we analyzed 20 patients with the Functional Assessment of Cancer Therapy–Lung: Lung Cancer Subscale (LCS) data available. Student’s t-test was used for the statistical analysis.
Results: DSCs between CTVI- and 3He MRI-based HVLs were 0.76 ± 0.04 for CTJac and 0.78 ± 0.03 for CTHU. Correlation coefficients of HVL V20Gy and MLD were all >0.85. Bland–Altman analysis indicated small bias (e.g., <1% for V20Gy) but relatively large random differences (e.g., 6-8% for V20Gy) attributable to differences in the HVL volume between CTVI and 3He MRI. Patients with a clinically meaningful decline of =3 points in the LCS score (n = 8) had higher HVL dose-function metrics than those with a decline of <3 points (n = 12), indicating a potential association between HVL dose and QOL decline.
Conclusion: Compared to 3He MRI, CTVI agreed strongly in determining HVL and showed strong correlations and small bias in the dose-function metrics, supporting CTVI as a surrogate of ventilation that can be used to guide functional avoidance RT.
Abstract 252 - Table 1:| V20Gy (%) | MLD (Gy) | |||||
| mean±SD | ||||||
| 3He MRI | CTJac | CTHU | 3He MRI | CTJac | CTHU | |
| LCS decline < 3 (n = 12) | 22.6±7.6 | 21.8±7.2 | 22.5±7.2 | 13.9±4.0 | 13.5±3.7 | 13.9±3.7 |
| LCS decline = 3 (n = 8) | 26.3±5.4 | 27.2±6.4 | 28.4±5.7 | 14.9±2.9 | 15.3±3.3 | 16.1±2.8 |
| P value | 0.28 | 0.12 | 0.08 | 0.54 | 0.31 | 0.19 |