TU‐G‐BRA‐03: Predicting Radiation Therapy Induced Ventilation Changes Using 4DCT Jacobian Calculations

Abstract

Purpose:Longitudinal changes in lung ventilation following radiation therapy can be mapped using four‐dimensional computed tomography(4DCT) and image registration. This study aimed to predict ventilation changes caused by radiation therapy(RT) as a function of pre‐RT ventilation and delivered dose.Methods:4DCT images were acquired before and 3 months after radiation therapy for 13 subjects. Jacobian ventilation maps were calculated from the 4DCT images, warped to a common coordinate system, and a Jacobian ratio map was computed voxel‐by‐voxel as the ratio of post‐RT to pre‐RT Jacobian calculations. A leave‐one‐out method was used to build a response model for each subject: post‐RT to pre‐RT Jacobian ratio data and dose distributions of 12 subjects were applied to the subject's pre‐RT Jacobian map to predict the post‐RT Jacobian. The predicted Jacobian map was compared to the actual post‐RT Jacobian map to evaluate efficacy. Within this cohort, 8 subjects had repeat pre‐RT scans that were compared as a reference for no ventilation change. Maps were compared using gamma pass rate criteria of 2mm distance‐to‐agreement and 6% ventilation difference. Gamma pass rates were compared using paired t‐tests to determine significant differences. Further analysis masked non‐radiation induced changes by excluding voxels below specified dose thresholds.Results:Visual inspection demonstrates the predicted post‐RT ventilation map is similar to the actual map in magnitude and distribution. Quantitatively, the percentage of voxels in agreement when excluding voxels receiving below specified doses are: 74%/20Gy, 73%/10Gy, 73%/5Gy, and 71%/0Gy. By comparison, repeat scans produced 73% of voxels within the 6%/2mm criteria. The agreement of the actual post‐RT maps with the predicted maps was significantly better than agreement with pre‐RT maps (p<0.02).Conclusion:This work validates that significant changes to ventilation post‐RT can be predicted. The differences between the predicted and actual outcome are similar to differences between repeat scans with equivalent ventilation.This work was supported by NIH grant CA166703 and a Pilot Grant from University of Iowa Carver College of Medicine