3673 - A Rapid Screening Strategy for DIBH Suitability without Additional DIBH-CT Scans
Presenter(s)
T. Lv1, Y. Zhang2, J. Dong1, Y. Wang3, X. Chen3, K. Men3, and W. Zhao4,5; 1Department of Radiotherapy, Beijing Hospital, Beijing, China, 2Shanghai General Hospital, Shanghai, China, 3Department of Radiation Oncology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China, 4Hangzhou International Innovation Institute, Beihang University, Zhejiang, China, 5School of Physics, Beihang University, Beijing, China
Purpose/Objective(s): Deep inspiration breath-holding (DIBH) technique for left breast cancer reduces cardiac radiation exposure but requires additional DIBH-CT scans to confirm patient suitability, increasing radiation dose and treatment costs. We propose a novel screening strategy leveraging synthetic DIBH-CT (sDIBH-CT) and dose prediction to eliminate the need for repeated DIBH-CT scans.
Materials/Methods: A total of 178 breast cancer patients underwent scans with both free-breathing (FB-CT) and DIBH-CT. A synthetic network was trained to generate sDIBH-CT from FB-CT, with data split into training (90 patients), validation (23 patients), and testing (30 patients). For each patient in the test set, the synthetic model was re-trained using FB-CT and DIBH orthogonal projections to obtain the refined sDIBH-CT. Image deformation registration was used to register from FB-CT to the refined sDIBH-CT to obtain final predicted DIBH-CT (pDIBH-CT) and associated contours (from FB-CT). A dose prediction network was trained (50 patients) and tested (15 patients) to estimate the mean heart dose (MHD) of pDIBH-CT and compare it with the FB-CT. Registration accuracy was quantified by Dice similarity coefficient (DSC). Image quality was evaluated via mean absolute error (MAE) between pDIBH-CT and DIBH-CT. A screening threshold for MHD reduction (FB vs. DIBH) was established to identify DIBH-suitable patients.
Results: The MAE between DIBH-CT, FB-CT, and pDIBH-CT decreased from 178.49±41.95 HU to 55.19±14.1 HU. The DSC between DIBH-CT, FB-CT, and pDIBH-CT for CTV, PTV, heart and lungs improved from 0.5±0.19 to 0.87±0.05, 0.59±0.17 to 0.89±0.04, 0.79±0.05 to 0.92±0.01, and 0.81±0.04 to 0.96±0.01, respectively. The accuracy of the radiotherapy strategy screening reached 86.7%.
Conclusion: We propose a rapid screening strategy for assessing DIBH suitability without the need for additional DIBH-CT scans. This approach provides predicted DIBH-CT and contours to assist with subsequent treatment planning, while utilizing a dose prediction network to rapidly provide MHD indicators for screening. This strategy has the potential to greatly optimize clinical workflow and reduce the burden on both patients and healthcare providers.