부산대학교 도서관

Background: This study aims to investigate the effects of different immobilization devices on dose distribution in head and neck patients treated with radiotherapy by using the deformable image registration (DIR) technology. Methods: DIR with different algorithms and parameter settings were investigated after different immobilization devices are used. With computed tomography image1 (plan CT) using immobilization device1 and CT2 (re-plan CT) using immobilization device2, the region including immobilization device1 and body regions were defined as Region1 (R1), and the region including body regions only was defined as Region2 (R2). The R1 and R2 regions were controlled by the integrated hybrid (Hybrid) algorithm and biomechanical (Biomechanical) algorithm. Contour propagation and dose deformation were used to analyze whether immobilization would cause errors when we evaluate and estimate the dose. Results: The mean dice similarity coefficients (DSCs) of Hybrid and Biomechanical for the R1 setting were 0.63 and 0.32, respectively. The Hybrid and Biomechanical mean DSCs for the R2 setting were 0.94 and 0.83, respectively. The contour consistency of the R2 setting was significantly better than that of R1. In dose deformation, the Hybrid and Biomechanical set by R1 have an average difference of 13.15% and 17.82% in dose compared to CT1 organs at risk (OARs). The average dose difference between Hybrid and Biomechanical under the R2 setting was 3.32% and 5.13%, respectively. The dose differences of OARs for R2 were all lower than for the R1 setting. The results show that dissimilar immobilization can lead to serious errors in DIR. Conclusion: The study suggests that when we use DIR for clinical treatment planning, immobilization device should be excluded from the registration range of the patient's anatomical region to avoid causing dose errors in evaluation.