Encrypted login | home

Program Information

Similar-Case-Based Framework of Beam Angle Configurations Using Treatment-Plan-Related Features in Lung Stereotactic Body Radiation Therapy

no image available
M Imai

M Imai1*, H Arimura1 , A Nonaka1 , K Nakamura1 , Y Shioyama2 , H Honda1 , H Hirata1 , (1) Kyushu University, Fukuoka, Japan (2) SAGA Heavy Ion Medical Accelerator in Tosu, Tosu, Japan


SU-E-T-696 (Sunday, July 12, 2015) 3:00 PM - 6:00 PM Room: Exhibit Hall

Purpose: Our goal of this study was to investigate a computational framework for beam angle configurations based on similar cases using treatment-plan-related features in lung stereotactic body radiation therapy.
Methods: A radiation treatment planning database including 81 lung cancer cases and 10 test cases were employed for development and test of the proposed framework. A lung shape feature, which is a shape similarity of two-dimensional (2D) lung region using a Dice's similarity coefficient (DSC), was newly introduced as a treatment-plan-related feature. First, ten cases with slice levels similar to a test case were selected based on a 2D lung shape similarity at an isocenter plane. Second, five similar cases were chosen using additional treatment-plan-related features with respect to lung and spinal cord. Finally, five beam angle configurations of the test case were determined from 5 similar cases by performing an affine linear registration of the lung region between the similar case and the test case. For evaluation of the proposed framework, dose evaluation indices with respect to planning target volume (PTV), lung, and spinal cord were computed to select a usable plan among five similar-case-based plans. Dose evaluation indices of usable plans produced by the proposed and conventional frameworks for 10 test cases were compared with original plans designed by radiation oncologists.
Results: The average homogeneity index in PTVs and V10 in lung for usable plans made by the proposed framework were statistically lower than those for original plans (p<0.05), whereas there were no statistically significant differences between the conventional framework and original plans in any evaluation indices.
Conclusion: The proposed framework with treatment-plan-related features had a potential to suggest usable plans, which can produce more homogeneous dose distributions in PTV and lower volumes with more than 10 Gy in lung, compared with the conventional framework.

Contact Email: