SU-E-J-117: Computer-Assisted Verification of Accumulated Dose Distribution during the Treatment Time Based on...

The accumulated dose distributions during the course of radiation treatment are substantially important for verifying whether treatment dose distributions are produced according to planned dose distributions. The purpose of this study was to develop a computer-assisted verification method of accumulated dose distribution during the irradiation of a tumor based on estimation of four-dimensional (4D) dose distribution using an electronic portal imaging device (EPID).

The 4D 'treatment' computed tomography (CT) images during the irradiation were estimated based on affine transformations including respiratory motions, which were derived by registration between a planning portal dose image and treatment portal dose dynamic image. Planning portal dose images were calculated from planning CT images and an algorithm for calculation of dose spatial distribution. Treatment portal dose images were estimated from EPID dynamic images obtained during a treatment time. The planning portal dose images were registered to the treatment portal dose images to obtain the affine transformation, which could include respiratory motion in a patient body. The CT images at a treatment time were determined by deforming the planning CT images using the affine transformation matrix. 4D dose distributions during a treatment delivery were obtained by applying a dose calculation algorithm to the 4D treatment CT images. Finally, accumulated dose distributions during the course of radiation treatment were verified with planned dose distributions.

We applied the proposed method to EPID dynamic images of 2 lung cancer patients, and evaluated the difference in accumulated dose distribution between the plan and treatment using a gamma evaluation (3mm/3%). The average pass rate for 2 cases was 78.2%.

The proposed method can be used for adaptively modifying the plan based on the dose discrepancy between the plan and treatment. This work was partially supported by Grant-in-Aid for Scientific Research (C) (22611011) and Okawa Foundation for Information and Telecommunications.