562015-03-24 , 法政大学大学院理工学・工学研究科
The purpose of this study is to realize a high precision radiotherapy. We developed an accurate dose calculation method using a cone beam computed tomography (CBCT) and proposed a new image processing method against the movement of a patient or tumor. First of all, we evaluated the quality of the CBCT images with phantom experiments and Monte Carlo simulations, and then studied the effect of scattered photons on an accuracy of the dose calculation. As a result, a CBCT image was strongly affected by scattered photons in an object, and the uniformity of the CBCT image was reduced by approximately one third of that of an image obtained by a multislice computed tomography (MSCT).To overcome this problem, we developed a new dose calculation method with combined image, in which, the CT values were obtained from an MSCT image and the organ contours were obtained from a CBCT image. The results showed that the accuracy of the proposed method improved about 70-80 % in the case with a phantom image, and about 20-30 % in the case with clinical images. It is thus considered that this method was useful to calculate the dose distribution with a high accuracy in any objects.Next, we studied a method against the patient movement during a radiation therapy. In this study, we assumed the stereotactic radiotherapy of a brain tumor. We realized a fast detection of the patient motion by using USB cameras. General-purpose computing on graphics processing units (GPGPU) was used to accelerate the motion detection. As a result, this system enabled us to detect the patient motion within an accuracy tolerance of the movement (2.0 mm) for the head stereotactic radiation therapy.Finally, we proposed an image processing method to reconstruct the tumor position during physiological movement. In this method, projection images acquired with the CBCT were selected only with the projection data in the same target position. As a result, motion artifacts of the target caused by the motion were suppressed, and the quality of reconstructed images was improved at nine different positions.In this thesis, several image processing techniques were applied to improve the quality of the radiation therapy and the validity of the proposed methods was proved both the simulation and clinical studies.