Fast and accurate transit portal dosimetry was investigated by developing a density-scaled layer model of electronic portal imaging device(EPID) and applying it to a clinical environment.Methods:
The model was developed for fast Monte Carlo dose calculation. The model was validated through comparison with measurements of dose on EPID using first open beams of varying field sizes under a 20-cm-thick flat phantom. After this basic validation, the model was further tested by applying it to transit dosimetry and dosereconstruction that employed our predetermined dose-response-based algorithm developed earlier. The application employed clinical intensity-modulated beams irradiated on a Rando phantom. The clinical beams were obtained through planning on pelvic regions of the Rando phantom simulating prostate and large pelvis intensity modulated radiation therapy. To enhance agreement between calculations and measurements of dose near penumbral regions, convolution conversion of acquired EPIDimages was alternatively used. In addition, thickness-dependent image-to-dosecalibration factors were generated through measurements of image and calculations of dose in EPID through flat phantoms of various thicknesses. The factors were used to convert acquired images in EPID into dose.Results:
For open beam measurements, the model showed agreement with measurements in dose difference better than 2% across open fields. For tests with a Rando phantom, the transit dosimetry measurements were compared with forwardly calculated doses in EPID showing gamma pass rates between 90.8% and 98.8% given 4.5 mm distance-to-agreement (DTA) and 3% dose difference (DD) for all individual beams tried in this study. The reconstructeddose in the phantom was compared with forwardly calculated doses showing pass rates between 93.3% and 100% in isocentric perpendicular planes to the beam direction given 3 mm DTA and 3% DD for all beams. On isocentric axial planes, the pass rates varied between 95.8% and 99.9% for all individual beams and they were 98.2% and 99.9% for the composite beams of the small and large pelvis cases, respectively. Three-dimensional gamma pass rates were 99.0% and 96.4% for the small and large pelvis cases, respectively.Conclusions:
The layer model of EPID built for Monte Carlo calculations offered fast (less than 1 min) and accurate calculation for transit dosimety and dosereconstruction.
J. O. Kim expresses gratitude to Dr. Fippel for providing him with the XVMC code. We acknowledge the help of Dr. Jimm Grimm who designed the median filter used in this study and Dr. Byungyong Yi who provided the inhouse software for the 3D gamma analysis. The dose evaluation in this study was performed using MapCHECK (Sun Nuclear, Inc., Melbourne, FL). This study was in part supported by Varian Medical Systems, Inc. The method of dose reconstruction used in this study is US-patented.
II. MATERIALS AND METHODS
II.A. EPID modeling with XVMC
II.A.1. Homogeneous layer model
II.B. Validation of EPID model
II.B.1. Open fields on flat phantom
II.C.1. IMRT fields on Rando phantom
II.C.2. Dosereconstruction: Advanced clinical use
III. RESULTS AND DISCUSSIONS
III.A. EPID modeling with XVMC
III.B. Validation of EPID model
III.B.1. Open fields on flat phantom
III.C.1. IMRT fields on Rando phantom
- Image guided radiation therapy
- Medical imaging
- Medical image reconstruction
Data & Media loading...
Article metrics loading...
Full text loading...