Room: ePoster Forums
Purpose: To calculate the gamma index between a reference and evaluated 2-D or 3-D image using MatLab. This code includes an image registration algorithm to find the best graphical fit of the considered images.
Methods: Our algorithm is based on the formalism for gamma calculation proposed by D.A Low et al. with the addition of a grid-based image-registration algorithm. The grid for the evaluated image is sampled at the same resolution of the reference image and the rigid transformation for the best match is calculated. A 4X4 squared field was delivered to the center of a 30X30X30cm3 solid water phantom to assess the performance of the algorithm. Gafchromic EBT3 film were placed in the three central axes. The dose distributions from the treatment planning system (TPS) and the films were compared using our algorithm, DoseLab 6.50 for 2-D gamma and analytically for 3-D gamma.
Results: The proposed algorithm was tested on 2-D data, 3-D simulated data, and 3-D real data. The verification for 2-D film data showed similar gamma index map as DoseLab with similar failing regions. 3-D simulated data included simple geometrical volumes such as a cube and an ellipse where the evaluated image was shifted relative to the reference image. Our 3-D gamma results matched analytically predicted results. For the 3-D data our algorithm was able to identify an artificial shift applied to TPS plan.
Conclusion: This implementation is able to calculate 2-D and 3-D gamma index effectively between two datasets using MatLab.