Room: Foyer
Purpose: The report from AAPM Task Group 132 was published in July 2017 with recommendations for validation of image registration algorithms, and included a set of digital phantoms to assist with validation tests. In this work, those tests and phantoms were used to validate and compare the registration results from two clinical software packages, MIM Maestro and the Raystation treatment planning system.
Methods: The software versions used for this work were MIM Maestro 6.7.10 and Raystation 5.0.3, both of which were previously commissioned for clinical use. Digital phantoms were downloaded from the TG-132 website and imported into both systems. Recommended tests from the TG-132 report were performed in each system. For rigid registrations, the translations and rotations given by each system were compared to the expected values. For deformable registrations, registrations were examined visually and, where available, quality metrics were applied on the deformation vector fields, landmark analysis, or contour deformation.
Results: Both systems performed acceptable rigid registrations on all TG-132 digital phantoms, which included registering phantoms simulated with different imaging modalities (CT, T1 MR, T2 MR, PET, and CBCT). Both systems gave results closer to the expected values when performing intramodality registration (CT to CT or CBCT) than for intermodality registration (CT to MR or PET). The deformable registration process relies more on user input than rigid registration and the results are difficult to quantify even for phantoms, but both systems performed similarly on the TG-132 recommended deformable registration tests.
Conclusion: MIM Maestro and Raystation 5.0.3 performed acceptable rigid and deformable registrations by TG-132 criteria. The TG-132 commissioning exercises were helpful to learn the strengths and limitations of each system, and particularly useful as a tool to learn what registration options are available in each system and how they can be used to improve registration results.