Purpose: To construct a phantom designed to test deformable image registration algorithms, by developing a set of deformable thermoplastic materials that mimic tissue for both kilovoltage and megavoltage computed tomography. Methods: A material formulation was developed to represent muscle, water, and adipose using two polyurethanes capable of deforming in boiling water to produce anatomical deformations. The CT Hounsfield Units (HU) were tuned by doping the polyurethanes with a primary additive to increase the HU and a secondary additive to decrease it. To mimic actual bone, a plastic mixture whose HU was controlled by a tertiary additive was formulated to remain rigid in boiling water. A phantom prototype representing a central sagittal head and neck slice was constructed using the optimized materials. To ensure the phantom was anthropomorphic, its geometry was designed by contouring the major anatomical sites of a patient. The phantom was constructed so it could be opened along the sagittal midsection where a grid of 300 nonradioopaque markers were placed with 5mm spacing. The grid served as a nonradioopaque means of characterizing the 3D magnitude of a deformation with an optical camera. Results: The phantom materials were successfully tuned to be both kV and MV compatible and made to represent different tissues. The phantom constructed from these materials exhibited a similar HU contrast to an actual patient and was successfully deformed around the neck section, producing a clinically realistic deformation of several mm in magnitude. Conclusion: Using the formulated set of materials, an anthropomorphic phantom depicting different tissue types was built and deformed in 3D in a clinically realistic fashion. By measuring the positional changes of the optical markers, the constructed phantom can be used to physically verify deformable image registration algorithms. The work was supported in part by a UC Proof of Concept Program Commercialization Gap Grant.
ASJC Scopus subject areas
- Radiology Nuclear Medicine and imaging