In this study, the authors have presented the functionality of this deformable HN phantom for testing the accuracy of DIR algorithms and verifying the ART dosimetric accuracy. The authors' experiments demonstrate the feasibility of this phantom serving as an end-to-end ART QA phantom.

Purpose: To design and construct a deformable head and neck (HN) phantom with in‐vivo dosimetry for adaptive radiotherapy (ART) verification purposes. Methods: The two‐dimensional deformable phantom recreates a single CT slice of a HN patient. It consists of two parallel acrylic plates with deformable material sandwiched in between. Gypsum plaster is used for bony anatomy and is rigidly attached to the plates. Soft tissue is mimicked with TX151 solidifying powder mixed with water. Areas of denser tissue are obtained by adding plaster powder into the mixture of TX151 and water. Parotids are represented with properties in between less dense and denser tissue. Tissue is distributed around the bony anatomy and constrained by an elastic rubber band around it. The pharynx is created with a hollow nylon tube. An inflatable silicone balloon filled with diluted barium sulfate acts as tumor. Thin plastic markers are glued on the phantom surfaces adjacent to the clear acrylic plates, providing quantifiable visual information of the deformation. Fixed detector holders are placed inside the cord and near the mandible while holders on rails that move with phantom deformation are placed close to the tumor and inside one parotid. Results: The deformable phantom has been tested for the validation of ART. We have studied the symmetric property and the response of an SRS diode that is used for dose measurements. The four point doses are obtained before and after deformation with the original plan and the re‐plan irradiated. The deformable image registration (DIR) algorithm accuracy is also verified by the deformation ground truth provided by surface markers. Conclusion: We have designed and constructed a deformable and durable HN phantom. This phantom can be used for DIR algorithm accuracy tests as well as ART verification.