Improving patient specific quality assurance for image registration: clinical use case of target contouring for PET deformable image registration

Deformable image registration (DIR) has proven to be an invaluable tool to maximize the clinical benefits of multimodality imaging in radiation oncology. In contrast to rigid image registration (RIR), which is employed at all stages of diagnosis and treatment, the uptake of DIR has been constrained by concerns over the potential for unsafe use. The AAPM Task Group 132 (TG132) published a report on the use of image registration, including many recommendations on clinical integration of registration in treatment planning and delivery. There is a remaining uncertainty on incorporating registration uncertainties into treatment margins (Sect. 6.A, TG 132), a challenge in clinical practice. The aim of this work was to report our experience in implementing a practical, patient specific quality assurance process based on the AAPM Task Group 132 report recommendations. This work includes refining our process of target contouring using PET with deformable image registration based on our experience of addressing vulnerabilities identified during implementation. A multidisciplinary team created a flowchart for patient specific quality assurance for image registration (RIR or DIR) based on use cases defined in the AAPM TG132 Report on the use of image registration in radiotherapy. Vulnerabilities identified from this implementation were assessed relative to AAPM TG132 recommendations. These findings were used to adapt our patient specific quality assurance to mitigate vulnerabilities. The main vulnerabilities were identified in the last steps of image registration. There was potential for inappropriate use of the registration for clinical use, such as target contouring where the image registration accuracy level was poor. Vulnerabilities were addressed by an adaptation in our quality assurance process. A new physics image registration QA task was introduced that independently checks registration accuracy and appropriateness of target contouring, addressing the vulnerability in the last steps of the AAPM TG132 flowchart. A multi-disciplinary team implemented the image registration process outlined by AAPM TG132. An improved patient specific quality assurance process was developed by introducing an independent physics image registration review that considers the acceptable registration uncertainty for the specific clinical use case in question.