Application of AAPM TG132 and TG100: Optimisation of image registration workflow for deformable image registration

Abstract

INTRODUCTION: The American Association of Physicists in Medicine: Task Group 132 (AAPM TG132)[1] details recommendations for both deformable image registration (DIR) and rigid image registration (RIR). TG132 was adapted locally with methodologies in AAPM Task Group 100 (TG100)[2] that employed a risk approach to quality management based on the severity (S), occurrence (O), and detectability (D) of failure modes-with uncertainty(C) scoring[3]. Local implementation involved a blend of presentation/training, assessment, scoring, and open feedback. Data was used for generation of adapted AAPM TG132 workflows based on Risk Priority Number (RPN) scores. METHOD: Iterative training of core subject knowledge was performed, including: confidence ratings, multiple choice questions, and open fields for comment. A core team generated process maps and failure modes, followed by departmental engagement to score S, O, and D- and further identification of failure modes. RESULTS: The TG132 process map was adapted to generate a unique image registration workflow (Fig. 1). Failure modes were scored for O, S, and D-scoring was measured with uncertainty ratings. Risk priority numbers (RPN) were used for risk-based quality management. The highest RPN value was for error handling of image registration errors, which received extra resources and management. Fig. 1 Simplified Image Registration Workflow documenting upstream errors, image registration and downstream errors with the highest failure mode in red (error handling RPN 229) CONCLUSION: AAPM TG100 and TG132 were utilised in tandem to train, survey, and engage the department. The core team used this data to optimise process maps to implement RIR and DIR safely and efficiently with risk-based quality management.