Author: Muhammad Ramish Ashraf, Clinton Gibson, Gregory Szalkowski, Lei Wang, Siqi Wang, Lei Xing 👨🔬
Affiliation: Department of Radiation Oncology, Stanford University, Department of Radiation Oncology, Stanford University School of Medicine, Stanford University 🌍
Purpose: To develop a neural network-based super-resolution framework for enhancing the resolution of sparse dosimetry measurements in patient-specific radiotherapy QA. Sparse detector arrays, such as MapCHECK (7.07 mm spacing) and ArcCHECK (10 mm spacing), create uncertainty in assessing neighboring regions and resolving fine-scale dose variations, particularly in high-gradient areas critical for treatment accuracy. Current gamma analysis is limited to individual detector points due to the limitations of interpolation methods. Our framework reconstructs high-resolution dose distributions from sparse measurements, enabling comprehensive full-map gamma analysis with high accuracy for the first time.
Methods: The framework processes a single treatment plan and an ArcCheck scan to achieve super-resolution results. For validation, additional measurements were acquired by shifting the ArcCheck device by 5 mm. These shifted measurements were matched against the reconstructed dose distributions, verifying the accuracy of the super-resolution results. The implicit neural representation (INR) model, pre-trained on treatment plans, refines its predictions using sparse ArcCheck data. The inclusion of shifted measurements provided an independent assessment of the framework’s robustness, demonstrating its ability to adapt to real-world deviations while maintaining spatial fidelity.
Results: The INR framework reconstructs complete dose distributions, enabling full-map gamma analysis, unlike traditional point-based methods. The reconstructed distributions achieved passing rates exceeding 90% under stringent criteria (1 mm/1%). Validation using 5 mm shifted measurements confirmed the reliability of the framework in accurately resolving dose distributions.
Conclusion: This study introduces a paradigm shift in radiotherapy QA by enabling reliable full-map gamma analysis. The framework’s ability to reconstruct detailed dose distributions from sparse measurements transforms QA workflows, allowing comprehensive analysis without additional measurement time. This advancement offers unprecedented insights into treatment accuracy, enhancing QA efficiency while maintaining high clinical standards.