TL;DR
This paper introduces a novel method for quality assurance (QA) of the multi-leaf collimator (MLC) in MR-guided linear accelerators (MR-linacs). By digitally stitching together two double-exposed films, researchers can now perform a comprehensive "picket fence" test that covers the entire MLC, overcoming limitations of existing technologies. This technique offers clinically acceptable accuracy and sensitivity for detecting MLC positioning errors, crucial for safe and effective MR-linac radiotherapy.
What Problem Does This Paper Address?
Modern radiotherapy, particularly MR-guided radiotherapy (MRgRT), relies on precise beam shaping by the MLC to deliver radiation accurately to tumors while sparing healthy tissues. The "picket fence" (PF) test is a standard QA procedure to verify MLC positioning accuracy. However, on the Elekta Unity MR-linac, this test faces significant challenges. The electronic portal imaging device (EPID) on the Unity has a limited field of view, only covering a small central portion of the MLC. Furthermore, the maximum field size of the Unity exceeds the dimensions of standard radiochromic films (like EBT3) typically used for this test. This means a single film cannot capture the entire MLC, and the EPID cannot provide a complete picture. This gap in QA coverage could compromise treatment safety and efficacy.
Methods
The authors propose a technique that utilizes two GAFCHROMIC EBT3 films and a solid water phantom to overcome the field-of-view limitations. Each film is positioned to cover half of the MLC's leaf bank. Crucially, each film undergoes a "double exposure" with two distinct radiation fields:
- Picket Fence (PF) Pattern: A series of narrow, parallel beams designed to test MLC leaf alignment.
- Open Field: A large, uniform rectangular field used for registration and alignment purposes.
The key innovation lies in the digital stitching of these two films. The open field serves as a reference for correcting film rotation and aligning the films horizontally. Vertical alignment is achieved by matching specific picket positions identified on both films and referencing EPID measurements. After stitching, individual leaf positions are extracted from the merged film using a gradient-based method to detect leaf edges. The technique's sensitivity to positioning errors is then evaluated by introducing known errors to the MLC.
Datasets
Not discussed in the paper. The study relies on measurements performed using EBT3 films and the Elekta Unity MR-linac's EPID.
Key Insights and Ideas
- Bridging the Gap: The core insight is the clever use of two overlapping films and digital stitching to achieve a comprehensive MLC QA that was previously impossible with either single films or the limited EPID.
- Open Field as a Universal Reference: The open field isn't just for exposure; it's a critical tool for geometric correction (rotation, horizontal alignment) and inter-film registration, simplifying the complex stitching process.
- Gradient Method for Leaf Detection: Using the gradient of the intensity profile to identify leaf positions is a robust approach that is less sensitive to intensity variations compared to methods relying on full-width at half-maximum, especially relevant for the Unity's flattening filter-free beam.
- EPID as a Validation Tool: While the EPID has limitations in coverage, its measurements of the central MLC leaves are used to validate the film-based results and aid in film registration, providing a cross-check.
Key Results
- Comprehensive Coverage: The digitally stitched films successfully covered nearly the entire MLC leaf banks, including their full travel range.
- Accuracy: The film-based method achieved a mean picket position deviation of -0.03 ± 0.13 mm (maximum 0.80 mm) and a mean picket width variation of 0.00 ± 0.13 mm (maximum 0.74 mm).
- Sensitivity: The technique demonstrated the ability to detect MLC positioning errors greater than 1.0 mm. In contrast, the EPID was able to detect errors of 0.5 mm or above with submillimeter accuracy.
- Reproducibility: EPID measurements showed excellent short-term reproducibility for the central MLC leaves, with position deviations under 0.3 mm.
Limitations and Future Work
- Limited Coverage of Outermost Leaves: Four leaf pairs on each side could not be fully imaged due to their position outside the treatment couch, and their signal was further compromised by electron return effects. The authors suggest extending the film and phantom setup to cover these leaves.
- Time Consumption: Due to the need for a large number of monitor units (MUs) to achieve sufficient film signal, the measurement process takes approximately 30 minutes on the Elekta Unity, which has a relatively low dose rate.
- Film Variability: The authors acknowledge that inherent variations in film response, scanning, and environmental conditions can contribute to uncertainty and potentially affect the sensitivity to introduced errors.
- Future Work: The authors suggest extending the film and phantom setup to cover the outermost leaf pairs for a complete MLC assessment. Further investigation into optimizing the MU delivery to reduce measurement time could also be beneficial.
Summary updated