TL;DR
This study reveals that the accuracy of portal dosimetry, a crucial quality assurance step in radiation therapy, is significantly impacted by how "off-center diagonal" (OCD) profile depths are configured in both the treatment planning system and the imaging device. For standard flattened beams, mismatched depth settings can lead to substantial drops in accuracy, highlighting the need for careful configuration in QA protocols.
What Problem Does This Paper Address?
Modern radiation therapy, especially techniques like IMRT and VMAT, requires precise delivery of radiation beams. To ensure this precision, a quality assurance (QA) step called portal dosimetry is used. This involves comparing a predicted radiation image, generated by the treatment planning system (TPS) using a portal dosimetry image prediction (PDIP) algorithm, with an actual image captured by an electronic portal imaging device (EPID) during treatment.
A key component in this process is the use of "off-center diagonal" (OCD) profiles. These are measurements of how radiation intensity changes across a beam at different depths and off-axis positions. These profiles are used by both the TPS (to predict the image) and the EPID (to correct the measured image). However, there's been limited guidance on how the specific depth chosen for these OCD profiles in the TPS and EPID might affect the overall accuracy of the portal dosimetry comparison. This paper addresses this gap by systematically investigating the impact of pairing different OCD profile depths between the TPS and EPID.
Methods
The researchers systematically evaluated the impact of OCD profile depth pairing on gamma pass rates, a common metric for assessing agreement between predicted and measured dose distributions.
- OCD Profile Acquisition: OCD profiles were measured at various depths (reference depth, 5, 10, 20, and 30 cm) for four photon beam types: 6X, 10X, 6FFF (flattening filter-free), and 10FFF.
- Treatment Planning and Prediction: For each beam type, five different OCD profile depths were configured in the TPS's PDIP model.
- EPID Configuration: For each TPS configuration, EPID measurements were performed using five different OCD profile depths for image correction.
- Experiment Design: This resulted in 25 unique TPS-EPID depth pairings for each of the four beam types. These were tested across seven different square field sizes (5x5 cm² to 30x30 cm²).
- Gamma Analysis: A standard 3%/3 mm gamma criterion with a 10% dose threshold was used to evaluate the agreement between the predicted and measured images.
- Statistical Analysis: A two-way ANOVA was employed to assess the effects of TPS OCD depth, EPID OCD depth, and their interaction on gamma pass rates for the 6X and 10X beams.
Datasets
The study utilized portal images acquired from a Varian TrueBeam linear accelerator equipped with an aS1200 EPID. The data consisted of measurements from four photon energies (6X, 10X, 6FFF, 10FFF) across seven field sizes. No new benchmark datasets were introduced; rather, existing clinical QA data and measurement protocols were used for evaluation.
Key Insights and Ideas
- Depth Matters for Flattened Beams: The most significant finding is that the choice of OCD profile depth is not trivial for flattened (FF) beams (6X and 10X). Mismatched depths between the TPS and EPID can lead to a noticeable decrease in gamma pass rates, indicating a less accurate verification.
- "Deeper is Better" Trend: For FF beams, gamma pass rates generally improved when the EPID's OCD depth was matched to or was deeper than the TPS's OCD depth. This suggests that the EPID's correction might be more robust when it accounts for deeper beam characteristics.
- FFF Beams are Robust: Flattening filter-free (FFF) beams (6FFF and 10FFF) showed remarkable insensitivity to OCD depth pairing. High gamma pass rates were consistently achieved across all configurations, suggesting that their beam profiles are less dependent on depth in a way that impacts this specific QA metric.
- Clinical Implications for QA: This study provides quantitative evidence that OCD depth alignment should be a critical consideration during the commissioning and ongoing QA maintenance of portal dosimetry systems, especially for FF beam deliveries.
Key Results
- For the 6X beam, gamma pass rates varied significantly with depth pairing. The highest pass rate (100%) was achieved with matched or deeper EPID depths (e.g., D20_E30, D30_E30), while the lowest (71.5%) occurred with a mismatched configuration (D30_Edr).
- For the 10X beam, similar trends were observed, with the maximum pass rate (100%) for D20_E30 and a minimum of 58.5% for D30_Edr.
- For 6FFF and 10FFF beams, gamma pass rates consistently exceeded 95% across all tested TPS-EPID OCD depth pairings and field sizes, with minimal standard deviations (<1%).
- A supplementary analysis using a stricter 2%/2 mm gamma criterion confirmed the depth dependence for FF beams and showed that FFF beams remained largely unaffected, reinforcing the robustness of the findings.
Limitations and Future Work
- Single Institution/System: The study was conducted at a single institution using a specific TPS and linac system. The generalizability of the optimal depth pairings to other platforms or vendors is not yet established.
- Limited FFF Analysis: Due to the minimal variability in FFF beam results, statistical analyses were primarily focused on FF beams. Future work could explore higher-resolution metrics or alternative comparison methods to detect subtler trends in FFF behavior.
- Other Model Parameters: The study focused solely on OCD profile depth pairing. Other crucial model parameters, such as effective source size, off-axis softening, and multileaf collimator modeling, were not investigated.
- Future Directions: The authors suggest that future research should include multi-institutional datasets and broader modeling parameters to validate and extend the clinical relevance of depth alignment. They also propose that identified optimal OCD depth pairings could serve as institution-specific reference points for commissioning and QA planning.
Summary updated