Abstract
Background: Due to no corresponding action limits for some new daily QA items, limits of the monthly or annual QA procedures have to be used for daily QA programs at present. The general purpose of this paper is to discuss whether it is feasible to do so. Methods: Three groups of daily quality control processes during 2017-2019 performed with Daily QA3 were analyzed using Statistical Process Control (SPC). In the calculation of capability indices for output, the daily QA tolerance of $\pm 3\%$ from TG-142 was used as the specification limit, while for flatness and symmetry, the annual QA tolerance of $\pm 1\%$ was used. Capability indices ($C_p, C_{pk}, C_{pm}, \text{and } C_{pmk}$) were calculated, and customized tolerance and action limits were established. Additionally, the consistency of data from five therapists was compared with that of a single physicist. Results: Process capability indices for output were all $\ge 1$, some reaching 3-4. Conversely, symmetry measurements often failed to meet the $\pm 1\%$ requirement. Calculated customized limits varied significantly between processes. Data collected by a single physicist showed narrower control limit ranges compared to therapist-collected data. Conclusions: It is necessary to establish customized limits for daily QA programs, as they improve the quality control level and operational constancy.
Introduction
Daily quality assurance (QA) is a cornerstone of patient safety in radiation therapy, ensuring that linear accelerators (linacs) operate within strict mechanical and dosimetric tolerances. Currently, many clinical practices rely on generalized standards, such as the American Association of Physicists in Medicine (AAPM) Task Group 142 (TG-142) report. While these guidelines provide a vital safety floor, they often utilize "one-size-fits-all" action limits that may not reflect the specific performance capabilities of modern linacs or the inherent variability of daily measurement devices like the Sun Nuclear Daily QA3. This study investigates the necessity of moving toward customized, process-based limits using Statistical Process Control (SPC).
Methodology
The study analyzed longitudinal data from an Elekta Synergy linac at Shanxi Bethune Hospital over a three-year period (2017–2019). The researchers employed SPC to evaluate the stability and capability of the daily QA process.
Statistical Process Control (SPC) Tools
- I-MR Charts: Individuals and Moving Range charts were used to track daily measurements, allowing for the differentiation between common-cause variation (random noise) and special-cause variation (systemic shifts).
- Process Capability Indices: The metrics $C_p, C_{pk}, C_{pm}, \text{and } C_{pmk}$ were calculated. A capability index $\ge 1$ indicates that the process is capable of meeting its specifications.
- Customized Limit Calculation: Rather than relying on static thresholds, the authors established action limits based on the actual distribution of their data using the Taguchi-type index ($C_{pm}$) and a specific scaling factor ($A$).
Data Collection and Human Factors
To assess the impact of operational consistency, the study compared two datasets:
- Therapist Group: 100 sets of daily data per year collected by five rotating therapists.
- Physicist Group: 30 sets of data collected by a single medical physicist over six weeks to evaluate the "best-case" operational constancy.
Results and Discussion
The Discrepancy in Output and Symmetry Limits
The analysis revealed significant flaws in applying generic limits to daily checks. For beam output, the TG-142 limit of $\pm 3\%$ was found to be excessively generous. The machine demonstrated capability indices between 3 and 4, indicating extreme stability. Using a $\pm 3\%$ limit in this context could allow significant machine drift to go unnoticed until it reaches a critical threshold.
In contrast, applying the annual symmetry tolerance of $\pm 1\%$ to daily measurements proved too restrictive. Most daily processes "failed" this standard. This failure was attributed not to linac malfunction, but to the higher inherent variability of the Daily QA3 device compared to the high-precision 3D water tanks used for annual calibrations. This highlights the need for device-specific daily tolerances to avoid "false positive" alarms.
Operational Constancy and Stabilization
The comparison between therapist and physicist data underscored the importance of the "human factor." The physicist’s measurements resulted in significantly narrower control limits on the I-MR charts. This suggests that variations in device setup—such as alignment with room lasers—contribute meaningfully to measurement noise. Furthermore, the study found that control lines on an SPC chart require approximately 30 to 40 data points (6–8 weeks) to stabilize, providing a practical timeframe for clinics to establish baseline performance for new equipment.
Detection of Equipment Aging
A key strength of the SPC approach was demonstrated in 2019, where the charts identified a systemic error after the 45th data point. While a standard threshold might have missed this shift, the SPC limits flagged it, allowing the team to trace the issue to the aging of the Daily QA3 detector rather than a fault in the linac itself.
Conclusion
The study concludes that "one-size-fits-all" QA limits are insufficient for high-precision radiotherapy. Establishing customized limits based on a machine's actual performance and the specific measurement tools used is essential. By adopting SPC-based customized limits and improving operational constancy, clinics can significantly enhance the sensitivity of their quality control programs, ensuring more reliable treatment delivery and proactive equipment maintenance.
Summary updated