Equivalency of beam scan data collection using a 1D tank and automated couch movements to traditional 3D tank measurements

Abstract This work shows the feasibility of collecting linear accelerator beam data using just a 1‐D water tank and automated couch movements with the goal to maximize the cost effectiveness in resource‐limited clinical settings. Two commissioning datasets were acquired: (a) using a standard of practice 3D water tank scanning system (3DS) and (b) using a novel technique to translate a commercial TG‐51 complaint 1D water tank via automated couch movements (1DS). The Extensible Markup Language (XML) was used to dynamically move the linear accelerator couch position (and thus the 1D tank) during radiation delivery for the acquisition of inline, crossline, and diagonal profiles. Both the 1DS and 3DS datasets were used to generate beam models (BM 1 DS and BM 3 DS) in a commercial treatment planning system (TPS). 98.7% of 1DS measured points had a gamma value (2%/2 mm) < 1 when compared with the 3DS. Static jaw defined field and dynamic MLC field dose distribution comparisons for the TPS beam models BM 1 DS and BM 3 DS had 3D gamma values (2%/2 mm) < 1 for all 24,900,000 data points tested and >99.5% pass rate with gamma value (1%/1 mm) < 1. In conclusion, automated couch motions and a 1D scanning tank were used to collect commissioning beam data with accuracy comparable to traditionally acquired data using a 3D scanning system. TPS beam models generated directly from 1DS measured data were clinically equivalent to a model derived from 3DS data.

looking at the discrepancy in the access to advanced treatment modalities for low-and middle-income countries vs high-income countries. 2 Efforts to solve this problem continue as the National Institute of Health has recently announced funding opportunities for the development of cancer-relevant technologies for low-and middle-income countries (RFA-CA-15-024). 3 There are many challenges to overcome in radiation oncology; acquisition of commissioning beam data is a prime example.
The World Health Organization estimated that approximately 750 of 3125 (24%) reported adverse advents in radiation oncology stemmed from the commissioning stage. 4 Beam data acquisition is an important step in the commissioning process, as it is the foundation for subsequent beam modeling. Errors made during beam data acquisition and modeling are particularly hazardous, since these errors will be systematic and propagate to impact every patient treated on a given machine. Therefore, it is crucial this process be accurate and error free. The beam data acquisition process involves the use of sophisticated scanning software to position the detector and take readings; however, this is often labor intensive. Beam scanning systems are not integrated with treatment systems as changes in the scanning software do not automatically translate to changes in the machine parameters (e.g., jaw settings or energy selection) and thus can be error prone (AAPM TG-106). 5 Furthermore, beam modeling becomes more critical as the complexity of treatment increases (e.g., SBRT & IMRT). 6 Currently, guidelines exist for ensuring best practices during the beam scanning process, 5 treatment planning system commissioning process, 6 and in the continued quality assurance of treatment planning systems. 7 The task groups underscore the importance of using precise and accurate equipment that, on the other hand, can come at a high financial cost. Furthermore, despite the presence of these guidelines, there is still substantial variability in the quality and accuracy of commissioning in the United States as seen by third party audits of institutions. [8][9][10][11] One possible cause could be a shortage for personnel proficient in these procedures to provide services. 12 This work presents a novel method to lower the financial and equipment barriers needed to acquire a full dosimetric commissioning dataset by presenting a departure from traditional nonintegrated 3D scanning systems (3DS), and by transitioning to the synergistic and efficient use of a compact 1D water tank and automated translation of the linear accelerator couch system (1DS) via the extensible markup language (XML). The logistical characteristics of the 1DS and 3DS systems per the manufacturer's technical data sheet highlight the differences between the two systems. In all datasets, central axis depth profiles and lateral profiles were Each profile collected via the 1DS was then compared and plotted to the paired profile from the 3DS dataset using a custom 1D gamma analysis code 14 Fig. 1(a).   Fig. 1(a)] and histogram of the gamma values from these profiles [ Fig. 1(b)] highlight this agreement.

3.A | Beam data collection
All measured profiles were compared at 2%/2 mm and these data are summarized in Table 2. Over 98.7% of all data points yielded gamma values <1. The mean gamma value across all profiles was 0.241. Figure 2(a) shows the off axis profiles for the various field sizes and depths. Note for plotting, all profiles were normalized to the central axis of a 10 × 10 cm 2 at 1.5 cm depth; however, for the gamma analysis each profile was normalized to its own central axis.
A histogram of the gamma values for these profiles is plotted in Fig. 2(b).

3.B | Beam modeling
To quantify the differences in the two beam models created (BM 1DS and BM 3DS ), beams of various field sizes were calculated on a water phantom in the treatment planning system and compared using a 3D gamma metric for each field size. The results are summarized in showing the 3D gamma value distribution, axial, coronal, and sagittal planes for the dynamic chair field is presented in Fig. 3. Histogram data of the gamma values were collected and are presented in Fig. 4. Even though there appears to be a logistical difference between 1DS and 3DS, many of the short comings of 3D tanks have been addressed in 2D tanks which are substantially cheaper and hence do find wide clinical acceptance worldwide. Even though the 1DS appears to be promising, there is substantial initial development time and QA cost as no commercial system is yet available. This is important as 2D/ 3D scanners need FDA-510k clearance before they can be sold. The latter assures the users about the quality of the system. Barring some electronic or motor drive assembly space, nearly the entire volume of the 2D/3D tank is available for data acquisition. This information is available as a specification of the tank by the company. With the 1DS, the free 3D space around couch restricts its range and can vary from linear accelerator type to another. The motion mechanism of a 2D/3D scanner is usually used very infrequently in a clinic and usually gets minimal wear and tear and hence results may be more trustworthy. The 1DS, on the other hand, relies on couch motion accuracy which is subjected to continuous and torturous use every day implying more wear and tear. Hence, every time the 1D tank is used, extensive QA on the couch needs to be carried out. Currently, the system could only be used if the linear accelerator is relatively new supporting XML language for its couch control. In a department which has a mix of different linear accelerator types, it might be more cost effective to have a 2D/3D tank based scanning system which can be used with any of them.

| DISCUSSION
[Correction added on September 14 2018, after first online publication: Under Discussion section "The latter assures the users about the quality of the system." sentence was modified.]

| CONCLUSION
Using a 1D tank and automated couch motions, a full 6 MV commissioning dataset was collected and produced a beam model clinically equivalent to traditional 3D tank based methods. This method could provide a valuable alternative option for commissioning a linac in developing and resource-limited countries, or for systems where the 3D tank is not feasible.

CONFLI CT OF INTERESTS
The authors declare no conflict of interest.