Accurate measurement of the dose characteristics of a treatment beam is a key requirement for accurate determination of the dose delivered to a patient during radiation therapy. In beam commissioning, thimble ionization chambers with a 0.1-cc volume are the gold standard for measuring beam data, including output factors, depth–dose curves, and lateral profiles. However, point-by-point measurements with an ionization chamber are time-consuming tasks, and it is tedious to collect all of the profiles with various field sizes and depths for different energies. In addition, due to its large measurement size of 5–6 mm, the ionization chamber causes a volume-averaging effect that blurs the penumbras of lateral profiles. An alternative method to overcome these drawbacks of ionization chambers is to use a linear diode array detector to measure the lateral profiles. However, due to the higher atomic mass of the silicon in silicon diodes, the detector overresponds to low-energy photons that are more abundant in scatter, which increases with large fields and larger depths. Therefore, diodes intended for use in photon fields commonly also have a shield made of a high atomic number material (usually tungsten) integrated into the encapsulation to selectively absorb low-energy photons to which silicon diodes would otherwise overrespond [1,2].

A difficulty is the spectral variation over field size, depth, and off-axis distance (OAD) from the beam center that cannot be uniformly compensated by a single configured metallic compensator. The linear diode array detector, LDA-99SC (IBA Dosimetry, Schwarzenbruck, Germany) used in this study is also designed to shield low-energy scattered photons by applying a tungsten compensator plate beneath the diode detectors.

The study aim was to investigate the feasibility of using a linear diode array by comparing the results with those of an ionization chamber for commissioning of a treatment planning system.

For the beam commissioning of a TrueBeam^® with Analytical Anisotropic Algorithm (AAA) Eclipse^TM treatment planning system ver.15.6 (Varian Medical Systems, Palo Alto, CA, USA), an IBA CC13 ionization chamber was used to measure all of the beam data collected for 6- and 10-MV photon energy beams under flattening filtered (FF) and flattening filter free (FFF) conditions. The beam data set for each beam energy included output factors, depth–dose curves, and lateral profiles at various field sizes and depths. The CC13 ionization chamber is widely used for beam commissioning and measurement of the representative beam data provided by Varian [3].

In addition, an IBA LDA-99SC linear diode array detector was used to measure the profile data for comparison with the CC13 ionization chamber data, which enabled assessment of the feasibility of the LDA-99SC. This detector consists of 99 Hi-pSi detectors spaced only 5 mm apart aligned in a row, so it can measure a range equal to the length of the detector array at the same time (Table 1).

Table 1 . Characteristics of the LDA-99SC linear diode array detector and CC13 ionization chamber.

Feature	Characteristic
	LDA-99SC	CC13
Type	Chip	Thimble cylinder
Detector material	Hi-pSi	C552*
Material density (g/cm3)	2.33	1.76
Typical sensitivity (nC/Gy)	100	4.00
Active detector diameter (mm)	1.6	6
Sensitive volume (cm3)	1.6×10-4†	0.13
Detector axis alignment to beam axis	Parallel	Perpendicular

The structure of the diode detector is different from that of the ionization chamber, and the structural characteristics cause a difference in the results when measuring the beam. In Table 1, the LDA-99SC and CC13 characteristics are compared. Ionization chambers re widely used for beam measurements because of their stable response over a range of energies, doses, and dose rates as well as its relatively low cost and high availability. However, the volume-averaging effect due to a finite detector size limits the spatial resolution, especially in small-field dosimetry [4]. Si diode detectors have advantages, such as a fast response time, high spatial resolution, and high sensitivity. On the other hand, there are also disadvantages, such as the dose rate, energy dependence, and overresponse to low-energy scattered radiation generated because of the relatively large photoelectric cross-section of silicon (z=14). This overresponse can be reduced by using a metal shield [1,2]. In the LDA-99SC, a thin tungsten plate is used under the diode chip to remove low-energy photons.

The LDA-99SC can save measurement time because its 99 diodes are arranged in an array. In particular, because the convolution effect is small when measuring the penumbral area, the profile drops sharply in the penumbra [5].

In this study, the following characteristic were compared and analyzed to investigate the feasibility of using the LDA-99SC for beam commissioning.

1. Comparison of the Varian Golden Beam (GB) profiles with the CC13 profiles

In this study, the CC13 ionization chamber was used to measure the profiles, which were used as the reference data. The CC13 profiles using a 0.12 cm step width, 0.3–0.5 cm/s scan speed, and 0.75 cm/s positioning speed were measured. To verify the reliability of the CC13 profiles measured in this study, before comparing the LDA-99SC, and CC13 profile measurements, the agreements between the measured CC13 profiles and Varian GB data were confirmed for the TrueBeam^®’s four energies (6 MV, 6 MV FFF, 10 MV, 10 MV FFF), field sizes (3×3 cm², 6×6 cm², 10×10 cm², 20×20 cm², 30×30 cm², 40×40 cm²), and depths (dmax, 5, 10, 20, 30 cm).

2. Comparison of the LDA-99SC profiles with the CC13 profiles

The LDA-99SC profile data were measured under the same conditions as used for the CC13 profile data, and the two datasets were compared and analyzed. The scan parameters were set with a 0.1 cm step distance, 0.3 cm/s in-scan positioning speed, and 1 second measurement time. Therefore, it took approximately 5 seconds to measure a single profile with the LDA-99SC. Although the scan time of the CC13 varied with the field size, it took 60–100 seconds on average to measure a single profile, which was 12–20 times longer than it took with the LDA-99SC. To quantitatively compare the profile of the FF beams (6 MV FF and 10 MV FF), and the FFF beams (6 MV FFF and 10 MV FFF), the penumbral widths determined by 80%–20% of the profile were compared. For the FFF beams, the penumbral widths were determined by Fogliata et al.’s method [6], which determined the inflection point of the dose profile to renormalize the FFF beam to the same dose level as that of the corresponding FF beam at the inflection point on the field edge. We applied the suggested renormalization factors of the FFF beams to determine the penumbral widths of the FFF beams (Table 2).

Table 2 . Comparison of the penumbral widths measured with the LDA-99SC and CC13 and calculated with the AAA_LDA99 and AAA_CC13 beam models.

Energy	Measured penumbral width (mm)			Calculated penumbral width (mm)
	LDA-99SC	CC13		AAALDA99	AAACC13
6 MV	3.2	5.6		3.6	3.7
6 MV FFF	3.1	5.4		3.3	3.2
10 MV	4.1	6.3		4.6	4.6
10 MV FFF	3.8	6.1		4.3	4.3

3. Comparison of the LDA-99SC and CC13 beam models

Two independent beam data sets, including each profile data set measured separately using the CC13 and LDA-99SC, were imported into the Varian Eclipse^TM Beam Configuration and processed by the AAA ver.15.6 beam model to give the AAA_CC13 and AAA_LDA99 beam models. The effective target spot size in the X- and Y-directions that were necessary for AAA beam modeling were set to the default value of zero without any further tuning. Another important beam modeling parameter, the dosimetric leaf gaps of the Varian HighDefinition 120 Multileaf Collimator were determined using the methodology suggested by Wasbø and Valen [7], which was ranged from 0.0732 cm for 6 MV FFF to 0.091 cm for 10 MV. The AAA_LDA99 and AAA_CC13 dose profiles were compared.

4. Comparison of the AAA_LDA99 and AAA_CC13 clinical plans

To confirm the effect of the difference in the beam data due to the different detector types on the actual treatment plan after beam modeling, the dose distributions calculated by applying the AAA_CC13 and AAA_LDA99 beam models to the clinical plans were compared. Ten clinical plan cases, including treatment sites, such as the spine, lung, chest, liver, pelvis, and brain, were selected, and compared. Four energies (6 MV, 6 MV FFF, 10 MVA, 10 MV FFF) were used. All of the plans were originally generated with the AAA_CC13 beam model and then recalculated with the AAA_LDA99 beam model. A dose-subtraction map was generated to measure the voxel-by-voxel differences of the calculated dose distributions between the AAA_CC13 and AAA_LDA99 models. In addition, the dose statistics, such as the mean, minimum, and maximum dose for the planning target volume (PTV) as well as the entire dose calculation volume, were analyzed using the Eclipse^TM’s dose statistics function.

1. Comparison of the CC13 and Varian GB data profiles

Fig. 1 shows a comparison of the typical dose profiles measured with the CC13 and Varian GB. After examination of all of the ranges of energies, field sizes, and depths, it was confirmed that the measured CC13 profiles in this study agreed within 1% with the GB profiles except for the penumbral region. In the penumbral region, the measurement results showed a difference of ≤4% between the CC13 and GB profiles. This greater difference was thought to be due to minute differences in jaw calibration depending on the equipment used.

Figure 1. Comparison of the profiles (40×40 cm², dmax) measured by CC13 in this study with those of Varian Golden Beam data. (a) 6MV, (b) 6MV FFF, (c) 10MV, (d) 10MVFFF.

2. Comparison of the LDA-99SC and CC13 profiles

Figs. 2 and 3 show comparisons of the CC13 and LDA-99SC profiles, respectively, measured at 6 MV, and 10 MV, with, and without the FF. In the penumbral region, ≤5% differences that rapidly changed from + to − were observed, which appeared to be caused by the difference in the detector sizes that caused a greater volume-averaging effect on the CC13 profiles than on the LDA-99SC profiles. For quantitative analysis, the penumbral width determined in 80%–20% of the dose profiles are summarized in Table 2. Regardless of the presence of the FF, the difference in the penumbral widths were greater at 10 MV than at 6 MV by approximately 1 mm. Additionally, as the OAD increased, the measurement became slightly lower when using the LDA-99SC than when using the CC13. The difference was slightly greater for the FF beam than for the FFF beam, which may have been caused by the metallic compensator of the LDA-99SC (discussed later).

Figure 2. Comparison of the LDA-99 and CC13 profiles measured at 6 MV with a flattening filter (FF) or flattening filter free (FFF) (field size=40×40 cm²): (a) 6 MV FF at dmax, (b) 6 MV FFF at dmax, (c) 6 MV FF at a depth of 30 cm, (d) 6 MV FFF at a depth of 30 cm.

Figure 3. Comparison of the LDA99 and CC13 profiles measured at 10 MV with a flattening filter (FF) or flattening filter free (FFF) (field size=40×40 cm²): (a) 10 MV FF at dmax, (b) 10 MV FFF at dmax, (c) 10 MV FF at a depth of 30 cm, (d) 10 MV FFF at a depth of 30 cm.

3. Comparison of the AAA_LDA99 and AAA_CC13 beam models

After beam modeling, the profiles calculated by the AAA_LDA99 were compared with those calculated by the AAA_CC13 at 6 MV and 6 MV FFF in Fig. 4 and for 10 MV and 10 MV FFF in Fig. 5, respectively.

Figure 4. Comparison of the AAA_LDA99 and AAA_CC13 profiles at 6 MV with a flattening filter (FF) and flattening filter free (FFF) (field size=40×40 cm²): (a) 6 MV FF at dmax, (b) 6 MV FFF at dmax, (c) 6 MV FF at 30 cm, (d) 6 MV FFF at 30 cm. AAA, Analytical Anisotropic Algorithm.

Figure 5. Comparison of the AAA_LDA99 and AAA_CC13 profiles at 10 MV with a flattening filter (FF) and flattening filter free (FFF) (field size=40×40 cm²): (a) 10 MV FF at dmax, (b) 10 MV FFF at dmax, (c) 10 MV FF at 30 cm, (d) 10 MV FFF at 30 cm. AAA, Analytical Anisotropic Algorithm.

A comparison of the LDA-99SC and CC13 measurement profiles (at 6 MV, Fig. 2; at 10 MV, Fig. 3) showed a reduction in the profile differences between the beam models. Especially in the penumbral region, the differences were significantly reduced to ≤1% (Fig. 4 and 5). This reduction was because after beam modeling, the penumbral widths of the CC13 became closer to those of the LDA-99SC, whereas the latter did not significantly change after beam modeling (Table 2). After modeling, the penumbral width differences between the two detectors were almost identical (≤0.1 mm).

4. Comparison of the AAA_LDA99 and AAA_CC13 clinical plans

As summarized in Table 3, the mean dose difference between AAA_LDA99 and AAA_CC13 agreed within 0.5% in the PTV and within 0.4% in the body for each clinical plan. The average maximum point dose difference between the AAA_LDA99 and AAA_CC13 was −1.5% (−3.3% to −0.2%). A typical AAA_LDA99 and AAA_CC13 dose-difference map is shown in Fig. 6, with the dose profiles across the PTV in the cranial–caudal direction. The map shows a rapid change from + to – when crossing the beam margin, which indicates a subtle difference in the penumbral width between the AAA_LDA99 and AAA_CC13 beam models.

Table 3 . Dose differences between AAA_LDA99 and AAA_CC13 for 10 clinical treatment plans with various disease sites, energies with a FF and FFF, and treatment volumes.

Energy	Plan technique	Treatment site	Prescription (Gy)	PTV (cm3)	Mean dose difference (AAALDA99−AAACC13)						Max dose difference (AAALDA99−AAACC13)
					PTV			Body
					cGy	%		cGy	%		cGy	%
6 MV FF	2 full arcs VMAT	Brain	35	22.21	6.1	0.1		−0.6	−0.2		43.9	−1.0
6 MV FF	2 full arcs VMAT	C-Spine	45	364.54	17.6	0.4		−3.6	−0.1		30.2	−0.7
6 MV FF	9 fields IMRT	Lung	40	812.75	−19.0	−0.3		−20.2	−0.4		127.8	−3.2
6 MV FF	7 fields IMRT	Chest	50	364.54	2.3	0.2		−0.9	−0.1		11.3	−0.2
6 MV FFF	2 half arcs VMAT	Lung	32	77.92	14.2	0.5		−0.2	0.0		104.1	−3.3
6 MV FFF	2 half arcs VMAT	Lung	54	5.49	15.9	0.3		0	0.0		99.7	−1.8
10 MV FF	2 full arcs VMAT	Pelvis	30	405.38	−0.1	0.0		−3.5	−0.1		32.7	−1.1
10 MV FF	2 half arcs VMAT	Pelvis	30	200.33	−2.1	0.0		−1.6	−0.1		24.4	−0.8
10 MV FFF	2 half arcs VMAT	Liver	45	32.69	−90.0	−0.2		−0.4	−0.1		74.4	−1.7
10 MV FFF	2 half arcs VMAT	Liver	45	20.79	9.5	0.2		−0.3	0.0		−43.0	−1.0

Figure 6. Dose difference between AAA_LDA99 and AAA_CC13 for liver stereotactic body radiotherapy. (a) Dose-difference color map in the sagittal plane. (b) Dose profile as it crosses the planning target volume in the cranio–caudal direction. AAA, Analytical Anisotropic Algorithm.

As shown in Figs. 2 and 3, in the out-of-field region, the dose profiles of LDA-99SC are relatively lower than those of CC13. Additionally, the dose profiles obtained with FF beams were ~1% lower than those obtained with FFF beams. However, this tendency seems to begin inside the field as the OAD increases because of the energy dependency of the LDA-99SC, which differs from that of CC13. Since silicon in the Si diode of the LDA-99SC has an atomic mass of 28, overresponse may occur by absorbing many more low-energy photons than water, which has a molecular weight of 18. Therefore, the LDA-99SC detector is structurally designed to shield low-energy scattered photons to some extent for the purpose of compensation [8]. However, considering the FF shape, the distribution of low photons filtered by the FF varies depending on the OAD. The closer to the beam center, the more low-energy photons are filtered, and thus beam hardening occurs; conversely, more low-energy photons are left with increasing OAD, which has a beam-softening effect. Therefore, the LDA-99SC would over-compensate for the low-energy scatter by increasing the OAD, resulting in LDA-99SC measured profiles that are slightly lower than the CC13 profiles. On the other hand, since the FFF beam has no FF, its photon beam spectra will be similar regardless of the OAD and thus less affected by the over-compensation. This explanation is not conclusive and thus needs to be further addressed. In addition, the over-compensation of the LDA-99SC increases as the depth increases with more low-energy photons due to increasing phantom scatter.

In Table 2, which shows a quantitative comparison of the penumbral width, for the LDA-99SC, there was no significant difference in penumbral widths during measurement and after modeling. As shown in Table 1, the active detector size of the LDA-99SC is as small as 1.6 mm and thus has little effect on the volume-averaging effect. On the other hand, the penumbral width for the CC13 measured value is relatively larger than the LDA-99SC value because the detector size is about 5–6 mm, resulting in a large volume-averaging effect. However, the AAA_CC13 penumbra, after Eclipse beam modeling, was similar to the AAA_LDA99 penumbra. This result can be explained by correction of the CC13 volume-averaging effect through modeling. Therefore, after beam modeling, the penumbral width difference between the two detectors was ≤0.1 mm.

As shown in Fig. 6, the dose difference between the AAA_LDA99 and AAA_CC13 changed rapidly from + to – when crossing the beam margin, which indicated a remaining subtle difference in the penumbral widths between the AAA_LDA99 and AAA_CC13 beam models.

Comparison of the LDA-99SC and CC13 measured and calculated profiles is considered the gold standard for evaluating a detector for beam commissioning. Subtle differences were observed in our comparison that were caused by differences in detector sizes and energy dependencies. However, the differences were not significant and even smaller when comparing the clinical IMRT and VMAT plans. Therefore, these differences were judged to be at clinically acceptable levels, indicating that the LDA-99SC detector is an efficient and accurate detector that is suitable for photon beam commissioning of a treatment planning system.

The authors have nothing to disclose.

The data that support the findings of this study are available on request from the corresponding author.

Conceptualization: Jungwon Kwak and Byungchul Cho. Data curation: Seung Mo Hong and Uiseob Lee. Formal analysis: Seung Mo Hong and Uiseob Lee. Investigation: Sung-woo Kim and Youngmoon Goh. Methodology: Min-Jae Park and Chiyoung Jeong. Writing – original draft: Seung Mo Hong and Uiseob Lee. Writing – review & editing: Jungwon Kwak and Byungchul Cho.