The modeling of the Elekta Synergy Agility linear accelerator head using the GAMOS code was conducted to simulate a 6 MV photon beam. This Monte Carlo simulation platform, dedicated to various medical applications such as radiotherapy, gamma spectrometry, shielding, Compton camera, proton therapy, and positron emission tomography (PET), is built upon Geant4. The modeling encompasses the three electromagnetic physics models and the three multiple scattering models of the GAMOS code. The modeled components of the accelerator head include the target, primary collimator, flattening filter (considered crucial for beam flattening at the exit of the primary collimator), ionization chamber, and X and Y jaws. All technical data required for modeling were provided by the manufacturer, covering dimensions, chemical composition, and density of different structures. However, adjustments were made to certain parameters such as energy and flattening filter. The visual representation of the LINAC irradiation head Figure 1 in combination with the water phantom was created using the VRML2FILE function of the GAMOS code with view3dscene. Data analysis was performed using the Python language and the Visual Studio Code (VSC) development environment. All obtained results were compared to experimental data from the International Cancer Center of Dakar in Senegal. This systematic approach and the use of GAMOS to model the accelerator head demonstrates our commitment to simulation accuracy, reinforced by systematic comparison with experimental data from ICCD.

Following the modeling of the linear accelerator components (target, primary collimator, flattening filter, ionization chamber, and X and Y jaws), we generated a phase space file following the format of the International Atomic Energy Agency (IAEA) [26] , aiming to become a Monte Carlo simulation standard. Positioned between the jaws and the water phantom, this file generated 3 × 10⁸ events using the following electromagnetic physics models: Standard, Low-Energy, and Penelope, along with the three multiple scattering models: Goudsmit-Saunderson, Urban, and Wentzel-VI. These simulations were conducted for field sizes of 5 × 5 cm² and 10 × 10 cm² at depths of 10 cm and 20 cm.

Physical characteristics of each particle passing through the recording plane were recorded in the phase space file, including information on particle types, energy, position, direction, and statistical weight. This phase space, positioned after the jaws, was used as a virtual source to calculate the dose distribution in the phantom for each of the three electromagnetic physics models and the three multiple scattering models. To enhance simulation efficiency, phase space particles were recycled 50 times for both electromagnetic physics and multiple scattering models, utilizing a XY mirror to reflect the reused particles.

In radiotherapy, dose distributions in a water phantom play a crucial role in characterizing the incident beam. The curve showing the variation of absorbed dose in water along the beam axis is referred to as the depth-dose curve. Meanwhile, the variation of absorbed dose in a plane perpendicular to the beam axis is designated as the dose profile. For our simulations, a water phantom with dimensions of 60 × 60 × 41 cm³ was utilized. We started with the initial voxel size of 120 × 120 × 41 along the x, y, z axes which was further divided into smaller voxels with dimensions of 5 × 5 × 10 mm³ and a density of 1 g.cm³.

This water phantom was placed at a source-surface distance (SSD) of 90 cm from the target, with field sizes of 5 × 5 cm² and 10 × 10 cm² at depths of 10 cm and 20 cm. These configurations were employed for the three electromagnetic physics models (Standard, Penelope, and Low-Energy) and the three multiple scattering models (Goudsmit-Saunderson, Urban, and Wentzel-VI) when modeling dose profiles. For depth-dose curves, the water phantom was positioned at 100 cm from the target. In experimental setups, measurements were conducted using a water phantom available at the International Cancer Center of Dakar in Senegal. Two detectors were utilized: a PTW-Freiburg microDiamond model (made in Germany) for dose profile data and a PTW-Freiburg Semiflex ionization chamber with a volume of 0.3 cm³ (made in Germany) for depth-dose measurements.

The Gamma Index (GI) used in our study is a mathematical tool employed for the quantitative comparison of two dose distributions in radiotherapy. The gamma index values represent the agreement between the measured and simulated dose profiles. Its purpose is to quantify the accuracy of the delivered dose during treatment by comparing a reference dose distribution (D_r) with an evaluated dose distribution (D_c). This method, introduced by [27] , and later by [28] , takes into account two parameters: the dose difference criterion (ΔD in %) and the distance difference criterion (Δd or DTA—Distance To Agreement in mm) between two points. The formula for the gamma index is defined as follows:

where D_r is reference dose distribution at distance d_r; D_c is the dose distribution to be evaluated at distance d_c; ΔD is the criterion for dose difference (in English, dose deviation DD); and Δd is the criterion for distance difference (in English, distance to agreement DTA). If the gamma index is less than 1, it means that the comparison between the measured and calculated points is acceptable according to the set tolerance criteria. However, if the gamma index is greater than 1, the test is not satisfied as it falls outside the acceptability ellipse. The acceptance criteria for the gamma index (GI) test used in our research are a dose deviation of 3% and a distance difference of 3 mm.

In this study, we characterized the incident beam in radiotherapy using dose distributions in a water phantom. Depth-dose curves were calculated along the central axis of the irradiation beam, and dose profiles were calculated at a depth perpendicular to the central axis, for field sizes of 5 × 5 cm² and 10 × 10 cm². We employed the three electromagnetic physics models: Standard, Penelope, and Low-Energy, as well as the three multiple scattering models: Goudsmit-Saunderson, Urban, and Wentzel-VI. All these curves were compared to the experimentally measured data. The adjustment of simulated values to experimental data was performed using maximum normalization, where the dose per voxel was divided by the maximum dose. Subsequently, we used the gamma index with an acceptable criterion of 3% for dose difference and 3 mm for distance-to-agreement, following the study by [27] . These criteria allowed us to assess the consistency between experimental and simulated values. The results demonstrate satisfactory agreement between experimental and simulated values Table 1 and Table 2 for the three electromagnetic physics models and the three multiple scattering models. The gamma index is 94% for depth-dose curves for a field size of 5 × 5 cm² Figure 2 and Figure 3 and 96% for a field size of 10 × 10 cm² Figure 4 and Figure 5.

These findings reinforce the validity of our modeling approach, highlighting the capability of GAMOS code models to faithfully reproduce the dosimetry characteristics of the irradiation beam.

Concerning dose profiles at depths of 10 cm and 20 cm, the Penelope, Standard, and Low-Energy electromagnetic physics models show 100% agreement Table 3 for a field size of 5 × 5 cm² Figure 6 and Figure 7. Similarly, at a depth of 10 cm and 20 cm with a field size of 5 × 5 cm² Figure 8 and Figure 9, the multiple scattering models: Goudsmit-Saunderson, Urban, and Wentzel-VI show 100% agreement Table 4. For a field size of 10 × 10 cm² at a depth of 10 cm, the Penelope model shows 98% agreement, followed by the Standard model at 96%, while the Low-Energy model exhibits a slightly lower agreement of 92% Figure 10 and Table 3.

At 20 cm depth, the Penelope and Standard models show 100% agreement, followed by the Low-Energy model at 99% Figure 11 and Table 3. For the same field size of 10 × 10 cm² at 10 cm depth, the multiple scattering models: Goudsmit-Saunderson, Urban, and Wentzel-VI exhibit 98% agreement each Figure 12 and Table 4, while at 20 cm depth, these models show 100% agreement each Figure 13 and Table 4.

It is important to note that [15] also used the Standard, Livermore, and Penelope electromagnetic physics models and strongly recommended the use of the Standard model.

We observe good agreement between experimental and simulated curves for the field size of 5 × 5 cm² at depths of 10 cm and 20 cm for all three electromagnetic physics models and the three multiple scattering models (100%).

However, for a field size of 10 × 10 cm², discrepancies in dose profile curves are noted for the three electromagnetic physics models. The difference between

the found models is due to statistical fluctuations.

These differences could be attributed to potential variations between manufacturer-provided values and the actual characteristics of the LINAC installed at the hospital, emphasizing the importance of precision in accelerator geometry modeling.

The good agreement observed in the comparison of depth dose curves and simulated dose profiles, utilizing electromagnetic physics models and multiple scattering models, with experimental measurements, indicates that our study offers vital information for selecting models that guarantee the most reliable results. This paves the way for broader adoption in treatment planning. Additionally, this study highlights that the phase spaces generated in each of the electromagnetic physics and multiple scattering models can be utilized as templates for simulating a 6 MV photon beam from an Elekta Synergy linear accelerator, making them accessible to researchers interested in conducting further studies.