Developing the advanced dosimetry of linac beam radiotherapy

<p dir="ltr">Photon beams remains the most widely used modality for radiotherapy of cancer worldwide. Continued technological development has enabled increasingly conformal treatment techniques such as intensity modulated radiotherapy (IMRT), volumetric modulated arc therapy (VMAT), and stereotactic radiotherapy (e.g., stereotactic body radiotherapy SBRT), which rely on small radiation fields, steep dose gradients, and tight geometric margins. While these advances improve the potential for tumor control and normal tissue sparing, they simultaneously reduce the tolerance for uncertainties throughout the radiotherapy chain. For example, limitations in small field dosimetry, residual differences between nominally beam-matched linear accelerators (linacs), accuracy of treatment planning system (TPS) beam modelling and performance of dose calculation algorithm under various conditions, can all introduce systematic deviations in absorbed dose delivery. The aim of this thesis is to improve the accuracy with which absorbed dose is delivered in external photon beam radiotherapy by addressing these sources of uncertainty.</p><p dir="ltr">The thesis comprises five studies spanning: beam matching, small field dosimetry, TPS beam model optimization, and dose calculation accuracy. Beam matching was studied across eight Varian TrueBeam linacs by extending tolerances beyond vendor acceptance criteria and focusing on clinically relevant dosimetric and mechanical parameters. Photon and electron beam symmetry, energy characteristics, jaw positioning accuracy, and multileaf collimator (MLC) calibration, were systematically evaluated and optimized. Verification measurements were performed to assess the impact of this integrated calibration strategy on dose delivery consistency.</p><p dir="ltr">Small field output factors (OFs) were measured using multiple solid state detectors on two sets of beam-matched Varian Clinac and TrueBeam linacs. Furthermore, the measurements on the TrueBeam linacs were repeated after five years of clinical operation to assess both inter-linac variability and temporal stability. Detector-specific output correction factors (OCF) published in the technical report series no. 483 (TRS-483) by the international atomic energy agency (IAEA), in collaboration with the American association for physicists in medicine (AAPM), were applied to evaluate their effectiveness in reducing detector-dependent differences and enabling clinically usable OF for small fields. Inter-linac equivalency was investigated by examining detector-reading ratios as a function of nominal field size.</p><p dir="ltr">TPS beam model optimization for stereotactic treatments was addressed by developing a practical protocol for tuning the effective spot size (ESS) and dosimetric leaf gap (DLG) parameters. Small MLC-defined field profiles were measured in water with careful control of detector positioning relative to MLC leaf geometry. A representative off-axis inline sampling strategy was introduced to mitigate bias arising from leaf-edge effects (referred to as off-axis inline translation, OIT). Beam models were optimized for both the anisotropic analytical algorithm (AAA) and Acuros XB algorithm (AXB), and verified using film- and electronic portal imaging device (EPID) measurements of clinical stereotactic plans.</p><p dir="ltr">Dose calculation accuracy of AXB was investigated through a physics-based comparison against full Monte Carlo (MC) simulations (using PENELOPE-based PRIMO), and the widely-implemented AAA. Calculations were performed in simplified phantom geometries containing well-defined material inserts (lung, adipose, muscle, cartilage and bone) and in a range of clinical patient geometries. Mean absorbed dose was evaluated in selected volumes to isolate material, interface, and volume-size effects. Dose-to-medium in medium (D_m,m) calculated by PRIMO and AXB, dose-to-water in water (D_w,w) calculated by PRIMO and AAA, and dose-to-water in medium (D_w,m) calculated by AXB, were examined to assess their consistency and clinical implications.</p><p dir="ltr">The results demonstrate that enhanced beam matching, incorporating imaging-based jaw calibration and rigorous MLC parameter alignment, achieved high dosimetric consistency across linacs for modulated deliveries. Pushing for more strict criteria for beam-matching among eight linacs was shown to be not only feasible, but advantageous. The presented methodology and the reported data may be used by other institutions as reference for guidance.</p><p dir="ltr">The application of TRS-483 OCF substantially reduces the intra-linac detector-dependent variability for all field sizes, especially below 4 cm x 4 cm. It is also shown that linac aging does not impact this application. However, for 0.5 cm x 0.5 cm, difference of several percent remains and is attributed to differences in beam characteristics between modelled linac from which OCF are derived, and actual linac which was used to measure the detector readings. Inter-linac variations over 1% is shown for field sizes below 1 cm x 1 cm (TrueBeam linacs) and 2 cm x 2 cm (Clinac linacs). This variation increases over five years of clinical use. Correlation with the penumbra width of 10 cm x 10 cm profiles, reflecting the focal spot size, was observed.</p><p dir="ltr">A proposed protocol for optimizing a beam model for stereotactic beams, incorporating both the DLG- and ESS parameters, was presented. The protocol presented an optimal OIT for measuring profiles with which the optimization can be performed. The protocol yielded improved agreement between calculated and delivered dose distributions, in comparison with the current practice.</p><p dir="ltr">AXB reproduced PRIMO-calculated D_m,m with high accuracy for all materials except for bone for which residual discrepancies were observed and linked to interface effects and algorithmic approximations regarding AXB assumption of uniform Bremsstrahlung production in the calculation volume. Comparison of D_w,w by PRIMO and AAA demonstrate a good agreement in all materials except for bone, which is well-established due to simplification in AAA. However, AXB-calculated D_w,m demonstrated even higher discrepancy compared to PRIMO-calculated D_w,w in bone, and to some extent, adipose. This discrepancy was attributed difference in fluence which is assumed to be similar between water and medium, in the calculation of D_w,m. For bone, the range of electrons and the size of the voxel further necessitate the partial application of the large cavity theory, requiring further corrections in the calculation of Dwm. Comparing corrected AXB Dwm against PRIMO-calculated D_w,w yielded similar accuracy as D_m,m.</p><p dir="ltr">In conclusion, this thesis demonstrates that accurate absorbed dose delivery in modern photon beam radiotherapy depends on several aspects, including small field dosimetry, linac equivalency, beam model configuration, and dose calculation algorithms. Improvements in individual components are necessary in realizing the overall aim of increased accuracy of the delivered dose.</p><h3 dir="ltr">List of scientific papers</h3><p dir="ltr">I. <b>Ghazal M,</b> Westermark M, Kaveckyte V, Carlsson-Tedgren Å, Benmakhlouf H. 6-MV small field output factors: intra-/intermachine comparison and implementation of TRS-483 using various detectors and several linear accelerators. Medical physics. 2019;46(11):5350-9. <a href="https://doi.org/10.1002/mp.13830" rel="noreferrer" target="_blank">https://doi.org/10.1002/mp.13830</a></p><p dir="ltr">II. <b>Ghazal M,</b> Södergren L, Westermark M, Söderström J, Pommer T. Dosimetric and mechanical equivalency of Varian TrueBeam linear accelerators. Journal of Applied Clinical Medical Physics. 2020;21(12):43-53. <a href="https://doi.org/10.1002/acm2.13058" rel="noreferrer" target="_blank">https://doi.org/10.1002/acm2.13058</a></p><p dir="ltr">III. Melhus T, <b>Ghazal M,</b> Benmakhlouf H. A novel protocol for parametrization of a beam model for small stereotactic beams. Journal of Applied Clinical Medical Physics. 2025;26(7):e70158. <a href="https://doi.org/10.1002/acm2.70158" rel="noreferrer" target="_blank">https://doi.org/10.1002/acm2.70158</a></p><p dir="ltr">IV. <b>Ghazal M,</b> Carlsson Tedgren Å, Benmakhlouf H. Physics-based dosimetric evaluation of Acuros XB photon dose calculation algorithm. [Manuscript]</p><p dir="ltr">V. <b>Ghazal M,</b> Aziz A, Benmakhlouf H. The impact of linac aging on the small field dosimetry of megavoltage photon beams. [Manuscript]</p>