On the quantification of the dosimetric accuracy of collapsed cone convolution superposition (CCCS) algorithm for small lung volumes using IMRT

GATE Monte Carlo approach to heterogeneity dose distribution in small fields used in radiation therapy

The Accurate dosage prediction in Radiation Therapy is challenging, prompting a need for precision beyond conventional clinical Treatment Planning Systems (TPS). Monte Carlo-based methods are sought for their superior accuracy. The aim of this study is to compare dose distributions between the ACUROS algorithm and the GATE platform in various tissue densities and field sizes, focusing on smaller fields. This study was initiated with a homogeneous validation of the TrueBeam STX system, using measurements obtained from the Centre Hospitalier Interregional Edith Cavell (CHIREC) in Brussels. The validation compared dosimetric functions (Percentage Depth Dose (PDD), Dose profile (DP) and Collimator scatter fraction (CSF)) employing the GAMMA index with a 2% / 2 mm criterion tolerance. Following this, heterogeneous studies examined dose distributions between the ACUROS algorithm and the GATE platform in various tissue densities and field sizes, with a specific focus on smaller fields. Simulations were conducted using both platforms on chest phantoms with heterogeneous slabs representing bone, lung, and heart, each housing a central tumor. The impact of electronic equilibrium on tumors for different small field sizes was evaluated. Results showed a remarkable 99% agreement between measurements and GATE calculations in the homogeneous validation of the TrueBeam STX system. However, in heterogeneous studies, ACUROS consistently overestimated lung doses by up to 8% compared to GATE simulation, especially evident with a flattening filter and smaller beam sizes at density interfaces. This highlights significant dose estimation discrepancies between ACUROS and GATE, emphasizing the need for precise calculations. The findings support exploring Monte Carlo-based methods for enhanced accuracy in Radiation Therapy treatment planning.

Effective clinical applications of Monte Carlo-based independent secondary dose verification software for helical tomotherapy

PURPOSE

To investigate the scope of the effective clinical application of Monte Carlo (MC)-based independent dose verification software for helical tomotherapy.

METHODS

DoseCHECK was selected as the MC-based dose calculation software. First, the dose calculation accuracy of DoseCHECK was evaluated with film and chamber measurements in a water-equivalent phantom. Second, the dose calculation accuracy was examined in several heterogeneous materials. Finally, dosimetric comparisons between DoseCHECK and the treatment planning system (TPS) were performed for clinical patient plans. Prostate IMRT, head and neck IMRT (HN), total body irradiation (TBI), and brain stereotactic radiotherapy (SRT) were evaluated.

RESULT

The DoseCHECK calculations agreed with the chamber and film measurements in the homogenous phantom. For heterogeneous phantom cases, the dose differences between DoseCHECK and TPS were within 3 %, except in air, in which large dose differences of 20 % were observed. In clinical patient plans, the median dose differences between the lung D_mean in TBI cases and the normal brain D_mean in brain SRT cases were significantly >3 %. For HN and brain SRT cases, the median target dose differences were >3 %.

CONCLUSION

Our results show that independent dose verification with the MC algorithm can detect systematic errors caused by the lack of heterogeneity correction in the TPS. In particular, MC-based independent dose verification is required for HN, TBI, and brain SRT cases in helical tomotherapy.

Monte Carlo modeling of the Elekta Versa HD and patient dose calculation with EGSnrc/BEAMnrc

Introduction

Numerous studies have proven the Monte Carlo method to be an accurate means of dose calculation. Although there are several commercial Monte Carlo treatment planning systems (TPSs), some clinics may not have access to these resources. We present a method for routine, independent patient dose calculations from treatment plans generated in a commercial TPS with our own Monte Carlo model using free, open‐source software.

Materials and methods

A model of the Elekta Versa HD linear accelerator was developed using the EGSnrc codes. A MATLAB script was created to take clinical patient plans and convert the DICOM RTP files into a format usable by EGSnrc. Ten patients’ treatment plans were exported from the Monaco TPS to be recalculated using EGSnrc. Treatment simulations were done in BEAMnrc, and doses were calculated using Source 21 in DOSXYZnrc. Results were compared to patient plans calculated in the Monaco TPS and evaluated in Verisoft with a gamma criterion of 3%/2 mm.

Results

Our Monte Carlo model was validated within 1%/1‐mm accuracy of measured percent depth doses and profiles. Gamma passing rates ranged from 82.1% to 99.8%, with 7 out of 10 plans having a gamma pass rate over 95%. Lung and prostate patients showed the best agreement with doses calculated in Monaco. All statistical uncertainties in DOSXYZnrc were less than 3.0%.

Conclusion

A Monte Carlo model for routine patient dose calculation was successfully developed and tested. This model allows users to directly recalculate DICOM RP files containing patients’ plans that have been exported from a commercial TPS.

Dose calculation accuracy for photon small fields in treatment planning systems with comparison by Monte Carlo simulations

Purpose: Advanced radiation therapy techniques use small fields in treatment planning and delivery. Small fields have the advantage of more accurate dose delivery, but with the cost of some complications in dosimetry. Different dose calculation algorithms imported in various treatment planning systems (TPSs) which each of them has different accuracy. Monte Carlo (MC) simulation has been reported as one of the accurate methods for calculating dose distribution in radiation therapy. The aim of this study was the evaluation of TPS dose calculation algorithms in small fields against 2 MC codes.

Methods: A linac head was simulated in 2 MC codes, MCNPX, and GATE. Then three small fields (0.5×0.5, 1×1 and 1.5×1.5 cm2) were simulated with 2 MC codes, and also these fields were planned with different dose calculation algorithms in Isogray and Monaco TPS. PDDs and lateral dose profiles were extracted and compared between MC simulations and dose calculation algorithms.

Results: For 0.5×0.5 cm2 field mean differences in PDDs with MCNPX were 2.28, 4.6, 5.3, and 7.4% and with GATE were -0.29, 2.3, 3 and 5% for CCC, superposition, FFT and Clarkson algorithms respectively. For 1×1 cm2 field mean differences in PDDs with MCNPX were 1.58, 0.6, 1.1 and 1.4% and with GATE were 0.77, 0.1, 0.6 and 0.9% for CCC, superposition, FFT and Clarkson algorithms respectively. For 1.5×1.5 cm2 field mean differences in PDDs with MCNPX were 0.82, 0.4, 0.6 and -0.4% and with GATE were 2.38, 2.5, 2.7 and 1.7% for CCC, superposition, FFT and Clarkson algorithms respectively.

Conclusions: Different dose calculation algorithms were evaluated and compared with MC simulation in small fields. Mean differences with MC simulation decreased with the increase of field sizes for all algorithms.

Predicting voxel-level dose distributions for esophageal radiotherapy using densely connected network with dilated convolutions

This work aims to develop a voxel-level dose prediction framework by integrating distance information between PTV and OARs, as well as image information, into a densely-connected network (DCNN). Firstly, a four-channel feature map, consisting of a PTV image, an OAR image, a CT image, and a distance image, is constructed. A densely connected neural network is then built and trained for voxel-level dose prediction. Considering that the shape and size of OARs are highly inconsistent, a dilated convolution is employed to capture features from multiple scales. Finally, the proposed network is evaluated with five-fold cross-validation, based on ninety-eight clinically approved treatment plans. The voxel-level mean absolute error(MAE _V ) of DCNN was 2.1% for PTV, 4.6% for left lung, 4.0% for right lung, 5.1% for heart, 6.0% for spinal cord, and 3.4% for body, which outperforms conventional U-Net, Resnet-antiResnet, U-Resnet-D by 0.1-0.8%. This result shows that with the introduction of a distance image and DCNN model, the accuracy of predicted dose distribution could be significantly improved. This approach offers a new dose prediction tool to support quality assurance and the automation of treatment planning in esophageal radiotherapy.

Incorporating biological modeling into patient‐specific plan verification

Purpose

Dose–volume histogram (DVH) measurements have been integrated into commercially available quality assurance systems to provide a metric for evaluating accuracy of delivery in addition to gamma analysis. We hypothesize that tumor control probability and normal tissue complication probability calculations can provide additional insight beyond conventional dose delivery verification methods.

Methods

A commercial quality assurance system was used to generate DVHs of treatment plan using the planning CT images and patient‐specific QA measurements on a phantom. Biological modeling was performed on the DVHs produced by both the treatment planning system and the quality assurance system.

Results

The complication‐free tumor control probability, P₊, has been calculated for previously treated intensity modulated radiotherapy (IMRT) patients with diseases in the following sites: brain (−3.9% ± 5.8%), head‐neck (+4.8% ± 8.5%), lung (+7.8% ± 1.3%), pelvis (+7.1% ± 12.1%), and prostate (+0.5% ± 3.6%).

Conclusion

Dose measurements on a phantom can be used for pretreatment estimation of tumor control and normal tissue complication probabilities. Results in this study show how biological modeling can be used to provide additional insight about accuracy of delivery during pretreatment verification.

Dose point kernels for 2,174 radionuclides

PURPOSE

Rapid adoption of targeted radionuclide therapy as an oncologic intervention has motivated the development of patient-specific voxel-wise approaches to radiation dosimetry. These approaches often rely on pretabulated dose point kernels for convolution-based calculations; however, these dose kernels are sparse in literature and often have suboptimal characteristics. The purpose of this work was to generate an extensive library of dose point kernels with sufficient size and resolution for general clinical application of voxel-wise dosimetry.

METHODS

Nuclear data were acquired for 2174 radionuclides from the National Nuclear Data Center (Brookhaven National Laboratory, accessed March 2018). Based on these data, isotropic point sources of radioactivity in water were simulated using Monte Carlo N-Particle transport v6.2 (MCNP6.2, Los Alamos National Laboratory). Simulations were separated by emission type for each radionuclide - photons (γ-rays, x rays), beta particles (positrons, electrons); and discrete electrons (conversion electrons, Auger electrons, Coster-Kronig electrons). Dose was tallied in concentric spherical shells about the point source using an energy deposition pulse-height tally (MCNP *F8 tally). Bins were spaced every 0.1 mm until a radius of 10 cm, and every 1 mm until a radius of 2 m. Positron emissions where treated as electrons for transport, with annihilation photons generated at the origin within the photon simulation. Alpha particle emissions were not simulated since their energy is deposited within ~0.2 mm of the source. Neutron and spallation effects were not considered. A subset of the resultant dose point kernels (¹¹ C, ¹⁸ F, ³² P, ^52g Mn, ⁶⁴ Cu, ⁶⁷ Ga, ⁸⁹ Sr, ⁸⁹ Zr, ⁹⁰ Y, ^99m Tc, ¹¹¹ In, ^117m Sn, ¹²³ I, ¹²⁴ I, ¹²⁵ I, ¹³¹ I, ¹⁵³ Sm, ¹⁷⁷ Lu, ¹⁸⁶ Re, ¹⁸⁸ Re, ²¹¹ As, ²¹² Pb, ²¹³ Bi, ²²³ Ra, and ²²⁵ Ac) were evaluated for accuracy based on conservation of energy, comparison to kernels in the literature, and statistical precision.

RESULTS

Among dose point kernels that were manually reviewed, good agreement with previously published dose point kernels was observed. Energy within the kernels was found to be conserved to within 1% of the value expected from nuclear data, suggesting that a radius of 2 m was sufficient to capture the almost all of the energy released during decay for all isotopes considered. Local dosimetric uncertainty, evaluated at the radius of 99% energy deposition, was found to be less than 9% for all radioisotopes evaluated. Rebinning data more coarsely by a factor of 10, similar to what would be done for a clinical dose calculation, results in all evaluated kernels having a relative error of less than 1.1% at R_50% , 1.5% at R_90% , and 2.7% at R_99% (the radius corresponding to 50%, 90%, and 99% of total energy deposition, respectively). The kernels produced in this work have been made freely available (https://zenodo.org/record/2564036).

CONCLUSIONS

An extensive library of high-resolution radial dose kernels was generated and validated against published data. In addition to enabling patient-specific voxel-wise internal dosimetry by convolution superposition, the generated dose point kernels data may prove useful to the wider health physics community.

Characterisation of small photon field outputs in a heterogeneous medium using X-ray voxel Monte Carlo dose calculation algorithm

Aim

To characterise small photon beams using the Monte Carlo dose calculation algorithm for small field ranges in a heterogeneous medium.

Materials and method

An in-house phantom constructed with three different mediums, foam, polymethyl methacrylate and delrin resembling the densities of lung, soft tissue and bone respectively, was used in this study. Photon beam energies of 6 and 15 MV and field sizes of 8×8, 16×16, 24×24, 32×32 and 40×40 mm using X-ray voxel Monte Carlo (XVMC) algorithm using different detectors were validated. The relative output factor was measured in three different mediums having six different tissue interfaces; at the depth of 0, 1, 2 and 3 cm. The planar dose verification was undertaken using gafchromic films and considered dose at the lung and bone medium interfaces. For all the measurements, 104×104 mm was taken as the reference field size. The relative output factor for all other field sizes was taken and compared with planning system calculated values.

Results

From field size 16×16 mm and above, the relative output factors were analysed in bone and soft tissue medium having lung as first medium. The maximum deviations were observed as 1·8 and 1·3% for 6 MV and 2·5 and 1·1% for 15 MV photon beams for bone and soft tissue, respectively. For lung as measurement medium, the maximum deviation of 14·8 and 19·2% were observed and having bone as first medium with 8×8 mm for 6 and 15 MV photon beams, respectively. The fluence verification of dose spectrum for the lung–bone interface scenarios with smaller field sizes were found within 2% of deviation with treatment planning system (TPS).

Conclusion

The accuracy of dose calculations for small field sizes in XVMC-based treatment planning algorithm was studied in different inhomogeneous mediums. It was found that the results correlated with measurement data for field size 16×16 mm and above. Noticeable deviation was observed for the smallest field size of 8×8 mm with interfaces of significant change in density. The observed results demands further analysis of work with smaller field sizes.

Comprehensive dosimetric planning comparison for early-stage, non-small cell lung cancer with SABR: fixed-beam IMRT versus VMAT versus TomoTherapy

Volumetric-modulated arc therapy (VMAT) is emerging as a leading technology in treating early-stage, non-small cell lung cancer (NSCLC) with stereotactic ablative radiotherapy (SABR). However, two other modalities capable of deliver-ing intensity-modulated radiation therapy (IMRT) include fixed-beam and helical TomoTherapy (HT). This study aims to provide an extensive dosimetric compari-son among these various IMRT techniques for treating early-stage NSCLC with SABR. Ten early-stage NSCLC patients were retrospectively optimized using three fixed-beam techniques via nine to eleven beams (high and low modulation step-and-shoot (SS), and sliding window (SW)), two VMAT techniques via two partial arcs (SmartArc (SA) and RapidArc (RA)), and three HT techniques via three different fan beam widths (1 cm, 2.5 cm, and 5 cm) for 80 plans total. Fixed-beam and VMAT plans were generated using flattening filter-free beams. SS and SA, HT treatment plans, and SW and RA were optimized using Pinnacle v9.1, Tomoplan v.3.1.1, and Eclipse (Acuros XB v11.3 algorithm), respectively. Dose-volume histogram statistics, dose conformality, and treatment delivery efficiency were analyzed. VMAT treatment plans achieved significantly lower values for contralat-eral lung V5Gy (p ≤ 0.05) compared to the HT plans, and significantly lower mean lung dose (p < 0.006) compared to HT 5 cm treatment plans. In the comparison between the VMAT techniques, a significant reduction in the total monitor units (p = 0.05) was found in the SA plans, while a significant decrease was observed in the dose falloff parameter, D2cm, (p = 0.05), for the RA treatments. The maximum cord dose was significantly reduced (p = 0.017) in grouped RA&SA plans com-pared to SS. Estimated treatment time was significantly higher for HT and fixed-beam plans compared to RA&SA (p < 0.001). Although, a significant difference was not observed in the RA vs. SA (p = 0.393). RA&SA outperformed HT in all parameters measured. Despite an increase in dose to the heart and bronchus, this study demonstrates that VMAT is dosimetrically advantageous in treating early-stage NSCLC with SABR compared to fixed-beam, while providing significantly shorter treatment times.

Dosimetric impact of orthopedic metal artifact reduction (O-MAR) on Spine SBRT patients

The dosimetric impact of orthopedic metal artifact reduction (O‐MAR) on spine SBRT patients has not been comprehensively studied, particularly with spinal prostheses in high‐dose gradient regions. Using both phantom and patient datasets, we investigated dosimetric effects of O‐MAR in combination of various metal locations and dose calculation algorithms. A physical phantom, with and without a titanium insert, was scanned. A clinical patient plan was applied to the artifact‐free reference, non‐O‐MAR, and O‐MAR phantom images with the titanium located either inside or outside of the tumor. Subsequently, five clinical patient plans were calculated with pencil beam and Monte Carlo (iPlan) on non‐O‐MAR and O‐MAR patient images using an extended CT‐density table. The dose differences for phantom plans and patient plans were analyzed using dose distributions, dose‐volume histograms (DVHs), gamma index, and selected dosimetric endpoints. From both phantom plans and patient plans, O‐MAR did not affect dose distributions and DVHs while minimizing metal artifacts. Among patient plans, we found that, when the same dose calculation method was used, the difference in the dosimetric endpoints between non‐O‐MAR and O‐MAR datasets were small. In conclusion, for spine SBRT patients with spinal prostheses, O‐MAR image reconstruction does not affect dose calculation accuracy while minimizing metal artifacts. Therefore, O‐MAR images can be safely used for clinical spine SBRT treatment planning.

PACS numbers: 87.53.Bn, 87.55.K‐, 87.57.Q‐, 87.57.cp

Comparison of CCC and ETAR dose calculation algorithms in pituitary adenoma radiation treatment planning; Monte Carlo evaluation

Aims

To verify the accuracy of two common absorbed dose calculation algorithms in comparison to Monte Carlo (MC) simulation for the planning of the pituitary adenoma radiation treatment.

Materials and methods

After validation of Linac's head modelling by MC in water phantom, it was verified in Rando phantom as a heterogeneous medium for pituitary gland irradiation. Then, equivalent tissue-air ratio (ETAR) and collapsed cone convolution (CCC) algorithms were compared for a conventional three small non-coplanar field technique. This technique uses 30 degree physical wedge and 18 MV photon beams.

Results

Dose distribution findings showed significant difference between ETAR and CCC of delivered dose in pituitary irradiation. The differences between MC and dose calculation algorithms were 6.40 ± 3.44% for CCC and 10.36 ± 4.37% for ETAR. None of the algorithms could predict actual dose in air cavity areas in comparison to the MC method.

Conclusions

Difference between calculation and true dose value affects radiation treatment outcome and normal tissue complication probability. It is of prime concern to select appropriate treatment planning system according to our clinical situation. It is further emphasised that MC can be the method of choice for clinical dose calculation algorithms verification.

Assessing the role of volumetric-modulated arc therapy in hepatocellular carcinoma

The role of volumetric-modulated arc therapy (VMAT) in hepatocellular carcinoma (HCC) remains controversial. The purpose of this study was to determine the potential clinical role of VMAT compared with three-dimensional conformal radiotherapy (3D CRT) for liver irradiation. Four-dimensional CT scans of 24patients with unresectable HCC were included and divided into two groups: (1) adjacent group (n = 11), with planning target volumes overlapping or within 1 cm adjacent to the alimentary tract; (2) nonadjacent group (n = 13), in which the normal liver itself was the dose-limiting structure. Target coverage, organs-at-risk (OARs) doses, delivery parameters, and treatment accuracy were evaluated. Superior target coverage, conformity, and homogeneity were achieved with VMAT compared with 3D CRT. In the adjacent group, VMAT provided superior sparing of serial functioning OARs including the stomach, small intestine, and spinal cord. In the nonadjacent group, VMAT provided inferior sparing of most OARs including the liver, stomach, and small intestine. For the whole group, the effective treatment time was 2.1 ± 0.3 min for 3D CRT and 3.1 ± 0.2 min for VMAT. For liver lesions adjacent to the alimentary tract, this study indicates that VMAT should be selected due to the plan quality, delivery efficiency, and superior sparing of stomach and small intestine. However, for liver lesions away from the alimentary tract, VMAT is not superior to 3D CRT for normal tissue protection.

Optimal partial-arcs in VMAT treatment planning

We present a method for improving the delivery efficiency of VMAT by extending the recently published VMAT treatment planning algorithm vmerge to automatically generate optimal partial-arc plans. A high-quality initial plan is created by solving a convex multicriteria optimization problem using 180 equi-spaced beams. This initial plan is used to form a set of dose constraints, and a set of partial-arc plans is created by searching the space of all possible partial-arc plans that satisfy these constraints. For each partial-arc, an iterative fluence map merging and sequencing algorithm (vmerge) is used to improve the delivery efficiency. Merging continues as long as the dose quality is maintained above a user-defined threshold. The final plan is selected as the partial-arc with the lowest treatment time. The complete algorithm is called pmerge. Partial-arc plans are created using pmerge for a lung, liver and prostate case, with final treatment times of 127, 245 and 147 . Treatment times using full arcs with vmerge are 211, 357 and 178 s. The mean doses to the critical structures for the vmerge and pmerge plans are kept within 5% of those in the initial plan, and the target volume covered by the prescription isodose is maintained above 98% for the pmerge and vmerge plans. Additionally, we find that the angular distribution of fluence in the initial plans is predictive of the start and end angles of the optimal partial-arc. We conclude that VMAT delivery efficiency can be improved by employing partial-arcs without compromising dose quality, and that partial-arcs are most applicable to cases with non-centralized targets.