A comprehensive evaluation of advanced dose calculation algorithms for brain stereotactic radiosurgery

PURPOSE

Accurate dose calculation is important in both target and low dose normal tissue regions for brain stereotactic radiosurgery (SRS). In this study, we aim to evaluate the dosimetric accuracy of the two advanced dose calculation algorithms for brain SRS.

METHODS

Retrospective clinical case study and phantom study were performed. For the clinical study, 138 SRS patient plans (443 targets) were generated using BrainLab Elements Voxel Monte Carlo (VMC). To evaluate the dose calculation accuracy, the plans were exported into Eclipse and recalculated with Acuros XB (AXB) algorithm with identical beam parameters. The calculated dose at the target center (Dref), dose to 95% target volume (D95), and the average dose to target (Dmean) were compared. Also, the distance from the skull was analyzed. For the phantom study, a cylindrical phantom and a head phantom were used, and the delivered dose was measured by an ion chamber and EBT3 film, respectively, at various locations. The measurement was compared with the calculated doses from VMC and AXB.

RESULTS

In clinical cases, VMC dose calculations tended to be higher than AXB. It was found that the difference in Dref showed > 5% in some cases for smaller volumes < 0.3 cm3 . Dmean and D95 differences were also higher for small targets. No obvious trend was found between the dose difference and the distance from the skull. In phantom studies, VMC dose was also higher than AXB for smaller targets, and VMC showed better agreement with the measurements than AXB for both point dose and high dose spread.

CONCLUSION

The two advanced calculation algorithms were extensively compared. For brain SRS, AXB sometimes calculates a noticeable lower target dose for small targets than VMC, and VMC tends to have a slightly closer agreement with measurements than AXB.

Dosimetry implications of VMAT delivery angles and intra-fraction motion for SBRT prostate treatments

For a cohort of retrospectively treated patients with prostate carcinoma using a Stereotactic body radiation therapy (SBRT) technique and with implanted localization fiducials, this study determined whether there were any correlations between theoretical fiducial visibility using intra-fraction megavoltage imaging and intra-fraction motion induced dosimetric effects. Treatment planning data for 20 retrospective patients that used a SBRT prostate technique were investigated in this study. An in-house script was created that split each of the 360-degree volumetric modulated arc therapy arcs into 12 sectors, with each sector being 30 degrees. The script created a total of 24 sectors for each SBRT plan ranging from 180 to 210 degrees to 180 to 150 degrees. Resulting data was assessed to determine whether there was dosimetric impact from intra-fractional prostate motion and if it correlated with the theoretical fiducial visibility. Forty percent of the 4 highest CTV D98% mean dose differences occurred within the angles of 240 and 270 degrees and 25% occurred between the angles of 90 and 120 degrees. The 4 highest average percentage differences in PTV D98% coverage during angular sectors of 270 to 240 degrees, 90 to 120 degrees, 240 to 270 degrees and 60 to 90 degrees were -1.19%, -1.14%, -1.10%, and 1.01% respectively. It was also the case for the PTV D95%in sectors 90 to 120 degrees, 240 to 270 degrees, 270 to 240 degrees and 270 to 300 degrees, which reduced by -0.97%, -0.93%, -0.92%, and -0.82% respectively. From the 4 highest rectal dose differences in V32Gy and V18Gy, it was found that 50% of the highest rectum V32Gyand V18Gy dose increases occurred between the angles of 90 and 120 degrees and 37.5% occurred between the angles of 240 and 270 degrees. The MU for each sector showed that 240_270, 240_210, 270_240, and 120_90 had the highest average MU with 150.8, 134.6, 129.2, and 124.3 respectively. This study demonstrated that there is a high correlation between dosimetric impact of intra-fractional motion and the theoretical fiducial visibility. As a result, modification to treatment plans to enable fiducial visibility at all angular sectors throughout a treatment may not be required. Further sector analysis tests are required to develop patient specific megavoltage imaging gantry angles for SBRT prostate patients.

Boosting radiotherapy dose calculation accuracy with deep learning

In radiotherapy, a trade‐off exists between computational workload/speed and dose calculation accuracy. Calculation methods like pencil‐beam convolution can be much faster than Monte‐Carlo methods, but less accurate. The dose difference, mostly caused by inhomogeneities and electronic disequilibrium, is highly correlated with the dose distribution and the underlying anatomical tissue density. We hypothesize that a conversion scheme can be established to boost low‐accuracy doses to high‐accuracy, using intensity information obtained from computed tomography (CT) images. A deep learning‐driven framework was developed to test the hypothesis by converting between two commercially available dose calculation methods: Anisotropic analytic algorithm (AAA) and Acuros XB (AXB). A hierarchically dense U‐Net model was developed to boost the accuracy of AAA dose toward the AXB level. The network contained multiple layers of varying feature sizes to learn their dose differences, in relationship to CT, both locally and globally. Anisotropic analytic algorithm and AXB doses were calculated in pairs for 120 lung radiotherapy plans covering various treatment techniques, beam energies, tumor locations, and dose levels. For each case, the CT and the AAA dose were used as the input and the AXB dose as the “ground‐truth” output, to train and test the model. The mean squared errors (MSEs) and gamma passing rates (2 mm/2% & 1 mm/1%) were calculated between the boosted AAA doses and the “ground‐truth” AXB doses. The boosted AAA doses demonstrated substantially improved match to the “ground‐truth” AXB doses, with average (± s.d.) gamma passing rate (1 mm/1%) 97.6% (±2.4%) compared to 87.8% (±9.0%) of the original AAA doses. The corresponding average MSE was 0.11(±0.05) vs 0.31(±0.21). Deep learning is able to capture the differences between dose calculation algorithms to boost the low‐accuracy algorithms. By combining a less accurate dose calculation algorithm with a trained deep learning model, dose calculation can potentially achieve both high accuracy and efficiency.

Optimal prescription isodose line in SBRT for lung tumor treatment with volumetric-modulated arc therapy

It is crucial to spare lung when treating early stage lung carcinoma with stereotactic body radiotherapy (SBRT) for minimizing the radiation induced toxicities, such as radiation pneumonitis and late fibrosis. A retrospective study was performed with a combination of approaches to determine the optimal range of prescription isodose line (P-IDL) within which lung tissue was best spared in SBRT plans with Volumetric-Modulated Arc Therapy (VMAT) and Monte-Carlo-like dose calculation algorithm. Twenty clinically-delivered SBRT lung plans were optimized using traditional LINAC MLC based approaches: an average P-IDL of (88.8 ± 0.5)% (the error bar of all the data is the 95% confidence interval (CI)). The plans were then re-optimized using a new combination of approaches with variation of P-IDL from 60% to 90% for each case. The combination of approaches included finding and utilizing an optimal P-IDL, implementing tuning ring structures internal and external to the target, as well as normal tissue objective and equivalent. The plans were evaluated with the following indexes: 1. R_50%, the ratio of 50% prescription isodose volume to the plan target volume (PTV); 2. V₂₀ and V₅, the volume of lung within 20Gy and 5Gy, respectively; 3. PCI, the Paddick comformity index; 4. D_2cm, the maximum dose at 2 cm from PTV in any direction; 5. MLD, the mean dose in total lung volume; 6. Focal Index (FI), an indicator of dose in the core of the target. The optimal P-IDL was found to be in the range of 75-80%. The average optimal P-IDL for the 20 cases was (77.9 ± 0.9)%. With the optimization strategies the average PCI was increased by (10.3 ± 2.1)%; the average R_50%, V₂₀, V₅, D_2cm and MLD were decreased by (29.1 ± 4.1)%, (26.9 ± 5.4)%, (13.9 ± 3.5)%, (13.4 ± 4.3)% and (16.7 ± 2.3)%, respectively. The FI was increased by (23.7 ± 1.3)%. The optimal P-IDL range was 75-80% for SBRT VMAT lung treatment plans. The application of the set of optimization approaches can significantly improve the lung sparing in SBRT VMAT plans with AXB dose calculation algorithm and makes treatment plans more conformal in high, intermediate and low dose regions, while higher dose is delivered to the target.

Dosimetric comparison of different algorithms in stereotactic body radiation therapy (SBRT) plan for non-small cell lung cancer (NSCLC)

Purposes

The main aim of the study was to investigate the dosimetric difference between acuros XB algorithm (AXB), anisotropic analytic algorithm (AAA), and pencil beam convolution (PBC) algorithm in stereotactic body radiation therapy (SBRT) plan for non-small cell lung cancer (NSCLC).

Patients and Methods

Thirty-eight NSCLC patients were included. GTV, PTV, and organs at risk were delineated by the radiation oncologists. Three optimized SBRT plans for each patients were gained using three algorithms of AXB, AAA, and PBC with the identical plan parameters. Dosimetric endpoints were collected and compared among the three plans, including dosimetric criteria: V100%, V90%, PTV D_min, D_max, D_mean, homogeneity index (HI), and Paddick conformity index (CI).

Results

AXB plan resulted in decreased V100% with a mean difference 6.14% compared with PBC plan (For V100%, AXB vs AAA vs PBC=93.44% vs 95.54% vs 99.58%, P<0.05). Three plans showed no significant difference as to the parameter V90%. AXB plan leaded to reduced D_min of PTV compared with other two algorithms (For D_min of PTV, AXB vs AAA vs PBC=4048cGy vs 4365Gy vs 4873Gy, P<0.05). PBC induced the enhanced trend of D_max of PTV compared with other two algorithms (D_max among three algorithms, P>0.05); and increased the D_mean of PTV in three algorithms with significant difference (For D_mean of PTV, AXB vs AAA vs PBC=5332cGy vs 5330Gy vs 5785Gy, P<0.05). AXB algorithm achieved a similar plan conformity with other two algorithms (For CI, AXB vs AAA vs PBC=0.80 vs 0.85 vs 0.71, P>0.05).

Conclusion

For SBRT plan of NSCLC, AAA and PBC algorithms overestimate target coverage, AXB algorithm is recommended for the SBRT plan of NSCLC.

A depth dose study between AAA and AXB algorithm against Monte Carlo simulation using AIP CT of a 4D dataset from a moving phantom

Aim

To identifying depth dose differences between the two versions of the algorithms using AIP CT of a 4D dataset.

Background

Motion due to respiration may challenge dose prediction of dose calculation algorithms during treatment planning.

Materials and methods

The two versions of depth dose calculation algorithms, namely, Anisotropic Analytical Algorithm (AAA) version 10.0 (AAAv10.0), AAA version 13.6 (AAAv13.6) and Acuros XB dose calculation (AXB) algorithm version 10.0 (AXBv10.0), AXB version 13.6 (AXBv13.6), were compared against a full MC simulated 6X photon beam using QUASAR respiratory motion phantom with a moving chest wall. To simulate the moving chest wall, a 4 cm thick wax mould was attached to the lung insert of the phantom. Depth doses along the central axis were compared in the anterior and lateral beam direction for field sizes 2 × 2 cm², 4 × 4 cm² and 10 × 10 cm².

Results

For the lateral beam direction, the moving chest wall highlighted differences of up to 105% for AAAv10.0 and 40% for AXBv10.0 from MC calculations in the surface and buildup doses. AAAv13.6 and AXBv13.6 agrees with MC predictions to within 10% at similar depth. For anterior beam doses, dose differences predicted for both versions of AAA and AXB algorithm were within 7% and results were consistent with static heterogeneous studies.

Conclusions

The presence of the moving chest wall was capable of identifying depth dose differences between the two versions of the algorithms. These differences could not be identified in the static chest wall as shown in the anterior beam depth dose calculations.

Dosimetric comparison of peripheral NSCLC SBRT using Acuros XB and AAA calculation algorithms

There is a concern for dose calculation in highly heterogenous environments such as the thorax region. This study compares the quality of treatment plans of peripheral non-small cell lung cancer (NSCLC) stereotactic body radiation therapy (SBRT) using 2 calculation algorithms, namely, Eclipse Anisotropic Analytical Algorithm (AAA) and Acuros External Beam (AXB), for 3-dimensional conformal radiation therapy (3DCRT) and volumetric-modulated arc therapy (VMAT). Four-dimensional computed tomography (4DCT) data from 20 anonymized patients were studied using Varian Eclipse planning system, AXB, and AAA version 10.0.28. A 3DCRT plan and a VMAT plan were generated using AAA and AXB with constant plan parameters for each patient. The prescription and dose constraints were benchmarked against Radiation Therapy Oncology Group (RTOG) 0915 protocol. Planning parameters of the plan were compared statistically using Mann-Whitney U tests. Results showed that 3DCRT and VMAT plans have a lower target coverage up to 8% when calculated using AXB as compared with AAA. The conformity index (CI) for AXB plans was 4.7% lower than AAA plans, but was closer to unity, which indicated better target conformity. AXB produced plans with global maximum doses which were, on average, 2% hotter than AAA plans. Both 3DCRT and VMAT plans were able to achieve D95%. VMAT plans were shown to be more conformal (CI = 1.01) and were at least 3.2% and 1.5% lower in terms of PTV maximum and mean dose, respectively. There was no statistically significant difference for doses received by organs at risk (OARs) regardless of calculation algorithms and treatment techniques. In general, the difference in tissue modeling for AXB and AAA algorithm is responsible for the dose distribution between the AXB and the AAA algorithms. The AXB VMAT plans could be used to benefit patients receiving peripheral NSCLC SBRT.

A comparison of two dose calculation algorithms-anisotropic analytical algorithm and Acuros XB-for radiation therapy planning of canine intranasal tumors

Although anisotropic analytical algorithm (AAA) and Acuros XB (AXB) are both radiation dose calculation algorithms that take into account the heterogeneity within the radiation field, Acuros XB is inherently more accurate. The purpose of this retrospective method comparison study was to compare them and evaluate the dose discrepancy within the planning target volume (PTV). Radiation therapy (RT) plans of 11 dogs with intranasal tumors treated by radiation therapy at the University of Georgia were evaluated. All dogs were planned for intensity-modulated radiation therapy using nine coplanar X-ray beams that were equally spaced, then dose calculated with anisotropic analytical algorithm. The same plan with the same monitor units was then recalculated using Acuros XB for comparisons. Each dog's planning target volume was separated into air, bone, and tissue and evaluated. The mean dose to the planning target volume estimated by Acuros XB was 1.3% lower. It was 1.4% higher for air, 3.7% lower for bone, and 0.9% lower for tissue. The volume of planning target volume covered by the prescribed dose decreased by 21% when Acuros XB was used due to increased dose heterogeneity within the planning target volume. Anisotropic analytical algorithm relatively underestimates the dose heterogeneity and relatively overestimates the dose to the bone and tissue within the planning target volume for the radiation therapy planning of canine intranasal tumors. This can be clinically significant especially if the tumor cells are present within the bone, because it may result in relative underdosing of the tumor.