Development of clinical application program for radiotherapy induced cancer risk calculation using Monte Carlo engine in volumetric-modulated arc therapy

A generalization performance study on the boosting radiotherapy dose calculation engine based on super-resolution

PURPOSE

During the radiation treatment planning process, one of the time-consuming procedures is the final high-resolution dose calculation, which obstacles the wide application of the emerging online adaptive radiotherapy techniques (OLART). There is an urgent desire for highly accurate and efficient dose calculation methods. This study aims to develop a dose super resolution-based deep learning model for fast and accurate dose prediction in clinical practice.

METHOD

A Multi-stage Dose Super-Resolution Network (MDSR Net) architecture with sparse masks module and multi-stage progressive dose distribution restoration method were developed to predict high-resolution dose distribution using low-resolution data. A total of 340 VMAT plans from different disease sites were used, among which 240 randomly selected nasopharyngeal, lung, and cervix cases were used for model training, and the remaining 60 cases from the same sites for model benchmark testing, and additional 40 cases from the unseen site (breast and rectum) was used for model generalizability evaluation. The clinical calculated dose with a grid size of 2 mm was used as baseline dose distribution. The input included the dose distribution with 4 mm grid size and CT images. The model performance was compared with HD U-Net and cubic interpolation methods using Dose-volume histograms (DVH) metrics and global gamma analysis with 1%/1 mm and 10% low dose threshold. The correlation between the prediction error and the dose, dose gradient, and CT values was also evaluated.

RESULTS

The prediction errors of MDSR were 0.06-0.84% of Dmean indices, and the gamma passing rate was 83.1-91.0% on the benchmark testing dataset, and 0.02-1.03% and 71.3-90.3% for the generalization dataset respectively. The model performance was significantly higher than the HD U-Net and interpolation methods (p < 0.05). The mean errors of the MDSR model decreased (monotonously by 0.03-0.004%) with dose and increased (by 0.01-0.73%) with the dose gradient. There was no correlation between prediction errors and the CT values.

CONCLUSION

The proposed MDSR model achieved good agreement with the baseline high-resolution dose distribution, with small prediction errors for DVH indices and high gamma passing rate for both seen and unseen sites, indicating a robust and generalizable dose prediction model. The model can provide fast and accurate high-resolution dose distribution for clinical dose calculation, particularly for the routine practice of OLART.

Assessment of dose perturbations for metal stent in photon and proton radiotherapy plans for hepatocellular carcinoma

BACKGROUND

The present study aimed to investigate the dosimetric impact of metal stent for photon and proton treatment plans in hepatocellular carcinoma.

METHODS

With computed tomography data of a water-equivalent solid phantom, dose perturbation caused by a metal stent included in the photon and proton treatment of hepatocellular carcinoma was evaluated by comparing Eclipse and RayStation treatment planning system (TPS) to a Monte Carlo (MC) based dose calculator. Photon and proton plans were created with anterior-posterior/posterior-anterior (AP/PA) fields using a 6 MV beam and AP/PA fields of a wobbling beam using 150 MeV and a 10 cm ridge filter. The difference in dose distributions and dosimetric parameters were compared depending on the stent's positions (the bile duct (GB) and intestinal tract (GI)) and angles (0°, 45°, and 90°). Additionally, the dose variation in the target volume including the stent was comparatively evaluated through dose volume histogram (DVH) analysis. And the comparison of clinical cases was carried out in the same way.

RESULTS

Percentage differences in the dosimetric parameters calculated by MC ranged from - 7.0 to 3.9% for the photon plan and - 33.7 to 4.3% for the proton plan, depending on the angle at which the GB and GI stents were placed, compared to those without the stent. The maximum difference was observed at the minimum dose (D_min), which was observed in both photon and proton plans in the GB and GI stents deployed at a 90° incidence angle. The parameter differences were greater in the proton plan than in photon plan. The target volume showed various dose variations depending on positions and angles of stent for both beams. Compared with no-stent, the doses within the target volume containing the GI and GB stents for the photon beam were overestimated in the high-dose area at 0°, nearly equal within 1% at 45°, and underestimated at 90°. These doses to the proton beam were underestimated at all angles, and the amount of underdose to the target volume increased with an increase in the stent angle. However, the difference was significantly greater with the proton plan than the photon plan.

CONCLUSIONS

Dose perturbations within the target volume due to the presence of the metal stent were not observed in the TPS calculations for photon and proton beams, but MC was used to confirm that there are dose variations within the target volume. The MC results found that delivery of the treatment beam avoiding the stent was the best method to prevent target volume underdose.

Experimental Validation of an Analytical Program and a Monte Carlo Simulation for the Computation of the Far Out-of-Field Dose in External Beam Photon Therapy Applied to Pediatric Patients

Background

The out-of-the-field absorbed dose affects the probability of primary second radiation-induced cancers. This is particularly relevant in the case of pediatric treatments. There are currently no methods employed in the clinical routine for the computation of dose distributions from stray radiation in radiotherapy. To overcome this limitation in the framework of conventional teletherapy with photon beams, two computational tools have been developed—one based on an analytical approach and another depending on a fast Monte Carlo algorithm. The purpose of this work is to evaluate the accuracy of these approaches by comparison with experimental data obtained from anthropomorphic phantom irradiations.

Materials and Methods

An anthropomorphic phantom representing a 5-year-old child (ATOM, CIRS) was irradiated considering a brain tumor using a Varian TrueBeam linac. Two treatments for the same planned target volume (PTV) were considered, namely, intensity-modulated radiotherapy (IMRT) and volumetric modulated arc therapy (VMAT). In all cases, the irradiation was conducted with a 6-MV energy beam using the flattening filter for a prescribed dose of 3.6 Gy to the PTV. The phantom had natLiF : Mg, Cu, P (MCP-N) thermoluminescent dosimeters (TLDs) in its 180 holes. The uncertainty of the experimental data was around 20%, which was mostly attributed to the MCP-N energy dependence. To calculate the out-of-field dose, an analytical algorithm was implemented to be run from a Varian Eclipse TPS. This algorithm considers that all anatomical structures are filled with water, with the exception of the lungs which are made of air. The fast Monte Carlo code dose planning method was also used for computing the out-of-field dose. It was executed from the dose verification system PRIMO using a phase-space file containing 3x109 histories, reaching an average standard statistical uncertainty of less than 0.2% (coverage factor k = 1 ) on all voxels scoring more than 50% of the maximum dose. The standard statistical uncertainty of out-of-field voxels in the Monte Carlo simulation did not exceed 5%. For the Monte Carlo simulation the actual chemical composition of the materials used in ATOM, as provided by the manufacturer, was employed.

Results

In the out-of-the-field region, the absorbed dose was on average four orders of magnitude lower than the dose at the PTV. For the two modalities employed, the discrepancy between the central values of the TLDs located in the out-of-the-field region and the corresponding positions in the analytic model were in general less than 40%. The discrepancy in the lung doses was more pronounced for IMRT. The same comparison between the experimental and the Monte Carlo data yielded differences which are, in general, smaller than 20%. It was observed that the VMAT irradiation produces the smallest out-of-the-field dose when compared to IMRT.

Conclusions

The proposed computational methods for the routine calculation of the out-of-the-field dose produce results that are similar, in most cases, with the experimental data. It has been experimentally found that the VMAT irradiation produces the smallest out-of-the-field dose when compared to IMRT for a given PTV.