Aleatoric and epistemic uncertainty extraction of patient-specific deep learning-based dose predictions in LDR prostate brachytherapy.

Objective.In brachytherapy, deep learning (DL) algorithms have shown the capability of predicting 3D dose volumes. The reliability and accuracy of such methodologies remain under scrutiny for prospective clinical applications. This study aims to establish fast DL-based predictive dose algorithms for low-dose rate (LDR) prostate brachytherapy and to evaluate their uncertainty and stability.Approach.Data from 200 prostate patients, treated with125I sources, was collected. The Monte Carlo (MC) ground truth dose volumes were calculated with TOPAS considering the interseed effects and an organ-based material assignment. Two 3D convolutional neural networks, UNet and ResUNet TSE, were trained using the patient geometry and the seed positions as the input data. The dataset was randomly split into training (150), validation (25) and test (25) sets. The aleatoric (associated with the input data) and epistemic (associated with the model) uncertainties of the DL models were assessed.Main results.For the full test set, with respect to the MC reference, the predicted prostateD90metric had mean differences of -0.64% and 0.08% for the UNet and ResUNet TSE models, respectively. In voxel-by-voxel comparisons, the average global dose difference ratio in the [-1%, 1%] range included 91.0% and 93.0% of voxels for the UNet and the ResUNet TSE, respectively. One forward pass or prediction took 4 ms for a 3D dose volume of 2.56 M voxels (128 × 160 × 128). The ResUNet TSE model closely encoded the well-known physics of the problem as seen in a set of uncertainty maps. The ResUNet TSE rectum D2cchad the largest uncertainty metric of 0.0042.Significance.The proposed DL models serve as rapid dose predictors that consider the patient anatomy and interseed attenuation effects. The derived uncertainty is interpretable, highlighting areas where DL models may struggle to provide accurate estimations. The uncertainty analysis offers a comprehensive evaluation tool for dose predictor model assessment.

FastDM,Mcalculation in LDR brachytherapy using deep learning methods.

Objective.The Monte Carlo (MC) method provides a complete solution to the tissue heterogeneity effects in low-energy low-dose rate (LDR) brachytherapy. However, long computation times limit the clinical implementation of MC-based treatment planning solutions. This work aims to apply deep learning (DL) methods, specifically a model trained with MC simulations, to predict accurate dose to medium in medium (DM,M) distributions in LDR prostate brachytherapy.Approach.To train the DL model, 2369 single-seed configurations, corresponding to 44 prostate patient plans, were used. These patients underwent LDR brachytherapy treatments in which125I SelectSeed sources were implanted. For each seed configuration, the patient geometry, the MC dose volume and the single-seed plan volume were used to train a 3D Unet convolutional neural network. Previous knowledge was included in the network as anr2kernel related to the first-order dose dependency in brachytherapy. MC and DL dose distributions were compared through the dose maps, isodose lines, and dose-volume histograms. Features enclosed in the model were visualized.Main results.Model features started from the symmetrical kernel and finalized with an anisotropic representation that considered the patient organs and their interfaces, the source position, and the low- and high-dose regions. For a full prostate patient, small differences were seen below the 20% isodose line. When comparing DL-based and MC-based calculations, the predicted CTVD90metric had an average difference of -0.1%. Average differences for OARs were -1.3%, 0.07%, and 4.9% for the rectumD2cc, the bladderD2cc, and the urethraD0.1cc. The model took 1.8 ms to predict a complete 3DDM,Mvolume (1.18 M voxels).Significance.The proposed DL model stands for a simple and fast engine which includes prior physics knowledge of the problem. Such an engine considers the anisotropy of a brachytherapy source and the patient tissue composition.

Validation of the TOPAS Monte Carlo toolkit for LDR brachytherapy simulations.

PURPOSE: This work has the purpose of validating the Monte Carlo toolkit TOol for PArticle Simulation (TOPAS) for low-dose-rate (LDR) brachytherapy uses. METHODS AND MATERIALS: Simulations of 12 LDR sources and 2 COMS eye plaques (10 mm and 20 mm in diameter) and comparisons with published reference data from the Carleton Laboratory for Radiotherapy Physics (CLRP), the TG-43 consensus data and the TG-129 consensus data were performed. Sources from the IROC Houston Source Registry were modeled. The OncoSeed 6711 and the SelectSeed 130.002 were also modeled for historical reasons. For each source, the dose rate constant, the radial dose function and the anisotropy functions at 0.5, 1 and 5 cm were extracted. For the eye plaques (loaded with 125I sources), dose distribution maps, dose profiles along the central axis and transverse axis were calculated. RESULTS: Dose rate constants for 11 of the 12 sources are within 4% of the consensus data and within 2% of the CLRP data. The radial dose functions and anisotropy functions are mostly within 2% of the CLRP data. In average, 92% of all voxels are within 1% of the CLRP data for the eye plaques dose distributions. The dose profiles are within 0.5% (central axis) and 1% (transverse axis) of the reference data. CONCLUSION: The TOPAS MC toolkit was validated for LDR brachytherapy applications. Single-seed and multi-seed results agree with the published reference data. TOPAS has several benefits such as a simplified approach to MC simulations and an accessible brachytherapy package including comprehensive learning resources.

Benchmark of the PenRed Monte Carlo framework for HDR brachytherapy.

PURPOSE: The purpose of this study is to validate the PenRed Monte Carlo framework for clinical applications in brachytherapy. PenRed is a C++ version of Penelope Monte Carlo code with additional tallies and utilities. METHODS AND MATERIALS: Six benchmarking scenarios are explored to validate the use of PenRed and its improved bachytherapy-oriented capabilities for HDR brachytherapy. A new tally allowing the evaluation of collisional kerma for any material using the track length kerma estimator and the possibility to obtain the seed positions, weights and directions processing directly the DICOM file are now implemented in the PenRed distribution. The four non-clinical test cases developed by the Joint AAPM-ESTRO-ABG-ABS WG-DCAB were evaluated by comparing local and global absorbed dose differences with respect to established reference datasets. A prostate and a palliative lung cases, were also studied. For them, absorbed dose ratios, global absorbed dose differences, and cumulative dose-volume histograms were obtained and discussed. RESULTS: The air-kerma strength and the dose rate constant corresponding to the two sources agree with the reference datatests within 0.3% (Sk) and 0.1% (Λ). With respect to the first three WG-DCAB test cases, more than 99.8% of the voxels present local (global) differences within ±1%(±0.1%) of the reference datasets. For test Case 4 reference dataset, more than 94.9%(97.5%) of voxels show an agreement within ±1%(±0.1%), better than similar benchmarking calculations in the literature. The track length kerma estimator scorer implemented increases the numerical efficiency of brachytherapy calculations two orders of magnitude, while the specific brachytherapy source allows the user to avoid the use of error-prone intermediate steps to translate the DICOM information into the simulation. In both clinical cases, only minor absorbed dose differences arise in the low-dose isodoses. 99.8% and 100% of the voxels have a global absorbed dose difference ratio within ±0.2% for the prostate and lung cases, respectively. The role played by the different segmentation and composition material in the bone structures was discussed, obtaining negligible absorbed dose differences. Dose-volume histograms were in agreement with the reference data. CONCLUSIONS: PenRed incorporates new tallies and utilities and has been validated for its use for detailed and precise high-dose-rate brachytherapy simulations.

Validation of the TOPAS Monte Carlo toolkit for HDR brachytherapy simulations.

PURPOSE: The goal of this work is to validate the user-friendly Geant4-based Monte Carlo toolkit TOol for PArticle Simulation (TOPAS) for brachytherapy applications. METHODS AND MATERIALS: Brachytherapy simulations performed with TOPAS were systematically compared with published TG-186 reference data. The photon emission energy spectrum, the air-kerma strength, and the dose-rate constant of the model-based dose calculation algorithm (MBDCA)-WG generic Ir-192 source were extracted. For dose calculations, a track-length estimator was implemented. The four Joint AAPM/ESTRO/ABG MBDCA-WG test cases were evaluated through histograms of the local and global dose difference volumes. A prostate, a palliative lung, and a breast case were simulated. For each case, the dose ratio map, the histogram of the global dose difference volume, and cumulative dose-volume histograms were calculated. RESULTS: The air-kerma strength was (9.772 ± 0.001) × 10-8 U Bq-1 (within 0.3% of the reference value). The dose-rate constant was 1.1107 ± 0.0005 cGy h-1 U-1 (within 0.01% of the reference value). For all cases, at least 96.9% of voxels had a local dose difference within [-1%, 1%] and at least 99.9% of voxels had a global dose difference within [-0.1%, 0.1%]. The implemented track-length estimator scorer was more efficient than the default analog dose scorer by a factor of 237. For all clinical cases, at least 97.5% of voxels had a global dose difference within [-1%, 1%]. Dose-volume histograms were consistent with the reference data. CONCLUSIONS: TOPAS was validated for high-dose-rate brachytherapy simulations following the TG-186 recommended approach for MBDCAs. Built on top of Geant4, TOPAS provides broad access to a state-of-the-art Monte Carlo code for brachytherapy simulations.