Generalisation of radiotherapy dose calculation for Monte Carlo algorithm combined with 3D Swin-Unet: a multi-institutional IMRT evaluation.

Objective.Accurate dose calculations are essential prerequisites for precise radiotherapy. The integration of deep learning into dosimetry could consider computational accuracy and efficiency and has potential applicability to clinical dose calculation. The generalisation of a deep learning dose calculation method (hereinafter referred to as TERMA-Monte Carlo network, T-MC net) was evaluated in clinical practice using intensity-modulated radiotherapy (IMRT) plans for various human body regions and multiple institutions, with the Monte Carlo (MC) algorithm serving as a benchmark.Approach. Sixty IMRT plans were selected from four institutions for testing the head and neck, chest and abdomen, and pelvis regions. Using the MC results as the benchmark, the T-MC net calculation results were used to perform three-dimensional dose distribution and dose-volume histogram (DVH) comparisons of the entire body, planning target volume (PTV) and organs at risk (OARs), respectively, and calculate the mean ±95% confidence interval of gamma pass rate (GPR), percentage of agreement (PA) and dose difference ratio (DDR) of dose indices D95, D50, and D5.Main results. For the entire body, the GPRs of 3%/3 mm, 2%/2 mm, 2%/1 mm, and the PA were 99.62 ± 0.32%, 98.50 ± 1.09%, 95.60 ± 2.90% and 97.80 ± 1.12%, respectively. For the PTV, the GPRs of 3%/3 mm, 2%/2 mm, 2%/1 mm and the PA were 98.90 ± 1.00%, 95.78 ± 2.83%, 92.23 ± 4.74% and 98.93 ± 0.62%, respectively. The absolute value of average DDR was less than 1.4%.Significance. We proposed a general dose calculation framework based on deep learning, using the MC algorithm as a benchmark, performing a generalisation test for IMRT treatment plans across multiple institutions. The framework provides high computational speed while maintaining the accuracy of MC and may become an effective dose algorithm engine in treatment planning, adaptive radiotherapy, and dose verification.

DVH Prediction for VMAT in NPC with GRU-RNN: An Improved Method by Considering Biological Effects.

PURPOSE: A recurrent neural network (RNN) and its variants such as gated recurrent unit-based RNN (GRU-RNN) were found to be very suitable for dose-volume histogram (DVH) prediction in our previously published work. Using the dosimetric information generated by nonmodulated beams of different orientations, the GRU-RNN model was capable of accurate DVH prediction for nasopharyngeal carcinoma (NPC) treatment planning. On the basis of our previous work, we proposed an improved approach and aimed to further improve the DVH prediction accuracy as well as study the feasibility of applying the proposed method to relatively small-size patient data. METHODS: Eighty NPC volumetric modulated arc therapy (VMAT) plans with local IRB's approval in recent two years were retrospectively and randomly selected in this study. All these original plans were created using the Eclipse treatment planning system (V13.5, Varian Medical Systems, USA) with ≥95% of PGTVnx receiving the prescribed doses of 70 Gy, ≥95% of PGTVnd receiving 66 Gy, and ≥95% of PTV receiving 60 Gy. Among them, fifty plans were used to train the DVH prediction model, and the remaining were used for testing. On the basis of our previously published work, we simplified the 3-layer GRU-RNN model to a single-layer model and further trained every organ at risk (OAR) separately with an OAR-specific equivalent uniform dose- (EUD-) based loss function. RESULTS: The results of linear least squares regression obtained by the new proposed method showed the excellent agreements between the predictions and the original plans with the correlation coefficient r = 0.976 and 0.968 for EUD results and maximum dose results, respectively, and the coefficient r of our previously published method was 0.957 and 0.946, respectively. The Wilcoxon signed-rank test results between the proposed and the previous work showed that the proposed method could significantly improve the EUD prediction accuracy for the brainstem, spinal cord, and temporal lobes with a p value < 0.01. CONCLUSIONS: The accuracy of DVH prediction achieved in different OARs showed the great improvements compared to the previous works, and more importantly, the effectiveness and robustness showed by the simplified GRU-RNN trained from relatively small-size DVH samples, fully demonstrated the feasibility of applying the proposed method to small-size patient data. Excellent agreements in both EUD results and maximum dose results between the predictions and original plans indicated the application prospect in a physically and biologically related (or a mixture of both) model for treatment planning.

Dose calculation in proton therapy using a discovery cross-domain generative adversarial network (DiscoGAN).

PURPOSE: Accurate dose calculation is a critical step in proton therapy. A novel machine learning-based approach was proposed to achieve comparable accuracy to that of Monte Carlo simulation while reducing the computational time. METHODS: Computed tomography-based patient phantoms were used and three treatment sites were selected (thorax, head, and abdomen), comprising different beam pathways and beam energies. The training data were generated using Monte Carlo simulations. A discovery cross-domain generative adversarial network (DiscoGAN) was developed to perform the mapping between two domains: stopping power and dose, with HU values from CT images incorporated as auxiliary features. The accuracy of dose calculation was quantitatively evaluated in terms of mean relative error (MRE) and mean absolute error (MAE). The relationship between the DiscoGAN performance and other factors such as absolute dose, beam energy and location within the beam cross-section (center and off-center lines) was examined. RESULTS: The DiscoGAN model is found to be effective in dose calculation. For the abdominal case, the MRE is found to 1.47% (mean), 3.30% (maximum) and 0.67% (minimum). For the thoracic case, the MRE is found to ~2.43% (mean), 4.80% (maximum) and 0.71% (minimum). For the head case, the MRE is found to ~2.83% (mean), 4.84% (maximum) and 1.01% (minimum). Comparable accuracy is found in the independent validation dataset (different CT images), achieving a mean MRE of ~1.65% (thorax), 4.02% (head) and 1.64% (abdomen). For the energy span between 80 and 130 MeV, no strong dependency of accuracy on beam energy is found. The results imply that no systematic deviation, either over-dose or under-dose, occurs between the predicted dose and raw dose. CONCLUSION: The DiscoGAN framework demonstrates great potential as a tool for dose calculation in proton therapy, achieving comparable accuracy yet being more efficient relative to Monte Carlo simulation. Its comparison with the pencil beam algorithm (PBA) will be the next step of our research. If successful, our proposed approach is expected to find its use in more advanced applications such as inverse planning and adaptive proton therapy.

Application of dose-volume histogram prediction in biologically related models for nasopharyngeal carcinomas treatment planning.

PURPOSE: In this study, we employed a gated recurrent unit (GRU)-based recurrent neural network (RNN) using dosimetric information induced by individual beam to predict the dose-volume histogram (DVH) and investigated the feasibility and usefulness of this method in biologically related models for nasopharyngeal carcinomas (NPC) treatment planning. METHODS AND MATERIALS: One hundred patients with NPC undergoing volumetric modulated arc therapy (VMAT) between 2018 and 2019 were randomly selected for this study. All the VMAT plans were created using the Monaco treatment planning system (Elekta, Sweden) and clinically approved: > 98% of PGTVnx received the prescribed doses of 70 Gy, > 98% of PGTVnd received the prescribed doses of 66 Gy and > 98% of PCTV received 60 Gy. Of these, the data from 80 patients were used to train the GRU-RNN, and the data from the other 20 patients were used for testing. For each NPC patient, the DVHs of different organs at risk were predicted by a trained GRU-based RNN using the information given by individual conformal beams. Based on the predicted DVHs, the equivalent uniform doses (EUD) were calculated and applied as dose constraints during treatment planning optimization. The regenerated VMAT experimental plans (EPs) were evaluated by comparing them with the clinical plans (CPs). RESULTS: For the 20 test patients, the regenerated EPs guided by the GRU-RNN predictive model achieved good consistency relative to the CPs. The EPs showed better consistency in PTV dose distribution and better dose sparing for many organs at risk, and significant differences were found in the maximum/mean doses to the brainstem, brainstem PRV, spinal cord, lenses, temporal lobes, parotid glands and larynx with P-values < 0.05. On average, compared with the CPs, the maximum/mean doses to these OARs were altered by - 3.44 Gy, - 1.94 Gy, - 1.88 Gy, 0.44 Gy, 1.98 Gy, - 1.82 Gy and 2.27 Gy, respectively. In addition, significant differences were also found in brainstem and spinal cord for the dose received by 1 cc volume with 4.11 and 1.67 Gy dose reduction in EPs on average. CONCLUSION: The GRU-RNN-based DVH prediction method was capable of accurate DVH prediction. The regenerated plans guided by the predicted EUDs were not inferior to the manual plans, had better consistency in PTVs and better dose sparing in critical OARs, indicating the usefulness and effectiveness of biologically related model in knowledge-based planning.

Dose-volume histogram prediction in volumetric modulated arc therapy for nasopharyngeal carcinomas based on uniform-intensity radiation with equal angle intervals.

In this study, we developed a gated recurrent unit (GRU)-based recurrent neural network (RNN) for dose-volume histogram (DVH) prediction in volumetric modulated arc therapy (VMAT) planning for nasopharyngeal carcinomas (NPCs) based on uniform-intensity radiation with equal angle intervals and investigated the feasibility and usefulness of this method for treatment optimization. One hundred twenty-four NPC patients were selected from a database containing clinical VMAT plans from 2015 to 2018; of these, the data from 100 patients were used to train the GRU-RNN, and the data of the other 24 patients were used for testing. For the prescribed doses to D95 (the absorbed dose for 95% of the planning target volume) of all the plans in 30 or 31 fractions, 70 Gy were delivered to PTV70 (the gross tumour volume with circumferential margin), 60 Gy were delivered to PTV60, 54 Gy were delivered to PTV54 and 66 Gy were delivered to PTV66 (lymph node gross tumour volume with circumferential margin). For each NPC patient, an equal-field-weight conformal radiotherapy plan was generated by a treatment planning system (TPS) to offer uniform-intensity radiation. By adjusting the field weights, the dose distribution induced by individual conformal beams was acquired, and the corresponding DVH was calculated. Direction-dependent DVHs were employed to predict the DVH for VMAT with the GRU-RNN, and the regenerated VMAT experimental plans (EPs), guided by the predicted DVHs, were evaluated by comparing them with the clinical plans (CPs). For the 24 test patients, the regenerated EPs guided by the GRU-RNN predictive model achieved good consistency relative to the CPs. The EPs resulted in better dose sparing for many organs at risk (OARs) while still meeting the acceptable criteria for the PTVs. Significant differences were found in the maximum/mean doses to the optic nerves, temporal lobes, lenses, mandibles, temporomandibular joints (TMJs), larynx and inner ears, with P-values of 0.03, 0.01, 0.01, <0.01, 0.02, 0.02 and <0.01, respectively. On average, compared to the CPs, the maximum/mean doses to these OARs were altered by -1.38 Gy, -0.92 Gy, 0.53 Gy, -1.19 Gy, -1.16 Gy, 2.39 Gy and -1.71 Gy, respectively. The results showed the accuracy and effectiveness of the proposed uniform-intensity radiation approach. The regenerated plans guided by the predictive method were not inferior to the manual plans, indicating their great potential for improved planning and quality control in clinical applications.

A trial for EBT3 film without batch-specific calibration using a neural network.

This note reports a trial to establish an ANN (artificial neural network) method applying to EBT3 films of different batches without batch-specific calibration. Based on Pytorch (Facebook, https://pytorch.org/), a feed-forward ANN model was built to convert the pixel values of scanned images from different batches into absorbed dose. Films from different batches exposed to x-ray doses were digitized in transmission mode on an Epson 11000XL scanner for training and testing. The calculated dose map of TPS (radiation therapy planning system) was used as a label (the desired output) for the ANN model. To verify the performance and generalization of the ANN method, a cross-validation experiment was performed. Using the trained ANN method, the scanned images were converted into absorbed dose maps, and the converted dose maps have good agreement with the calculated dose maps from TPS. For films irradiated via the sliding window mode, the MSEs (mean square errors) of the trained batches were less than [Formula: see text] and the MSEs of the tested batches were less than [Formula: see text]. For patient intensity-modulated radiotherapy (IMRT) films, the γ(3%, 3 mm) between the dose maps obtained from the trained films and TPS exceeded 97.5%. The γ(3%, 3 mm) between most of the dose maps obtained from the tested films and TPS exceeded 97.0%. This shows that it is feasible to establish a method for EBT3 films from certain batches to convert pixel values into an absorbed dose without batch-specific calibration, and the method can be applied to other cases.