Automatic treatment planning based on three-dimensional dose distribution predicted from deep learning technique.

To develop an automated treatment planning strategy for external beam intensity-modulated radiation therapy (IMRT), including a deep learning-based three-dimensional (3D) dose prediction and a dose distribution-based plan generation algorithm. A residual neural network-based deep learning model is trained to predict a dose distribution based on patient-specific geometry and prescription dose. A total of 270 head-and-neck cancer cases were enrolled in this study, including 195 cases in the training dataset, 25 cases in the validation dataset, and 50 cases in the testing dataset. All patients were treated with IMRT with a variety of different prescription patterns. The model input consists of CT images and contours delineating the organs at risk (OARs) and planning target volumes (PTVs). The algorithm output is trained to predict the dose distribution on the CT image slices. The obtained prediction model is used to predict dose distributions for new patients. Then, an optimization objective function based on these predicted dose distributions is created for automatic plan generation. Our results demonstrate that the deep learning method can predict clinically acceptable dose distributions. There is no statistically significant difference between prediction and real clinical plan for all clinically relevant dose-volume histogram (DVH) indices, except brainstem, right and left lens. However, the predicted plans were still clinically acceptable. The results of plan generation show no statistically significant differences between the automatic generated plan and the predicted plan except PTV70.4 , but the difference is only 0.5% which is still clinically acceptable. This study developed a new automated radiotherapy treatment planning system based on 3D dose prediction and 3D dose distribution-based optimization. It is a promising approach for realizing automated treatment planning in the future.

Fan J ，Wang J ，Chen Z ，Hu C ，Zhang Z ，Hu W ... -  《-》

A hybrid optimization strategy for deliverable intensity-modulated radiotherapy plan generation using deep learning-based dose prediction.

To propose a clinically feasible automatic planning solution for external beam intensity-modulated radiotherapy, including dose prediction via a deep learning and voxel-based optimization strategy. The dose distribution of patients was predicted using a U-Net-based deep learning network based on the patient's anatomy information. One hundred seventeen patients with nasopharyngeal cancer (NPC) and 200 patients with rectal cancer were enrolled in this study. For NPC cases, 94 cases were included in the training dataset, 13 in the validation dataset, and 10 in the testing dataset. For rectal cancer cases, 172 cases were included in the training set, 18 in the validation set, and 10 in the testing set. A voxel-based optimization strategy, "Voxel," was proposed to achieve treatment planning optimization by dividing body voxels into two parts: inside planning target volumes (PTVs) and outside PTVs. Fixed dose-volume objectives were attached to the total objective function to realize individualized planning intended as the "hybrid" optimizing strategy. Automatically generated plans were compared with clinically approved plans to evaluate clinical gains, according to dosimetric indices and dose-volume histograms (DVHs). Similarities were found between the DVH of the predicted dose and clinical plan, although significant differences were found in some organs at risk. Better organ sparing and suboptimal PTV coverage were shown using the voxel strategy; however, the deviations in homogeneity indices (HIs) and conformity indices (CIs) of the PTV between automatically generated plans and manual plans were reduced by the hybrid strategy ([manual plans]/[voxel plans[/[hybrid plans]: HI of PTV70 [1.06/1.12/1.02] and CI of PTV70 [0.79/0.58/0.76]). The optimization time for each patient was within 1 min and included fluence map optimization, leaf sequencing, and control point optimization. All the generated plans (voxel and hybrid strategy) could be delivered on uRT-linac 506c (United Imaging Healthcare, Shanghai, China). Deliverable plans can be generated by incorporating a voxel-based optimization strategy into a commercial treatment planning system (TPS). The hybrid optimization method shows the benefit and clinical feasibility in generating clinically acceptable plans.

Sun Z ，Xia X ，Fan J ，Zhao J ，Zhang K ，Wang J ，Hu W ... -  《-》

A deep learning method for prediction of three-dimensional dose distribution of helical tomotherapy.

To develop a deep learning method for prediction of three-dimensional (3D) voxel-by-voxel dose distributions of helical tomotherapy (HT). Using previously treated HT plans as training data, a deep learning model named U-ResNet-D was trained to predict a 3D dose distribution. First, the contoured structures and dose volumes were converted from plan database to 3D matrix with a program based on a developed visualization toolkit (VTK), then transferred to U-ResNet-D for correlating anatomical features and dose distributions at voxel-level. One hundred and ninety nasopharyngeal cancer (NPC) patients treated by HT with multiple planning target volumes (PTVs) in different prescription patterns were studied. The model was typically trained from scratch with weights randomly initialized rather than using transfer-learning method, and used to predict new patient's 3D dose distributions. The predictive accuracy was evaluated with three methods: (a) The dose difference at the position r, δ(r, r) = Dc (r) - Dp (r), was calculated for each voxel. The mean (μδ(r,r) ) and standard deviation (σδ(r,r) ) of δ(r, r) were calculated to assess the prediction bias and precision; (b) The mean absolute differences of dosimetric indexes (DIs) including maximum and mean dose, homogeneity index, conformity index, and dose spillage for PTVs and organ at risks (OARs) were calculated and statistically analyzed with the paired-samples t test; (c) Dice similarity coefficients (DSC) between predicted and clinical isodose volumes were calculated. The U-ResNet-D model predicted 3D dose distribution accurately. For twenty tested patients, the prediction bias ranged from -2.0% to 2.3% and prediction error varied from 1.5% to 4.5% (relative to prescription) for 3D dose differences. The mean absolute dose differences for PTVs and OARs are within 2.0% and 4.2%, and nearly all the DIs for PTVs and OARs had no significant differences. The averaged DSC ranged from 0.95 to 1 for different isodose volumes. The study developed a new deep learning method for 3D voxel-by-voxel dose prediction, and shown to be able to produce accurately dose predictions for nasopharyngeal patients treated by HT. The predicted 3D dose map can be useful for improving radiotherapy planning design, ensuring plan quality and consistency, making clinical technique comparison, and guiding automatic treatment planning.

Liu Z ，Fan J ，Li M ，Yan H ，Hu Z ，Huang P ，Tian Y ，Miao J ，Dai J ... -  《-》

Attention-aware 3D U-Net convolutional neural network for knowledge-based planning 3D dose distribution prediction of head-and-neck cancer.

Deep learning-based knowledge-based planning (KBP) methods have been introduced for radiotherapy dose distribution prediction to reduce the planning time and maintain consistent high-quality plans. This paper presents a novel KBP model using an attention-gating mechanism and a three-dimensional (3D) U-Net for intensity-modulated radiation therapy (IMRT) 3D dose distribution prediction in head-and-neck cancer. A total of 340 head-and-neck cancer plans, representing the OpenKBP-2020 AAPM Grand Challenge data set, were used in this study. All patients were treated with the IMRT technique and a dose prescription of 70 Gy. The data set was randomly divided into 64%/16%/20% as training/validation/testing cohorts. An attention-gated 3D U-Net architecture model was developed to predict full 3D dose distribution. The developed model was trained using the mean-squared error loss function, Adam optimization algorithm, a learning rate of 0.001, 120 epochs, and batch size of 4. In addition, a baseline U-Net model was also similarly trained for comparison. The model performance was evaluated on the testing data set by comparing the generated dose distributions against the ground-truth dose distributions using dose statistics and clinical dosimetric indices. Its performance was also compared to the baseline model and the reported results of other deep learning-based dose prediction models. The proposed attention-gated 3D U-Net model showed high capability in accurately predicting 3D dose distributions that closely replicated the ground-truth dose distributions of 68 plans in the test set. The average value of the mean absolute dose error was 2.972 ± 1.220 Gy (vs. 2.920 ± 1.476 Gy for a baseline U-Net) in the brainstem, 4.243 ± 1.791 Gy (vs. 4.530 ± 2.295 Gy for a baseline U-Net) in the left parotid, 4.622 ± 1.975 Gy (vs. 4.223 ± 1.816 Gy for a baseline U-Net) in the right parotid, 3.346 ± 1.198 Gy (vs. 2.958 ± 0.888 Gy for a baseline U-Net) in the spinal cord, 6.582 ± 3.748 Gy (vs. 5.114 ± 2.098 Gy for a baseline U-Net) in the esophagus, 4.756 ± 1.560 Gy (vs. 4.992 ± 2.030 Gy for a baseline U-Net) in the mandible, 4.501 ± 1.784 Gy (vs. 4.925 ± 2.347 Gy for a baseline U-Net) in the larynx, 2.494 ± 0.953 Gy (vs. 2.648 ± 1.247 Gy for a baseline U-Net) in the PTV_70, and 2.432 ± 2.272 Gy (vs. 2.811 ± 2.896 Gy for a baseline U-Net) in the body contour. The average difference in predicting the D99 value for the targets (PTV_70, PTV_63, and PTV_56) was 2.50 ± 1.77 Gy. For the organs at risk, the average difference in predicting the D m a x ${D_{max}}$ (brainstem, spinal cord, and mandible) and D m e a n ${D_{mean}}$ (left parotid, right parotid, esophagus, and larynx) values was 1.43 ± 1.01 and 2.44 ± 1.73 Gy, respectively. The average value of the homogeneity index was 7.99 ± 1.45 for the predicted plans versus 5.74 ± 2.95 for the ground-truth plans, whereas the average value of the conformity index was 0.63 ± 0.17 for the predicted plans versus 0.89 ± 0.19 for the ground-truth plans. The proposed model needs less than 5 s to predict a full 3D dose distribution of 64 × 64 × 64 voxels for a new patient that is sufficient for real-time applications. The attention-gated 3D U-Net model demonstrated a capability in predicting accurate 3D dose distributions for head-and-neck IMRT plans with consistent quality. The prediction performance of the proposed model was overall superior to a baseline standard U-Net model, and it was also competitive to the performance of the best state-of-the-art dose prediction method reported in the literature. The proposed model could be used to obtain dose distributions for decision-making before planning, quality assurance of planning, and guiding-automated planning for improved plan consistency, quality, and planning efficiency.

Osman AFI ，Tamam NM 《Journal of Applied Clinical Medical Physics》

Automatic dose prediction using deep learning and plan optimization with finite-element control for intensity modulated radiation therapy.

Automatic solutions for generating radiotherapy treatment plans using deep learning (DL) have been investigated by mimicking the voxel's dose. However, plan optimization using voxel-dose features has not been extensively studied. This study aims to investigate the efficiency of a direct optimization strategy with finite elements (FEs) after DL dose prediction for automatic intensity-modulated radiation therapy (IMRT) treatment planning. A double-UNet DL model was adapted for 220 cervical cancer patients (200 for training and 20 for testing), who underwent IMRT between 2016 and 2020 at our clinic. The model inputs were computed tomography (CT) slices, organs at risk (OARs), and planning target volumes (PTVs), and the outputs were dose distributions of uniformly generated high-dose region-controlled plans. The FEs were discretized into equal intervals of the dose prediction value within the [OARs avoid PTV(O-P)] and [body avoids OARs & PTV(B-OP)] regions in the test cohort and used to define the objectives for IMRT plan optimization. The plans were optimized using a two-step process. In the beginning, the plans of two extra cases with and without low-dose region control were compared to pursue robust and optimal dose adjustment degree pattern of FEs. In the first step, the mean dose of O-P FEs were constrained to differing degrees according to the pattern. The further the FEs from the PTV, the tighter the constraints. In the second step, the mean dose of O-P FEs from first step were constrained again but weakly and the dose of the B-OP FEs from dose prediction and PTV were tightly regulated. The dosimetric parameters of the OARs and PTV were evaluated and compared using an interstep approach. In another 10 cases, the plans optimized via the aforementioned steps (method 1) were compared with those directly generated by the double-UNet dose prediction model trained by low and high region-controlled plans (method 2). The mean differences in dose metrics between the UNet-predicted dose and the clinical plans were: 0.47 Gy for bladder D50% ; 0.62 Gy for rectum D50% ; 0% for small intestine V30Gy ; 1% for small intestine V40Gy ; 4% for left femoral head V30Gy ; and 6% for right femoral head V30Gy . The reductions in mean dose (p < 0.001) after FE-based optimization were: 4.0, 1.9, 2.8, 5.9, and 5.7 Gy for the bladder, rectum, small intestine, left femoral head, and right femoral head, respectively, with flat PTV homogeneity and conformity. Method 1 plans produced lower mean doses than those of method 2 for the bladder (0.7 Gy), rectum (1.0 Gy), and small intestine (0.6 Gy), while maintaining  PTV homogeneity and conformity. FE-based direct optimization produced lower OAR doses and adequate PTV doses after DL prediction. This solution offers rapid and automatic plan optimization without manual adjustment, particularly in low-dose regions.

Shen Y ，Tang X ，Lin S ，Jin X ，Ding J ，Shao M ... -  《-》