Purpose: To develop a method for predicting optimal dose distributions, given the planning image and segmented anatomy, by applying deep learning techniques to a database of previously optimized and approved Intensity‐modulated radiation therapy treatment plans.Methods: Eighty cases of early‐stage nasopharyngeal cancer (NPC) were included in the study. Seventy cases were chosen randomly as the training set and the remaining as the test set. The inputs were the images with structures, with each target and organs at risk (OARs) assigned a unique label. The outputs were dose maps, including coarse dose maps and converted fine dose maps (FDM) from convolution. Two types of input images with structures were used in the model building. One type of input included the images (with associated structures) without manipulation. The second type of input involved modifying the image gray label with information from radiation beam geometry. ResNet101 was chosen as the deep learning network for both. The accuracy of predicted dose distributions was evaluated against the corresponding dose as used in the clinic. A global three‐dimensional gamma analysis was calculated for the evaluation.Results: The proposed model trained with the two different sets of input images and structures could both predict patient‐specific dose distributions accurately. For the out‐of‐field dose distributions, the model obtained from the input with radiation geometry performed better (dose difference in %, 4.7 ± 6.1% vs 5.5 ± 7.9%, P < 0.05). The mean Gamma pass rates of dose distributions predicted with both types of input were comparable for most OARs (P > 0.05), except for the bilateral optic nerves and the optic chiasm.Conclusions: The proposed system with radiation geometry added to the input is a promising method to generate patient‐specific dose distributions for radiotherapy. It can be applied to obtain the dose distributions slice‐by‐slice for planning quality assurance and for guiding automated planning.

Radiation pneumonitis (RP) is one of the major toxicities of thoracic radiation therapy. RP incidence has been proven to be closely associated with the dosimetric factors and normal tissue control possibility (NTCP) factors. However, because these factors only utilize limited information of the dose distribution, the prediction abilities of these factors are modest. We adopted the dosiomics method for RP prediction. The dosiomics method first extracts spatial features of the dose distribution within ipsilateral, contralateral, and total lungs, and then uses these extracted features to construct prediction model via univariate and multivariate logistic regression (LR). The dosiomics method is validated using 70 non-small cell lung cancer (NSCLC) patients treated with volumetric modulated arc therapy (VMAT) radiotherapy. Dosimetric and NTCP factors based prediction models are also constructed to compare with the dosiomics features based prediction model. For the dosimetric, NTCP and dosiomics factors/features, the most significant single factors/features are the mean dose, parallel/serial (PS) NTCP and gray level co-occurrence matrix (GLCM) contrast of ipsilateral lung, respectively. And the area under curve (AUC) of univariate LR is 0.665, 0.710 and 0.709, respectively. The second significant factors are V₅ of contralateral lung, equivalent uniform dose (EUD) derived from PS NTCP of contralateral lung and the low gray level run emphasis of gray level run length matrix (GLRLM) of total lungs. The AUC of multivariate LR is improved to 0.676, 0.744, and 0.782, respectively. The results demonstrate that the univariate LR of dosiomics features has approximate predictive ability with NTCP factors, and the multivariate LR outperforms both the dosimetric and NTCP factors. In conclusion, the spatial features of dose distribution extracted by the dosiomics method effectively improves the prediction ability.

Purpose: More and more automatic segmentation tools are being introduced in routine clinical practice. However, physicians need to spend a considerable amount of time in examining the generated contours slice by slice. This greatly reduces the benefit of the tool's automaticity. In order to overcome this shortcoming, we developed an automatic quality assurance (QA) method for automatic segmentation using convolutional neural networks (CNNs). Materials and Methods: The study cohort comprised 680 patients with early-stage breast cancer who received whole breast radiation. The overall architecture of the automatic QA method for deep learning-based segmentation included the following two main parts: a segmentation CNN model and a QA network that was established based on ResNet-101. The inputs were from computed tomography, segmentation probability maps, and uncertainty maps. Two kinds of Dice similarity coefficient (DSC) outputs were tested. One predicted the DSC quality level of each slice ([0.95, 1] for “good,” [0.8, 0.95] for “medium,” and [0, 0.8] for “bad” quality), and the other predicted the DSC value of each slice directly. The performances of the method to predict the quality levels were evaluated with quantitative metrics: balanced accuracy, F score, and the area under the receiving operator characteristic curve (AUC). The mean absolute error (MAE) was used to evaluate the DSC value outputs. Results: The proposed methods involved two types of output, both of which achieved promising accuracy in terms of predicting the quality level. For the good, medium, and bad quality level prediction, the balanced accuracy was 0.97, 0.94, and 0.89, respectively; the F score was 0.98, 0.91, and 0.81, respectively; and the AUC was 0.96, 0.93, and 0.88, respectively. For the DSC value prediction, the MAE was 0.06 ± 0.19. The prediction time was approximately 2 s per patient. Conclusions: Our method could predict the segmentation quality automatically. It can provide useful information for physicians regarding further verification and revision of automatic contours. The integration of our method into current automatic segmentation pipelines can improve the efficiency of radiotherapy contouring.

Radiation pneumonitis (RP) is one of the major side effects of thoracic radiotherapy. The aim of this study is to build a dose distribution based prediction model, and investigate the correlation of RP incidence and high-order features of dose distribution. A convolution 3D (C3D) neural network was used to construct the prediction model. The C3D network was pre-trained for action recognition. The dose distribution was used as input of the prediction model. With the C3D network, the convolution operation was performed in 3D space. The guided gradient-weighted class activation map (grad-CAM) was utilized to locate the regions of dose distribution which were strongly correlated with grade≥2 and grade<2 RP cases, respectively. The features learned by the convolution filters were generated with gradient ascend to understand the deep network. The performance of the C3D prediction model was evaluated by comparing with three multivariate logistic regression (LR) prediction models, which used the dosimetric, normal tissue complication probability (NTCP) or dosiomics factors as input, respectively. All the prediction models were validated using 70 non-small cell lung cancer (NSCLC) patients treated with volumetric modulated arc therapy (VMAT). The area under curve (AUC) of C3D prediction model was 0.842. While the AUC of the three LR models were 0.676, 0.744 and 0.782, respectively. The guided grad-CAM indicated that the low-dose region of contralateral lung and high-dose region of ipsilateral lung were strongly correlated with the grade≥2 and grade<2 RP cases, respectively. The features learned by shallow filters were simple and globally consistent, and of monotonous color. The features of deeper filters displayed more complicated pattern, which was hard or impossible to give strict mathematical definition. In conclusion, we built a C3D model for thoracic radiotherapy toxicity prediction. The results demonstrate its performance is superior over the classical LR models. In addition, CNN also offers a new perspective to further understand RP incidence.

Purpose: The record of daily quality control (QC) items shows machine performance patterns and potentially provides warning messages for preventive actions. This study developed a neural network model that could predict the record and trend of data variations quantitively.Methods and materials: The record of 24 QC items for a radiotherapy machine was investigated in our institute. The QC records were collected daily for 3 years. The stacked long short‐term memory (LSTM) model was used to develop the neural network model. A total of 867 records were collected to predict the record for the next 5 days. To compare the stacked LSTM, the autoregressive integrated moving average model (ARIMA) was developed on the same data set. The accuracy of the model was quantified by the mean absolute error (MAE), root‐mean‐square error (RMSE), and coefficient of determination (R²). To validate the robustness of the model, the record of four QC items was collected for another radiotherapy machine, which was input into the stacked LSTM model without changing any hyperparameters and ARIMA model.Results: The mean MAE, RMSE, and with 24 QC items were 0.013, 0.020, and 0.853 in LSTM, while 0.021, 0.030, and 0.618 in ARIMA, respectively. The results showed that the stacked LSTM outperforms the ARIMA. Moreover, the mean MAE, RMSE, and with four QC items were 0.102, 0.151, and 0.770 in LSTM, while 0.162, 0.375, and 0.550 in ARIMA, respectively.Conclusions: In this study, the stacked LSTM model can accurately predict the record and trend of QC items. Moreover, the stacked LSTM model is robust when applied to another radiotherapy machine. Predicting future performance record will foresee possible machine failure, allowing early machine maintenance and reducing unscheduled machine downtime.