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

Abstract

Purpose To develop a deep learning method for prediction of three‐dimensional (3D) voxel‐by‐voxel dose distributions of helical tomotherapy (HT). Methods 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. Results 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. Conclusions 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.

A method of using deep learning to predict three‐dimensional dose distributions for intensity‐modulated radiotherapy of rectal cancer

“…There have been some researches on 3D dose distribution prediction using deep learning 43‐47. Nguyen et al employed Hierarchically Densely Connected U‐Net model to implement dose prediction for IMRT treatment plan of the prostate, and the results showed that the averaged D max and D mean of dose differences for all contoured structures were within 5.1% of the prescription dose and the average DSC between the predicted and clinical truth was 0.91 43.…”

“…Nguyen et al employed Hierarchically Densely Connected U‐Net model to implement dose prediction for IMRT treatment plan of the prostate, and the results showed that the averaged D max and D mean of dose differences for all contoured structures were within 5.1% of the prescription dose and the average DSC between the predicted and clinical truth was 0.91 43. Liu et al used 2D residual network to achieve dose prediction for helical tomotherapy of nasopharyngeal cases, which reported the mean absolute differences of D max and D mean for OARs were within 4.2% and 2.4%, respectively, and averaged 3D dose prediction bias ranged from 2.0% to 2.3% 44. It is difficult to directly compare the CNN models developed by us with those by these researchers, since different patient databases and treatment modality were used.…”

Deep learning‐based virtual noncontrast CT for volumetric modulated arc therapy planning: Comparison with a dual‐energy CT‐based approach

“…[19][20][21][22] A DL-based algorithm with convolutional neural networks can automatically extract complex features by backpropagation to output appropriate results, eliminating the need for manual feature extraction. 23,24 We hypothesized that a DL-based method could generate realistic VNC images that suppress the contrast material signal while maintaining the CT numbers of bones. Here, we describe a fully automated method that uses DL to generate more realistic VNC images (VNC DL ) than VNC DECT images to improve the accuracy of treatment planning.…”

A Review on Application of Deep Learning Algorithms in External Beam Radiotherapy Automated Treatment Planning

“…In recent years, a number of deep learning (DL)-based ATP techniques have been proposed using various DL neural networks (18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33). Several review articles on AI in radiation oncology (34)(35)(36), and radiotherapy treatment planning (37)(38)(39), have been published, which demonstrated the interests on AI and the significance of ATP, summarization of the achievements and challenges, as well as insightful discussion on future studies.…”