A feasibility study on deep learning‐based individualized 3D dose distribution prediction

Abstract

Purpose Radiation therapy treatment planning is a trial‐and‐error, often time‐consuming process. An approximately optimal dose distribution corresponding to a specific patient's anatomy can be predicted by using pre‐trained deep learning (DL) models. However, dose distributions are often optimized based not only on patient‐specific anatomy but also on physicians’ preferred trade‐offs between planning target volume (PTV) coverage and organ at risk (OAR) sparing or among different OARs. Therefore, it is desirable to allow physicians to fine‐tune the dose distribution predicted based on patient anatomy. In this work, we developed a DL model to predict the individualized 3D dose distributions by using not only the patient's anatomy but also the desired PTV/OAR trade‐offs, as represented by a dose volume histogram (DVH), as inputs. Methods In this work, we developed a modified U‐Net network to predict the 3D dose distribution by using patient PTV/OAR masks and the desired DVH as inputs. The desired DVH, fine‐tuned by physicians from the initially predicted DVH, is first projected onto the Pareto surface, then converted into a vector, and then concatenated with feature maps encoded from the PTV/OAR masks. The network output for training is the dose distribution corresponding to the Pareto optimal DVH. The training/validation datasets contain 77 prostate cancer patients, and the testing dataset has 20 patients. Results The trained model can predict a 3D dose distribution that is approximately Pareto optimal while having the DVH closest to the input desired DVH. We calculated the difference between the predicted dose distribution and the optimized dose distribution that has a DVH closest to the desired one for the PTV and for all OARs as a quantitative evaluation. The largest absolute error in mean dose was about 3.6% of the prescription dose, and the largest absolute error in the maximum dose was about 2.0% of the prescription dose. Conclusions In this feasibility study, we have developed a 3D U‐Net model with the patient's anatomy and the desired DVH curves as inputs to predict an individualized 3D dose distribution that is approximately Pareto optimal while having the DVH closest to the desired one. The predicted dose distributions can be used as references for dosimetrists and physicians to rapidly develop a clinically acceptable treatment plan.

Dose Prediction Using a Three-Dimensional Convolutional Neural Network for Nasopharyngeal Carcinoma With Tomotherapy

“…Precise automatic dose prediction can significantly improve clinical planning efficiency and safety ( 23 ). 3D dose prediction results can refer to current RT plan optimization in TPS ( 24 , 25 ).…”

Application and progress of artificial intelligence in radiation therapy dose prediction

“…Since Nguyen pioneered the U-Net variant, more and more studies have focused on the DL developed based on 3D U-Net. By changing the internal structure of the model, it can be combined with different ML methods, such as DenseNet [41] , HD U-Net, DVHnet, and ResDevNet.Or by using various input data, such as distance information, PTV and OARs contour information, etc., with corresponding development in the fields of head and neck cancer [1] , [42] , [43] , [44] , [45] , [46] , [47] , [48] , breast cancer [49] , [50] , lung cancer [51] , [52] , prostate cancer [36] , [44] , [53] , [54] , [55] , [56] , [57] , [58] , [59] , and cervical cancer [60] , respectively.…”

“… Xing et al [36] /2020 Prostate cancer:78 Training set & validation set:70 Testing set:8 A modified Hierarchically Densely Connected U-net (HD U-net)model Pre-calculated inaccurate dose distribution and patient CT Dose distribution The dose difference between DL and CS < 0.25 Gy;for volume:<0.16 %. Ma et al [43] /2021 Prostate cancer:97 Training/Validation set:77 Testing set:20 3D U-net Patient PTV/OAR masks and the desired DVH. Dose distribution The largest average error: mean dose 1.6 %,maximum dose 1.8 % Ma et al [53] /2019 Prostate cancer:70 Training set:60 Test set:10 CNN The contours of six structures in CT images Dose distribution Mean SARs:0.029 ± 0.020(bladder),0.077 ± 0.030(rectum).…”

The role of artificial intelligence in veterinary radiation oncology

“…Some systems that automate contouring, planning, and plan evaluation of less complex treatment plans, such as human breast, prostate, and GI cancers, include the Automated Treatment Planning Assistant developed by MD Anderson, and the automated planning system developed at the Princess Margaret Hospital 36–38 . Deep learning has also been applied in generating optimized treatment plans and predict dose distributions based on a given set of inputs such as gantry angles, weights, and multi‐leaf collimator (MLC) settings 39–41 …”

“…[36][37][38] Deep learning has also been applied in generating optimized treatment plans and predict dose distributions based on a given set of inputs such as gantry angles, weights, and multi-leaf collimator (MLC) settings. [39][40][41] Experience with MCO and knowledge-based planning have yet to be reported in veterinary radiation oncology, but these technologies are currently available in commercial treatment planning systems. As the number of cases normally used to train such systems in the human domain are typically on the order of a 100, developing, training, and validating such systems in the veterinary domain becomes challenging.…”