Feasibility of automated planning for whole‐brain radiation therapy using deep learning

Abstract

Purpose: The purpose of this study was to develop automated planning for wholebrain radiation therapy (WBRT) using a U-net-based deep-learning model for predicting the multileaf collimator (MLC) shape bypassing the contouring processes. Methods: A dataset of 55 cases, including 40 training sets, five validation sets, and 10 test sets, was used to predict the static MLC shape. The digitally reconstructed radiograph (DRR) reconstructed from planning CT images as an input layer and the MLC shape as an output layer are connected one-to-one via the U-net modeling. The Dice similarity coefficient (DSC) was used as the loss function in the training and ninefold cross-validation. Dose-volume-histogram (DVH) curves were constructed for assessing the automatic MLC shaping performance. Deep-learning (DL) and manually optimized (MO) approaches were compared based on the DVH curves and dose distributions.Results: The ninefold cross-validation ensemble test results were consistent with DSC values of 94.6 ± 0.4 and 94.7 ± 0.9 in training and validation learnings, respectively. The dose coverages of 95% target volume were (98.0 ± 0.7)% and (98.3 ± 0.8)%, and the maximum doses for the lens as critical organ-at-risk were 2.9 Gy and 3.9 Gy for DL and MO, respectively. The DL technique shows the consistent results in terms of the DVH parameter except for MLC shaping prediction for dose saving of small organs such as lens.Conclusions: Comparable with the MO plan result, the WBRT plan quality obtained using the DL approach is clinically acceptable. Moreover, the DL approach enables WBRT auto-planning without the time-consuming manual MLC shaping and target contouring.whole-brain radiation therapy, deep learning, automatic planning | INTRODUCTIONMetastatic brain cancer is the most common type of intracranial tumor. 1 From common primary cancer sites such as lung, breast, and melanoma, brain metastasis occurs in 15% to 40% of the cancer patients. 2,3 The treatment of metastatic brain tumors depends on the number of metastatic tumors, extracranial tumor status, and performance status. [4][5][6] Currently, whole-brain radiation therapy (WBRT) is considered a well-established treatment for patients with multiple brain metastases.

Assessment of efficacy in automated plan generation for Varian Ethos intelligent optimization engine

“…Although KBP further facilitates the treatment planning process while maintaining clinical quality, it is limited to the number and quality of treatment plans in the database, which stem from one institution. Several groups have also investigated the feasibility of artificial intelligence (AI)‐based automated planning 21–26 . Currently, one TPS Vendor (RaySearch Laboratories, Stockholm, Sweden) provides AI‐based automated treatment planning 27 .…”

“…Several groups have also investigated the feasibility of artificial intelligence (AI)‐based automated planning. 21 , 22 , 23 , 24 , 25 , 26 Currently, one TPS Vendor (RaySearch Laboratories, Stockholm, Sweden) provides AI‐based automated treatment planning. 27 AI‐based auto‐segmentation is available for either a reference CT or online daily cone‐beam computed tomography (CBCT) scans to create an adaptive treatment plan.…”

A Survey on Deep Learning for Precision Oncology

“…Yu et al [ 4 ] employed a U-Net to predict the multileaf collimator (MLC) shape in the task for automatic treatment planning for whole-brain radiotherapy (WBRT). The input of their model is the digitally reconstructed radiograph (DRR) from CT images and the output is the MLC shape.…”

“…Han et al [ 16 ] employed a DeepLab-V3+ for automated treatment planning for whole-brain radiotherapy (WBRT) using CT images. Yu et al [ 4 ] employed a U-Net for automated treatment planning for WBRT using CT images to predict the multileaf collimator (MLC) shape bypassing the contouring processes. They constructed the dose-volume histogram (DVH) curves to assess the automatic MLC shaping performance.…”

“…Wang et al [ 3 ] demonstrated that deep learning could be used for automated radiotherapy planning and has gained enormous attention to improve the quality and efficiency of treatment planning. Applications of deep learning in precision oncology can be coarsely divided into dose distribution for treatment planning, survival analysis and risk estimation after treatment, prediction of treatment response, and patient selection for treatment planning with the type of treatment, such as radiotherapy [ 4 ], chemotherapy [ 5 ], immunotherapy [ 6 ], chemoradiation [ 7 ], targeted therapy [ 8 ], surgery [ 9 ], radiosurgery [ 10 ], and multiple therapy [ 11 ]. One particular review of the deep-learning application to radiotherapy planning was published by Wang et al [ 3 ].…”