Automatic Inverse Treatment Planning for Gamma Knife Radiosurgery via Deep Reinforcement Learning

Purpose: Several inverse planning algorithms have been developed for Gamma Knife (GK) radiosurgery to determine a large number of plan parameters via solving an optimization problem, which typically consists of multiple objectives. The priorities among these objectives need to be repetitively adjusted to achieve a clinically good plan for each patient. This study aimed to achieve automatic and intelligent priority-tuning, by developing a deep reinforcement learning (DRL) based method to model the tuning behaviors of human planners. Methods: We built a priority-tuning policy network using deep convolutional neural networks. Its input was a vector composed of the plan metrics that were used in our institution for GK plan evaluation. The network can determine which tuning action to take, based on the observed quality of the intermediate plan. We trained the network using an end-to-end DRL framework to approximate the optimal action-value function. A scoring function was designed to measure the plan quality. Results: Vestibular schwannoma was chosen as the test bed in this study. The number of training, validation and testing cases were 5, 5, and 16, respectively. For these three datasets, the average plan scores with initial priorities were 3.63 $pm$ 1.34, 3.83 $pm$ 0.86 and 4.20 $pm$ 0.78, respectively, while can be improved to 5.28 $pm$ 0.23, 4.97 $pm$ 0.44 and 5.22 $pm$ 0.26 through manual priority tuning by human expert planners. Our network achieved competitive results with 5.42 $pm$ 0.11, 5.10 $pm$ 0. 42, 5.28 $pm$ 0.20, respectively. Conclusions: Our network can generate GK plans of comparable or slightly higher quality comparing with the plans generated by human planners via manual priority tuning. The network can potentially be incorporated into the clinical workflow to improve GK planning efficiency.

Generative Adversarial Network for Image Synthesis

This chapter reviews recent developments of generative adversarial networks (GAN)-based methods for medical and biomedical image synthesis tasks. These methods are classified into conditional GAN and Cycle-GAN according to the network architecture designs. For each category, a literature survey is given, which covers discussions of the network architecture designs, highlights important contributions and identifies specific challenges.

Synthetic MRI-aided Head-and-Neck Organs-at-Risk Auto-Delineation for CBCT-guided Adaptive Radiotherapy

Purpose: Organ-at-risk (OAR) delineation is a key step for cone-beam CT (CBCT) based adaptive radiotherapy planning that can be a time-consuming, labor-intensive, and subject-to-variability process. We aim to develop a fully automated approach aided by synthetic MRI for rapid and accurate CBCT multi-organ contouring in head-and-neck (HN) cancer patients. MRI has superb soft-tissue contrasts, while CBCT offers bony-structure contrasts. Using the complementary information provided by MRI and CBCT is expected to enable accurate multi-organ segmentation in HN cancer patients. In our proposed method, MR images are firstly synthesized using a pre-trained cycle-consistent generative adversarial network given CBCT. The features of CBCT and synthetic MRI are then extracted using dual pyramid networks for final delineation of organs. CBCT images and their corresponding manual contours were used as pairs to train and test the proposed model. Quantitative metrics including Dice similarity coefficient (DSC) were used to evaluate the proposed method. The proposed method was evaluated on a cohort of 65 HN cancer patients. CBCT images were collected from those patients who received proton therapy. Overall, DSC values of 0.87, 0.79/0.79, 0.89/0.89, 0.90, 0.75/0.77, 0.86, 0.66, 0.78/0.77, 0.96, 0.89/0.89, 0.832, and 0.84 for commonly used OARs for treatment planning including brain stem, left/right cochlea, left/right eye, larynx, left/right lens, mandible, optic chiasm, left/right optic nerve, oral cavity, left/right parotid, pharynx, and spinal cord, respectively, were achieved. In this study, we developed a synthetic MRI-aided HN CBCT auto-segmentation method based on deep learning. It provides a rapid and accurate OAR auto-delineation approach, which can be used for adaptive radiation therapy.

An institutional study on plan quality and variation of manual forward planning for Gamma Knife radiosurgery for vestibular schwannoma

Due to the complexity and cumbersomeness of Gamma Knife (GK) manual forward planning, the quality of the resulting treatment plans heavily depends on the planners skill, experience and the amount of effort devoted to plan development. Hence, GK plan quality may vary significantly among institutions and planners, and even for a same planner at different cases. This is particularly a concern for challenging cases with complicated geometry, such as vestibular schwannoma cases. The purpose of this retrospective study is to investigate the plan quality and variation in the manually forward planned, clinically acceptable GK treatment plans of 22 previous vestibular schwannoma cases. Considering the impacts of different patient geometry and different trade-offs among the planning objectives in GK planning, it is difficult to objectively assess the plan quality across different cases. To reduce these confounding factors on plan quality assessment, we employed our recently developed multiresolution-level inverse planning algorithm to generate a golden plan for each case, which is expected to be on or close to the pareto surface with a similar trade-off as used in the manual plan. The plan quality of the manual plan is then quantified in terms of its deviation from the golden plan. A scoring criterion between 0-100 was designed to calculate a final score for each manual plan to simplify our analysis. Large quality variation was observed in these 22 cases, with two cases having a score lower than 75, three cases scoring between 80 and 85, two cases between 85 and 90, eight cases between 90 and 95, and seven cases higher than 95. Inter- and intra- planner variability was also observed in our study. This large variation in GK manual planning deserves high attention, and merits further investigation on how to reduce the variation in GK treatment plan quality.

A plan quality control method of treatment planning for Gamma Knife radiosurgery

With many variables to adjust, conventional manual forward planning for Gamma Knife (GK) radiosurgery is very complicated and cumbersome. The resulting plan quality heavily depends on planners skills, experiences and devoted efforts, and varies significantly among cases, planners, and institutions. Quality control for GK planning is desired to consistently provide high-quality plan to each patient. In this study, we proposed a quality control method for GK planning by building a database of high-quality GK plans. Patient anatomy was described by target volume, target shape complexity, and spatial relationship between target and nearby organs, which determine GK planning difficulty level. Plan quality was evaluated using target coverage, selectivity, intermediate dose spillage, maximum dose to 0.1 cc of brainstem, mean dose of ipsilateral cochlea, and beam-on time. When a new plan is created, a high-quality plan that has the most similar target volume size and shape complexity will be identified from the database. A model has also been built to predict the dose to brainstem and cochlea based on their overlap volume histograms. The identified reference plan and the predicted organ dose will help planners to make quality control decisions accordingly. To validate this method, we have built a database for vestibular schwannoma, which are considered to be challenging for GK planning due to the irregularly-shaped target and its proximity to brainstem and cochlea. Five cases were tested, among which one case was considered to be of high quality and four cases had a lower plan quality than prediction. These four cases were replanned and got substantially improved. Our results have demonstrated the efficacy of our proposed quality control method. This method may also be used as a plan quality prediction method to facilitate the development of automatic treatment planning for GK radiosurgery.

Learning-Based Synthetic Dual Energy CT Imaging from Single Energy CT for Stopping Power Ratio Calculation in Proton Radiation Therapy

Purpose: Dual-energy CT (DECT) has been shown to derive stopping power ratio (SPR) map with higher accuracy than conventional single energy CT (SECT) by obtaining the energy dependence of photon interactions. However, DECT is not as widely implemented as SECT in proton radiation therapy simulation. This work presents a learning-based method to synthetize DECT images from SECT for proton radiation therapy. Methods: The proposed method uses a residual attention generative adversarial network. Residual blocks with attention gates were used to force the model focus on the difference between DECT maps and SECT images. To evaluate the accuracy of the method, we retrospectively investigated 20 head-and-neck cancer patients with both DECT and SECT scans available. The high and low energy CT images acquired from DECT acted as learning targets in the training process for SECT datasets and were evaluated against results from the proposed method using a leave-one-out cross-validation strategy. To evaluate our method in the context of a practical application, we generated SPR maps from sDECT using physics-based dual-energy stoichiometric method and compared the maps to those generated from DECT. Results: The synthesized DECT images showed an average mean absolute error around 30 Hounsfield Unit (HU) across the whole-body volume. The corresponding SPR maps generated from synthetic DECT showed an average normalized mean square error of about 1% with reduced noise level and artifacts than those from original DECT. Conclusions: The accuracy of the synthesized DECT image by our machine-learning-based method was evaluated on head and neck patient, and potential feasibility for proton treatment planning and dose calculation was shown by generating SPR map using the synthesized DECT.

Learning-Based Stopping Power Mapping on Dual Energy CT for Proton Radiation Therapy

Purpose: Dual-energy CT (DECT) has been used to derive relative stopping power (RSP) map by obtaining the energy dependence of photon interactions. The DECT-derived RSP maps could potentially be compromised by image noise levels and the severity of artifacts when using physics-based mapping techniques, which would affect subsequent clinical applications. This work presents a noise-robust learning-based method to predict RSP maps from DECT for proton radiation therapy. Methods: The proposed method uses a residual attention cycle-consistent generative adversarial (CycleGAN) network. CycleGAN were used to let the DECT-to-RSP mapping be close to a one-to-one mapping by introducing an inverse RSP-to-DECT mapping. We retrospectively investigated 20 head-and-neck cancer patients with DECT scans in proton radiation therapy simulation. Ground truth RSP values were assigned by calculation based on chemical compositions, and acted as learning targets in the training process for DECT datasets, and were evaluated against results from the proposed method using a leave-one-out cross-validation strategy. Results: The predicted RSP maps showed an average normalized mean square error (NMSE) of 2.83% across the whole body volume, and average mean error (ME) less than 3% in all volumes of interest (VOIs). With additional simulated noise added in DECT datasets, the proposed method still maintained a comparable performance, while the physics-based stoichiometric method suffered degraded inaccuracy from increased noise level. The average differences in DVH metrics for clinical target volumes (CTVs) were less than 0.2 Gy for D95% and Dmax with no statistical significance. Conclusion: These results strongly indicate the high accuracy of RSP maps predicted by our machine-learning-based method and show its potential feasibility for proton treatment planning and dose calculation.