Dan Nguyen

Biling Wang, Michael Dohopolski, Ti Bai et al.

We evaluated the temporal performance of a deep learning (DL) based artificial intelligence (AI) model for auto segmentation in prostate radiotherapy, seeking to correlate its efficacy with changes in clinical landscapes. Our study involved 1328 prostate cancer patients who underwent definitive radiotherapy from January 2006 to August 2022 at the University of Texas Southwestern Medical Center. We trained a UNet based segmentation model on data from 2006 to 2011 and tested it on data from 2012 to 2022 to simulate real world clinical deployment. We measured the model performance using the Dice similarity coefficient (DSC), visualized the trends in contour quality using exponentially weighted moving average (EMA) curves. Additionally, we performed Wilcoxon Rank Sum Test to analyze the differences in DSC distributions across distinct periods, and multiple linear regression to investigate the impact of various clinical factors. The model exhibited peak performance in the initial phase (from 2012 to 2014) for segmenting the prostate, rectum, and bladder. However, we observed a notable decline in performance for the prostate and rectum after 2015, while bladder contour quality remained stable. Key factors that impacted the prostate contour quality included physician contouring styles, the use of various hydrogel spacer, CT scan slice thickness, MRI-guided contouring, and using intravenous (IV) contrast. Rectum contour quality was influenced by factors such as slice thickness, physician contouring styles, and the use of various hydrogel spacers. The bladder contour quality was primarily affected by using IV contrast. This study highlights the challenges in maintaining AI model performance consistency in a dynamic clinical setting. It underscores the need for continuous monitoring and updating of AI models to ensure their ongoing effectiveness and relevance in patient care.

Xiao Liang, Howard Morgan, Ti Bai et al.

Cone-beam CT (CBCT)-based online adaptive radiotherapy calls for accurate auto-segmentation to reduce the time cost for physicians to edit contours. However, deep learning (DL)-based direct segmentation of CBCT images is a challenging task, mainly due to the poor image quality and lack of well-labelled large training datasets. Deformable image registration (DIR) is often used to propagate the manual contours on the planning CT (pCT) of the same patient to CBCT. In this work, we undertake solving the problems mentioned above with the assistance of DIR. Our method consists of three main components. First, we use deformed pCT contours derived from multiple DIR methods between pCT and CBCT as pseudo labels for initial training of the DL-based direct segmentation model. Second, we use deformed pCT contours from another DIR algorithm as influencer volumes to define the region of interest for DL-based direct segmentation. Third, the initially trained DL model is further fine-tuned using a smaller set of true labels. We found that DL-based direct segmentation on CBCT trained with pseudo labels and without influencer volumes shows poor performance compared to DIR-based segmentation. However, adding deformed pCT contours as influencer volumes in the direct segmentation network dramatically improves segmentation performance, reaching the accuracy level of DIR-based segmentation. The DL model with influencer volumes can be further improved through fine-tuning using a smaller set of true labels. Experiments showed that 7 out of 19 structures have an at least 0.2 Dice similarity coefficient increase compared to DIR-based segmentation. A DL-based direct CBCT segmentation model can be improved to outperform DIR-based segmentation models by using deformed pCT contours as pseudo labels and influencer volumes for initial training, and by using a smaller set of true labels for model fine tuning.

Michael Dohopolski, Kai Wang, Biling Wang et al.

Prediction uncertainty estimation has clinical significance as it can potentially quantify prediction reliability. Clinicians may trust 'blackbox' models more if robust reliability information is available, which may lead to more models being adopted into clinical practice. There are several deep learning-inspired uncertainty estimation techniques, but few are implemented on medical datasets -- fewer on single institutional datasets/models. We sought to compare dropout variational inference (DO), test-time augmentation (TTA), conformal predictions, and single deterministic methods for estimating uncertainty using our model trained to predict feeding tube placement for 271 head and neck cancer patients treated with radiation. We compared the area under the curve (AUC), sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) trends for each method at various cutoffs that sought to stratify patients into 'certain' and 'uncertain' cohorts. These cutoffs were obtained by calculating the percentile "uncertainty" within the validation cohort and applied to the testing cohort. Broadly, the AUC, sensitivity, and NPV increased as the predictions were more 'certain' -- i.e., lower uncertainty estimates. However, when a majority vote (implementing 2/3 criteria: DO, TTA, conformal predictions) or a stricter approach (3/3 criteria) were used, AUC, sensitivity, and NPV improved without a notable loss in specificity or PPV. Especially for smaller, single institutional datasets, it may be important to evaluate multiple estimations techniques before incorporating a model into clinical practice.

Anjali Balagopal, Dan Nguyen, Ti Bai et al.

When a pre-trained general auto-segmentation model is deployed at a new institution, a support framework in the proposed Prior-guided DDL network will learn the systematic difference between the model predictions and the final contours revised and approved by clinicians for an initial group of patients. The learned style feature differences are concatenated with the new patients (query) features and then decoded to get the style-adapted segmentations. The model is independent of practice styles and anatomical structures. It meta-learns with simulated style differences and does not need to be exposed to any real clinical stylized structures during training. Once trained on the simulated data, it can be deployed for clinical use to adapt to new practice styles and new anatomical structures without further training. To show the proof of concept, we tested the Prior-guided DDL network on six different practice style variations for three different anatomical structures. Pre-trained segmentation models were adapted from post-operative clinical target volume (CTV) segmentation to segment CTVstyle1, CTVstyle2, and CTVstyle3, from parotid gland segmentation to segment Parotidsuperficial, and from rectum segmentation to segment Rectumsuperior and Rectumposterior. The mode performance was quantified with Dice Similarity Coefficient (DSC). With adaptation based on only the first three patients, the average DSCs were improved from 78.6, 71.9, 63.0, 52.2, 46.3 and 69.6 to 84.4, 77.8, 73.0, 77.8, 70.5, 68.1, for CTVstyle1, CTVstyle2, and CTVstyle3, Parotidsuperficial, Rectumsuperior, and Rectumposterior, respectively, showing the great potential of the Priorguided DDL network for a fast and effortless adaptation to new practice styles

Anjali Balagopal, Michael Dohopolski, Young Suk Kwon et al.

Background and purpose: Radiation-induced erectile dysfunction (RiED) is commonly seen in prostate cancer patients. Clinical trials have been developed in multiple institutions to investigate whether dose-sparing to the internal-pudendal-arteries (IPA) will improve retention of sexual potency. The IPA is usually not considered a conventional organ-at-risk (OAR) due to segmentation difficulty. In this work, we propose a deep learning (DL)-based auto-segmentation model for the IPA that utilizes CT and MRI or CT alone as the input image modality to accommodate variation in clinical practice. Materials and methods: 86 patients with CT and MRI images and noisy IPA labels were recruited in this study. We split the data into 42/14/30 for model training, testing, and a clinical observer study, respectively. There were three major innovations in this model: 1) we designed an architecture with squeeze-and-excite blocks and modality attention for effective feature extraction and production of accurate segmentation, 2) a novel loss function was used for training the model effectively with noisy labels, and 3) modality dropout strategy was used for making the model capable of segmentation in the absence of MRI. Results: The DSC, ASD, and HD95 values for the test dataset were 62.2%, 2.54mm, and 7mm, respectively. AI segmented contours were dosimetrically equivalent to the expert physician's contours. The observer study showed that expert physicians' scored AI contours (mean=3.7) higher than inexperienced physicians' contours (mean=3.1). When inexperienced physicians started with AI contours, the score improved to 3.7. Conclusion: The proposed model achieved good quality IPA contours to improve uniformity of segmentation and to facilitate introduction of standardized IPA segmentation into clinical trials and practice.

Ti Bai, Muhan Lin, Xiao Liang et al.

Medical image registration is a fundamental and vital task which will affect the efficacy of many downstream clinical tasks. Deep learning (DL)-based deformable image registration (DIR) methods have been investigated, showing state-of-the-art performance. A test time optimization (TTO) technique was proposed to further improve the DL models' performance. Despite the substantial accuracy improvement with this TTO technique, there still remained some regions that exhibited large registration errors even after many TTO iterations. To mitigate this challenge, we firstly identified the reason why the TTO technique was slow, or even failed, to improve those regions' registration results. We then proposed a two-levels TTO technique, i.e., image-specific optimization (ISO) and region-specific optimization (RSO), where the region can be interactively indicated by the clinician during the registration result reviewing process. For both efficiency and accuracy, we further envisioned a three-step DL-based image registration workflow. Experimental results showed that our proposed method outperformed the conventional method qualitatively and quantitatively.

Aaron Babier, Rafid Mahmood, Binghao Zhang et al.

We establish an open framework for developing plan optimization models for knowledge-based planning (KBP) in radiotherapy. Our framework includes reference plans for 100 patients with head-and-neck cancer and high-quality dose predictions from 19 KBP models that were developed by different research groups during the OpenKBP Grand Challenge. The dose predictions were input to four optimization models to form 76 unique KBP pipelines that generated 7600 plans. The predictions and plans were compared to the reference plans via: dose score, which is the average mean absolute voxel-by-voxel difference in dose a model achieved; the deviation in dose-volume histogram (DVH) criterion; and the frequency of clinical planning criteria satisfaction. We also performed a theoretical investigation to justify our dose mimicking models. The range in rank order correlation of the dose score between predictions and their KBP pipelines was 0.50 to 0.62, which indicates that the quality of the predictions is generally positively correlated with the quality of the plans. Additionally, compared to the input predictions, the KBP-generated plans performed significantly better (P<0.05; one-sided Wilcoxon test) on 18 of 23 DVH criteria. Similarly, each optimization model generated plans that satisfied a higher percentage of criteria than the reference plans. Lastly, our theoretical investigation demonstrated that the dose mimicking models generated plans that are also optimal for a conventional planning model. This was the largest international effort to date for evaluating the combination of KBP prediction and optimization models. In the interest of reproducibility, our data and code is freely available at https://github.com/ababier/open-kbp-opt.

Qiming Yang, Hongyang Chao, Dan Nguyen et al.

The automatic standardization of nomenclature for anatomical structures in radiotherapy (RT) clinical data is a critical prerequisite for data curation and data-driven research in the era of big data and artificial intelligence, but it is currently an unmet need. Existing methods either cannot handle cross-institutional datasets or suffer from heavy imbalance and poor-quality delineation in clinical RT datasets. To solve these problems, we propose an automated structure nomenclature standardization framework, 3D Non-local Network with Voting (3DNNV). This framework consists of an improved data processing strategy, namely, adaptive sampling and adaptive cropping (ASAC) with voting, and an optimized feature extraction module. The framework simulates clinicians' domain knowledge and recognition mechanisms to identify small-volume organs at risk (OARs) with heavily imbalanced data better than other methods. We used partial data from an open-source head-and-neck cancer dataset to train the model, then tested the model on three cross-institutional datasets to demonstrate its generalizability. 3DNNV outperformed the baseline model, achieving higher average true positive rates (TPR) overall categories on the three test datasets (+8.27%, +2.39%, and +5.53%, respectively). More importantly, the 3DNNV outperformed the baseline on the test dataset, 28.63% to 91.17%, in terms of F1 score for a small-volume OAR with only 9 training samples. The results show that 3DNNV can be applied to identify OARs, even error-prone ones. Furthermore, we discussed the limitations and applicability of the framework in practical scenarios. The framework we developed can assist in standardizing structure nomenclature to facilitate data-driven clinical research in cancer radiotherapy.

Margerie Huet-Dastarac, Dan Nguyen, Steve Jiang et al.

Estimating the uncertainty of deep learning models in a reliable and efficient way has remained an open problem, where many different solutions have been proposed in the literature. Most common methods are based on Bayesian approximations, like Monte Carlo dropout (MCDO) or Deep ensembling (DE), but they have a high inference time (i.e. require multiple inference passes) and might not work for out-of-distribution detection (OOD) data (i.e. similar uncertainty for in-distribution (ID) and OOD). In safety critical environments, like medical applications, accurate and fast uncertainty estimation methods, able to detect OOD data, are crucial, since wrong predictions can jeopardize patients safety. In this study, we present an alternative direct uncertainty estimation method and apply it for a regression U-Net architecture. The method consists in the addition of a branch from the bottleneck which reconstructs the input. The input reconstruction error can be used as a surrogate of the model uncertainty. For the proof-of-concept, our method is applied to proton therapy dose prediction in head and neck cancer patients. Accuracy, time-gain, and OOD detection are analyzed for our method in this particular application and compared with the popular MCDO and DE. The input reconstruction method showed a higher Pearson correlation coefficient with the prediction error (0.620) than DE and MCDO (between 0.447 and 0.612). Moreover, our method allows an easier identification of OOD (Z-score of 34.05). It estimates the uncertainty simultaneously to the regression task, therefore requires less time or computational resources.

Biling Wang, Austen Maniscalco, Ti Bai et al.

Purpose: This study presents a Deep Learning (DL)-based quality assessment (QA) approach for evaluating auto-generated contours (auto-contours) in radiotherapy, with emphasis on Online Adaptive Radiotherapy (OART). Leveraging Bayesian Ordinal Classification (BOC) and calibrated uncertainty thresholds, the method enables confident QA predictions without relying on ground truth contours or extensive manual labeling. Methods: We developed a BOC model to classify auto-contour quality and quantify prediction uncertainty. A calibration step was used to optimize uncertainty thresholds that meet clinical accuracy needs. The method was validated under three data scenarios: no manual labels, limited labels, and extensive labels. For rectum contours in prostate cancer, we applied geometric surrogate labels when manual labels were absent, transfer learning when limited, and direct supervision when ample labels were available. Results: The BOC model delivered robust performance across all scenarios. Fine-tuning with just 30 manual labels and calibrating with 34 subjects yielded over 90% accuracy on test data. Using the calibrated threshold, over 93% of the auto-contours' qualities were accurately predicted in over 98% of cases, reducing unnecessary manual reviews and highlighting cases needing correction. Conclusion: The proposed QA model enhances contouring efficiency in OART by reducing manual workload and enabling fast, informed clinical decisions. Through uncertainty quantification, it ensures safer, more reliable radiotherapy workflows.

Xiao Liang, Jaehee Chun, Howard Morgan et al.

Online adaptive radiotherapy (ART) requires accurate and efficient auto-segmentation of target volumes and organs-at-risk (OARs) in mostly cone-beam computed tomography (CBCT) images. Propagating expert-drawn contours from the pre-treatment planning CT (pCT) through traditional or deep learning (DL) based deformable image registration (DIR) can achieve improved results in many situations. Typical DL-based DIR models are population based, that is, trained with a dataset for a population of patients, so they may be affected by the generalizability problem. In this paper, we propose a method called test-time optimization (TTO) to refine a pre-trained DL-based DIR population model, first for each individual test patient, and then progressively for each fraction of online ART treatment. Our proposed method is less susceptible to the generalizability problem, and thus can improve overall performance of different DL-based DIR models by improving model accuracy, especially for outliers. Our experiments used data from 239 patients with head and neck squamous cell carcinoma to test the proposed method. Firstly, we trained a population model with 200 patients, and then applied TTO to the remaining 39 test patients by refining the trained population model to obtain 39 individualized models. We compared each of the individualized models with the population model in terms of segmentation accuracy. The number of patients with at least 0.05 DSC improvement or 2 mm HD95 improvement by TTO averaged over the 17 selected structures for the state-of-the-art architecture Voxelmorph is 10 out of 39 test patients. The average time for deriving the individualized model using TTO from the pre-trained population model is approximately four minutes. When adapting the individualized model to a later fraction of the same patient, the average time is reduced to about one minute and the accuracy is slightly improved.

Ti Bai, Anjali Balagopal, Michael Dohopolski et al.

Automatic segmentation of anatomical structures is critical for many medical applications. However, the results are not always clinically acceptable and require tedious manual revision. Here, we present a novel concept called artificial intelligence assisted contour revision (AIACR) and demonstrate its feasibility. The proposed clinical workflow of AIACR is as follows given an initial contour that requires a clinicians revision, the clinician indicates where a large revision is needed, and a trained deep learning (DL) model takes this input to update the contour. This process repeats until a clinically acceptable contour is achieved. The DL model is designed to minimize the clinicians input at each iteration and to minimize the number of iterations needed to reach acceptance. In this proof-of-concept study, we demonstrated the concept on 2D axial images of three head-and-neck cancer datasets, with the clinicians input at each iteration being one mouse click on the desired location of the contour segment. The performance of the model is quantified with Dice Similarity Coefficient (DSC) and 95th percentile of Hausdorff Distance (HD95). The average DSC/HD95 (mm) of the auto-generated initial contours were 0.82/4.3, 0.73/5.6 and 0.67/11.4 for three datasets, which were improved to 0.91/2.1, 0.86/2.4 and 0.86/4.7 with three mouse clicks, respectively. Each DL-based contour update requires around 20 ms. We proposed a novel AIACR concept that uses DL models to assist clinicians in revising contours in an efficient and effective way, and we demonstrated its feasibility by using 2D axial CT images from three head-and-neck cancer datasets.

Maryam Mashayekhi, Itzel Ramirez Tapia, Anjali Balagopal et al.

Typically, the current dose prediction models are limited to small amounts of data and require re-training for a specific site, often leading to suboptimal performance. We propose a site-agnostic, 3D dose distribution prediction model using deep learning that can leverage data from any treatment site, thus increasing the total data available to train the model. Applying our proposed model to a new target treatment site requires only a brief fine-tuning of the model to the new data and involves no modifications to the model input channels or its parameters. Thus, it can be efficiently adapted to a different treatment site, even with a small training dataset.

Noah Bice, Mohamad Fakhreddine, Ruiqi Li et al.

Volumetric modulated arc therapy planning is a challenging problem in high-dimensional, non-convex optimization. Traditionally, heuristics such as fluence-map-optimization-informed segment initialization use locally optimal solutions to begin the search of the full arc therapy plan space from a reasonable starting point. These routines facilitate arc therapy optimization such that clinically satisfactory radiation treatment plans can be created in about 10 minutes. However, current optimization algorithms favor solutions near their initialization point and are slower than necessary due to plan overparameterization. In this work, arc therapy overparameterization is addressed by reducing the effective dimension of treatment plans with unsupervised deep learning. An optimization engine is then built based on low-dimensional arc representations which facilitates faster planning times.

Jaehee Chun, Justin C. Park, Sven Olberg et al.

In this study, we propose a tailored DL framework for patient-specific performance that leverages the behavior of a model intentionally overfitted to a patient-specific training dataset augmented from the prior information available in an ART workflow - an approach we term Intentional Deep Overfit Learning (IDOL). Implementing the IDOL framework in any task in radiotherapy consists of two training stages: 1) training a generalized model with a diverse training dataset of N patients, just as in the conventional DL approach, and 2) intentionally overfitting this general model to a small training dataset-specific the patient of interest (N+1) generated through perturbations and augmentations of the available task- and patient-specific prior information to establish a personalized IDOL model. The IDOL framework itself is task-agnostic and is thus widely applicable to many components of the ART workflow, three of which we use as a proof of concept here: the auto-contouring task on re-planning CTs for traditional ART, the MRI super-resolution (SR) task for MRI-guided ART, and the synthetic CT (sCT) reconstruction task for MRI-only ART. In the re-planning CT auto-contouring task, the accuracy measured by the Dice similarity coefficient improves from 0.847 with the general model to 0.935 by adopting the IDOL model. In the case of MRI SR, the mean absolute error (MAE) is improved by 40% using the IDOL framework over the conventional model. Finally, in the sCT reconstruction task, the MAE is reduced from 68 to 22 HU by utilizing the IDOL framework.

Dan Nguyen, Fernando Kay, Jun Tan et al.

Since the outbreak of the COVID-19 pandemic, worldwide research efforts have focused on using artificial intelligence (AI) technologies on various medical data of COVID-19-positive patients in order to identify or classify various aspects of the disease, with promising reported results. However, concerns have been raised over their generalizability, given the heterogeneous factors in training datasets. This study aims to examine the severity of this problem by evaluating deep learning (DL) classification models trained to identify COVID-19-positive patients on 3D computed tomography (CT) datasets from different countries. We collected one dataset at UT Southwestern (UTSW), and three external datasets from different countries: CC-CCII Dataset (China), COVID-CTset (Iran), and MosMedData (Russia). We divided the data into 2 classes: COVID-19-positive and COVID-19-negative patients. We trained nine identical DL-based classification models by using combinations of the datasets with a 72% train, 8% validation, and 20% test data split. The models trained on a single dataset achieved accuracy/area under the receiver operating characteristics curve (AUC) values of 0.87/0.826 (UTSW), 0.97/0.988 (CC-CCCI), and 0.86/0.873 (COVID-CTset) when evaluated on their own dataset. The models trained on multiple datasets and evaluated on a test set from one of the datasets used for training performed better. However, the performance dropped close to an AUC of 0.5 (random guess) for all models when evaluated on a different dataset outside of its training datasets. Including the MosMedData, which only contained positive labels, into the training did not necessarily help the performance on the other datasets. Multiple factors likely contribute to these results, such as patient demographics and differences in image acquisition or reconstruction, causing a data shift among different study cohorts.

Anjali Balagopal, Howard Morgan, Michael Dohopoloski et al.

Automatic segmentation of medical images with DL algorithms has proven to be highly successful. With most of these algorithms, inter-observer variation is an acknowledged problem, leading to sub-optimal results. This problem is even more significant in post-operative clinical target volume (post-op CTV) segmentation due to the absence of macroscopic visual tumor in the image. This study, using post-op CTV segmentation as the test bed, tries to determine if physician styles are consistent and learnable, if there is an impact of physician styles on treatment outcome and toxicity; and how to explicitly deal with physician styles in DL algorithms to facilitate its clinical acceptance. A classifier is trained to identify which physician has contoured the CTV from just the contour and corresponding CT scan, to determine if physician styles are consistent and learnable. Next, we evaluate if adapting automatic segmentation to physician styles would be clinically feasible based on a lack of difference between outcomes. For modeling different physician styles of CTV segmentation, a concept called physician style-aware (PSA) segmentation is proposed which is an encoder-multidecoder network trained with perceptual loss. With the proposed physician style-aware network (PSA-Net), Dice similarity coefficient (DSC) accuracy increases on an average of 3.4% for all physicians from a general model that is not style adapted. We show that stylistic contouring variations also exist between institutions that follow the same segmentation guidelines and show the effectiveness of the proposed method in adapting to new institutional styles. We observed an accuracy improvement of 5% in terms of DSC when adapting to the style of a separate institution.

Anjali Balagopal, Dan Nguyen, Maryam Mashayekhi et al.

Inter-observer variation is a significant problem in clinical target volume(CTV) segmentation in postoperative settings, where there is no gross tumor present. In this scenario, the CTV is not an anatomically established structure, but one determined by the physician based on the clinical guideline used, the preferred tradeoff between tumor control and toxicity, their experience and training background, and other factors. This results in high inter-observer variability between physicians. This variability has been considered an issue, but the absence of multiple physician CTV contours for each patient and the significant amount of time required for dose planning have made it impractical to study its dosimetric consequences. In this study, we analyze the impact that variations in physician style have on dose to organs-at-risk(OAR) by simulating the clinical workflow via deep learning. For a given patient previously treated by one physician, we use deep learning-based tools to simulate how other physicians would contour the CTV and how the corresponding dose distributions would look for this patient. To simulate multiple physician styles, we use a previously developed in-house CTV segmentation model that can produce physician style-aware segmentations. The corresponding dose distribution is predicted using another in-house deep learning tool, which, can predict dose within 3% of the prescription dose, on average, on the test data. For every test patient, four different physician style CTVs are considered, and four different dose distributions are analyzed. OAR dose metrics are compared, showing that even though physician style variations result in organs getting different doses, all the important dose metrics except Maximum Dose point are within the clinically acceptable limit.

Ti Bai, Dan Nguyen, Biling Wang et al.

Low Dose Computed Tomography (LDCT) is clinically desirable due to the reduced radiation to patients. However, the quality of LDCT images is often sub-optimal because of the inevitable strong quantum noise. Inspired by their unprecedent success in computer vision, deep learning (DL)-based techniques have been used for LDCT denoising. Despite the promising noise removal ability of DL models, people have observed that the resolution of the DL-denoised images is compromised, decreasing their clinical value. Aiming at relieving this problem, in this work, we developed a more effective denoiser by introducing a high-resolution network (HRNet). Since HRNet consists of multiple branches of subnetworks to extract multiscale features which are later fused together, the quality of the generated features can be substantially enhanced, leading to improved denoising performance. Experimental results demonstrated that the introduced HRNet-based denoiser outperforms the benchmarked UNet-based denoiser in terms of superior image resolution preservation ability while comparable, if not better, noise suppression ability. Quantitative metrics in terms of root-mean-squared-errors (RMSE)/structure similarity index (SSIM) showed that the HRNet-based denoiser can improve the values from 113.80/0.550 (LDCT) to 55.24/0.745 (HRNet), in comparison to 59.87/0.712 for the UNet-based denoiser.

Ti Bai, Biling Wang, Dan Nguyen et al.

Monte Carlo (MC) simulation is considered the gold standard method for radiotherapy dose calculation. However, achieving high precision requires a large number of simulation histories, which is time consuming. The use of computer graphics processing units (GPUs) has greatly accelerated MC simulation and allows dose calculation within a few minutes for a typical radiotherapy treatment plan. However, some clinical applications demand real time efficiency for MC dose calculation. To tackle this problem, we have developed a real time, deep learning based dose denoiser that can be plugged into a current GPU based MC dose engine to enable real time MC dose calculation. We used two different acceleration strategies to achieve this goal: 1) we applied voxel unshuffle and voxel shuffle operators to decrease the input and output sizes without any information loss, and 2) we decoupled the 3D volumetric convolution into a 2D axial convolution and a 1D slice convolution. In addition, we used a weakly supervised learning framework to train the network, which greatly reduces the size of the required training dataset and thus enables fast fine tuning based adaptation of the trained model to different radiation beams. Experimental results show that the proposed denoiser can run in as little as 39 ms, which is around 11.6 times faster than the baseline model. As a result, the whole MC dose calculation pipeline can be finished within 0.15 seconds, including both GPU MC dose calculation and deep learning based denoising, achieving the real time efficiency needed for some radiotherapy applications, such as online adaptive radiotherapy.

Ti Bai, Biling Wang, Dan Nguyen et al.

Low dose computed tomography (LDCT) is desirable for both diagnostic imaging and image guided interventions. Denoisers are openly used to improve the quality of LDCT. Deep learning (DL)-based denoisers have shown state-of-the-art performance and are becoming one of the mainstream methods. However, there exists two challenges regarding the DL-based denoisers: 1) a trained model typically does not generate different image candidates with different noise-resolution tradeoffs which sometimes are needed for different clinical tasks; 2) the model generalizability might be an issue when the noise level in the testing images is different from that in the training dataset. To address these two challenges, in this work, we introduce a lightweight optimization process at the testing phase on top of any existing DL-based denoisers to generate multiple image candidates with different noise-resolution tradeoffs suitable for different clinical tasks in real-time. Consequently, our method allows the users to interact with the denoiser to efficiently review various image candidates and quickly pick up the desired one, and thereby was termed as deep interactive denoiser (DID). Experimental results demonstrated that DID can deliver multiple image candidates with different noise-resolution tradeoffs, and shows great generalizability regarding various network architectures, as well as training and testing datasets with various noise levels.

Dan Nguyen, Azar Sadeghnejad Barkousaraie, Gyanendra Bohara et al.

Recently, artificial intelligence technologies and algorithms have become a major focus for advancements in treatment planning for radiation therapy. As these are starting to become incorporated into the clinical workflow, a major concern from clinicians is not whether the model is accurate, but whether the model can express to a human operator when it does not know if its answer is correct. We propose to use Monte Carlo dropout (MCDO) and the bootstrap aggregation (bagging) technique on deep learning models to produce uncertainty estimations for radiation therapy dose prediction. We show that both models are capable of generating a reasonable uncertainty map, and, with our proposed scaling technique, creating interpretable uncertainties and bounds on the prediction and any relevant metrics. Performance-wise, bagging provides statistically significant reduced loss value and errors in most of the metrics investigated in this study. The addition of bagging was able to further reduce errors by another 0.34% for Dmean and 0.19% for Dmax, on average, when compared to the baseline framework. Overall, the bagging framework provided significantly lower MAE of 2.62, as opposed to the baseline framework's MAE of 2.87. The usefulness of bagging, from solely a performance standpoint, does highly depend on the problem and the acceptable predictive error, and its high upfront computational cost during training should be factored in to deciding whether it is advantageous to use it. In terms of deployment with uncertainty estimations turned on, both frameworks offer the same performance time of about 12 seconds. As an ensemble-based metaheuristic, bagging can be used with existing machine learning architectures to improve stability and performance, and MCDO can be applied to any deep learning models that have dropout as part of their architecture.

Roya Norouzi Kandalan, Dan Nguyen, Nima Hassan Rezaeian et al.

This work aims to study the generalizability of a pre-developed deep learning (DL) dose prediction model for volumetric modulated arc therapy (VMAT) for prostate cancer and to adapt the model to three different internal treatment planning styles and one external institution planning style. We built the source model with planning data from 108 patients previously treated with VMAT for prostate cancer. For the transfer learning, we selected patient cases planned with three different styles from the same institution and one style from a different institution to adapt the source model to four target models. We compared the dose distributions predicted by the source model and the target models with the clinical dose predictions and quantified the improvement in the prediction quality for the target models over the source model using the Dice similarity coefficients (DSC) of 10% to 100% isodose volumes and the dose-volume-histogram (DVH) parameters of the planning target volume and the organs-at-risk. The source model accurately predicts dose distributions for plans generated in the same source style but performs sub-optimally for the three internal and one external target styles, with the mean DSC ranging between 0.81-0.94 and 0.82-0.91 for the internal and the external styles, respectively. With transfer learning, the target model predictions improved the mean DSC to 0.88-0.95 and 0.92-0.96 for the internal and the external styles, respectively. Target model predictions significantly improved the accuracy of the DVH parameter predictions to within 1.6%. We demonstrated model generalizability for DL-based dose prediction and the feasibility of using transfer learning to solve this problem. With 14-29 cases per style, we successfully adapted the source model into several different practice styles. This indicates a realistic way to widespread clinical implementation of DL-based dose prediction.

Gyanendra Bohara, Azar Sadeghnejad Barkousaraie, Steve Jiang et al.

We propose to develop deep learning models that can predict Pareto optimal dose distributions by using any given set of beam angles, along with patient anatomy, as input to train the deep neural networks. We implement and compare two deep learning networks that predict with two different beam configuration modalities. We generated Pareto optimal plans for 70 patients with prostate cancer. We used fluence map optimization to generate 500 IMRT plans that sampled the Pareto surface for each patient, for a total of 35,000 plans. We studied and compared two different models, Model I and Model II. Model I directly uses beam angles as a second input to the network as a binary vector. Model II converts the beam angles into beam doses that are conformal to the PTV. Our deep learning models predicted voxel-level dose distributions that precisely matched the ground truth dose distributions. Quantitatively, Model I prediction error of 0.043 (confirmation), 0.043 (homogeneity), 0.327 (R50), 2.80% (D95), 3.90% (D98), 0.6% (D50), 1.10% (D2) was lower than that of Model II, which obtained 0.076 (confirmation), 0.058 (homogeneity), 0.626 (R50), 7.10% (D95), 6.50% (D98), 8.40% (D50), 6.30% (D2). Treatment planners who use our models will be able to use deep learning to control the tradeoffs between the PTV and OAR weights, as well as the beam number and configurations in real time. Our dose prediction methods provide a stepping stone to building automatic IMRT treatment planning.

Ti Bai, Dan Nguyen, Biling Wang et al.

Despite the indispensable role of X-ray computed tomography (CT) in diagnostic medicine field, the associated ionizing radiation is still a major concern considering that it may cause genetic and cancerous diseases. Decreasing the exposure can reduce the dose and hence the radiation-related risk, but will also induce higher quantum noise. Supervised deep learning can be used to train a neural network to denoise the low-dose CT (LDCT). However, its success requires massive pixel-wise paired LDCT and normal-dose CT (NDCT) images, which are rarely available in real practice. To alleviate this problem, in this paper, a shift-invariant property based neural network was devised to learn the inherent pixel correlations and also the noise distribution by only using the LDCT images, shaping into our probabilistic self-learning framework. Experimental results demonstrated that the proposed method outperformed the competitors, producing an enhanced LDCT image that has similar image style as the routine NDCT which is highly-preferable in clinic practice.

Anjali Balagopal, Dan Nguyen, Howard Morgan et al.

In post-operative radiotherapy for prostate cancer, the cancerous prostate gland has been surgically removed, so the clinical target volume (CTV) to be irradiated encompasses the microscopic spread of tumor cells, which cannot be visualized in typical clinical images such as computed tomography or magnetic resonance imaging. In current clinical practice, physicians segment CTVs manually based on their relationship with nearby organs and other clinical information, per clinical guidelines. Automating post-operative prostate CTV segmentation with traditional image segmentation methods has been a major challenge. Here, we propose a deep learning model to overcome this problem by segmenting nearby organs first, then using their relationship with the CTV to assist CTV segmentation. The model proposed is trained using labels clinically approved and used for patient treatment, which are subject to relatively large inter-physician variations due to the absence of a visual ground truth. The model achieves an average Dice similarity coefficient (DSC) of 0.87 on a holdout dataset of 50 patients, much better than established methods, such as atlas-based methods (DSC<0.7). The uncertainties associated with automatically segmented CTV contours are also estimated to help physicians inspect and revise the contours, especially in areas with large inter-physician variations. We also use a 4-point grading system to show that the clinical quality of the automatically segmented CTV contours is equal to that of approved clinical contours manually drawn by physicians.

Xiao Liang, Dan Nguyen, Steve Jiang

Generalizability is a concern when applying a deep learning (DL) model trained on one dataset to other datasets. Training a universal model that works anywhere, anytime, for anybody is unrealistic. In this work, we demonstrate the generalizability problem, then explore potential solutions based on transfer learning (TL) by using the cone-beam computed tomography (CBCT) to computed tomography (CT) image conversion task as the testbed. Previous works have converted CBCT to CT-like images. However, all of those works studied only one or two anatomical sites and used images from the same vendor's scanners. Here, we investigated how a model trained for one machine and one anatomical site works on other machines and other sites. We trained a model on CBCT images acquired from one vendor's scanners for head and neck cancer patients and applied it to images from another vendor's scanners and for other disease sites. We found that generalizability could be a significant problem for this particular application when applying a trained DL model to datasets from another vendor's scanners. We then explored three practical solutions based on TL to solve this generalization problem: the target model, which is trained on a target domain from scratch; the combined model, which is trained on both source and target domain datasets from scratch; and the adapted model, which fine-tunes the trained source model to a target domain. We found that when there are sufficient data in the target domain, all three models can achieve good performance. When the target dataset is limited, the adapted model works the best, which indicates that using the fine-tuning strategy to adapt the trained model to an unseen target domain dataset is a viable and easy way to implement DL models in the clinic.

Azar Sadeghnejad-Barkousaraie, Gyanendra Bohara, Steve Jiang et al.

Due to the large combinatorial problem, current beam orientation optimization algorithms for radiotherapy, such as column generation (CG), are typically heuristic or greedy in nature, leading to suboptimal solutions. We propose a reinforcement learning strategy using Monte Carlo Tree Search capable of finding a superior beam orientation set and in less time than CG.We utilized a reinforcement learning structure involving a supervised learning network to guide Monte Carlo tree search (GTS) to explore the decision space of beam orientation selection problem. We have previously trained a deep neural network (DNN) that takes in the patient anatomy, organ weights, and current beams, and then approximates beam fitness values, indicating the next best beam to add. This DNN is used to probabilistically guide the traversal of the branches of the Monte Carlo decision tree to add a new beam to the plan. To test the feasibility of the algorithm, we solved for 5-beam plans, using 13 test prostate cancer patients, different from the 57 training and validation patients originally trained the DNN. To show the strength of GTS to other search methods, performances of three other search methods including a guided search, uniform tree search and random search algorithms are also provided. On average GTS outperforms all other methods, it find a solution better than CG in 237 seconds on average, compared to CG which takes 360 seconds, and outperforms all other methods in finding a solution with lower objective function value in less than 1000 seconds. Using our guided tree search (GTS) method we were able to maintain a similar planning target volume (PTV) coverage within 1% error, and reduce the organ at risk (OAR) mean dose for body, rectum, left and right femoral heads, but a slight increase of 1% in bladder mean dose.

Dan Nguyen, Rafe McBeth, Azar Sadeghnejad Barkousaraie et al.

We propose a novel domain specific loss, which is a differentiable loss function based on the dose volume histogram, and combine it with an adversarial loss for the training of deep neural networks to generate Pareto optimal dose distributions. The mean squared error (MSE) loss, dose volume histogram (DVH) loss, and adversarial (ADV) loss were used to train 4 instances of the neural network model: 1) MSE, 2) MSE+ADV, 3) MSE+DVH, and 4) MSE+DVH+ADV. 70 prostate patients were acquired, and the dose influence arrays were calculated for each patient. 1200 Pareto surface plans per patient were generated by pseudo-randomizing the tradeoff weights (84,000 plans total). We divided the data into 54 training, 6 validation, and 10 testing patients. Each model was trained for 100,000 iterations, with a batch size of 2. The prediction time of each model is 0.052 seconds. Quantitatively, the MSE+DVH+ADV model had the lowest prediction error of 0.038 (conformation), 0.026 (homogeneity), 0.298 (R50), 1.65% (D95), 2.14% (D98), 2.43% (D99). The MSE model had the worst prediction error of 0.134 (conformation), 0.041 (homogeneity), 0.520 (R50), 3.91% (D95), 4.33% (D98), 4.60% (D99). For both the mean dose PTV error and the max dose PTV, Body, Bladder and rectum error, the MSE+DVH+ADV outperformed all other models. All model's predictions have an average mean and max dose error less than 2.8% and 4.2%, respectively. Expert human domain specific knowledge can be the largest driver in the performance improvement, and adversarial learning can be used to further capture nuanced features. The real-time prediction capabilities allow for a physician to quickly navigate the tradeoff space, and produce a dose distribution as a tangible endpoint for the dosimetrist to use for planning. This can considerably reduce the treatment planning time, allowing for clinicians to focus their efforts on challenging cases.

Ana M. Barragan-Montero, Dan Nguyen, Weiguo Lu et al.

The use of neural networks to directly predict three-dimensional dose distributions for automatic planning is becoming popular. However, the existing methods only use patient anatomy as input and assume consistent beam configuration for all patients in the training database. The purpose of this work is to develop a more general model that, in addition to patient anatomy, also considers variable beam configurations, to achieve a more comprehensive automatic planning with a potentially easier clinical implementation, without the need of training specific models for different beam settings.

Anjali Balagopal, Samaneh Kazemifar, Dan Nguyen et al.

Accurate segmentation of prostate and surrounding organs at risk is important for prostate cancer radiotherapy treatment planning. We present a fully automated workflow for male pelvic CT image segmentation using deep learning. The architecture consists of a 2D localization network followed by a 3D segmentation network for volumetric segmentation of prostate, bladder, rectum, and femoral heads. We used a multi-channel 2D U-Net followed by a 3D U-Net with encoding arm modified with aggregated residual networks, known as ResNeXt. The models were trained and tested on a pelvic CT image dataset comprising 136 patients. Test results show that 3D U-Net based segmentation achieves mean (SD) Dice coefficient values of 90 (2.0)% ,96 (3.0)%, 95 (1.3)%, 95 (1.5)%, and 84 (3.7)% for prostate, left femoral head, right femoral head, bladder, and rectum, respectively, using the proposed fully automated segmentation method.

Dan Nguyen, Xun Jia, David Sher et al.

The treatment planning process for patients with head and neck (H&N) cancer is regarded as one of the most complicated due to large target volume, multiple prescription dose levels, and many radiation-sensitive critical structures near the target. Treatment planning for this site requires a high level of human expertise and a tremendous amount of effort to produce personalized high quality plans, taking as long as a week, which deteriorates the chances of tumor control and patient survival. To solve this problem, we propose to investigate a deep learning-based dose prediction model, Hierarchically Densely Connected U-net, based on two highly popular network architectures: U-net and DenseNet. We find that this new architecture is able to accurately and efficiently predict the dose distribution, outperforming the other two models, the Standard U-net and DenseNet, in homogeneity, dose conformity, and dose coverage on the test data. Averaging across all organs at risk, our proposed model is capable of predicting the organ-at-risk max dose within 6.3% and mean dose within 5.1% of the prescription dose on the test data. The other models, the Standard U-net and DenseNet, performed worse, having an averaged organ-at-risk max dose prediction error of 8.2% and 9.3%, respectively, and averaged mean dose prediction error of 6.4% and 6.8%, respectively. In addition, our proposed model used 12 times less trainable parameters than the Standard U-net, and predicted the patient dose 4 times faster than DenseNet.

Zohaib Iqbal, Da Luo, Peter Henry et al.

Deep learning has started to revolutionize several different industries, and the applications of these methods in medicine are now becoming more commonplace. This study focuses on investigating the feasibility of tracking patients and clinical staff wearing Bluetooth Low Energy (BLE) tags in a radiation oncology clinic using artificial neural networks (ANNs) and convolutional neural networks (CNNs). The performance of these networks was compared to relative received signal strength indicator (RSSI) thresholding and triangulation. By utilizing temporal information, a combined CNN+ANN network was capable of correctly identifying the location of the BLE tag with an accuracy of 99.9%. It outperformed a CNN model (accuracy = 94%), a thresholding model employing majority voting (accuracy = 95%), and a triangulation classifier utilizing majority voting (accuracy = 95%). Future studies will seek to deploy this affordable real time location system in hospitals to improve clinical workflow, efficiency, and patient safety.

Dan Nguyen, Troy Long, Xun Jia et al.

With the advancement of treatment modalities in radiation therapy for cancer patients, outcomes have improved, but at the cost of increased treatment plan complexity and planning time. The accurate prediction of dose distributions would alleviate this issue by guiding clinical plan optimization to save time and maintain high quality plans. We have modified a convolutional deep network model, U-net (originally designed for segmentation purposes), for predicting dose from patient image contours of the planning target volume (PTV) and organs at risk (OAR). We show that, as an example, we are able to accurately predict the dose of intensity-modulated radiation therapy (IMRT) for prostate cancer patients, where the average Dice similarity coefficient is 0.91 when comparing the predicted vs. true isodose volumes between 0% and 100% of the prescription dose. The average value of the absolute differences in [max, mean] dose is found to be under 5% of the prescription dose, specifically for each structure is [1.80%, 1.03%](PTV), [1.94%, 4.22%](Bladder), [1.80%, 0.48%](Body), [3.87%, 1.79%](L Femoral Head), [5.07%, 2.55%](R Femoral Head), and [1.26%, 1.62%](Rectum) of the prescription dose. We thus managed to map a desired radiation dose distribution from a patient's PTV and OAR contours. As an additional advantage, relatively little data was used in the techniques and models described in this paper.