Gross Tumor Volume Contours Research Articles

Paired conditional generative adversarial network for highly accelerated liver 4D MRI

Purpose. 4D MRI with high spatiotemporal resolution is desired for image-guided liver radiotherapy. Acquiring densely sampling k-space data is time-consuming. Accelerated acquisition with sparse samples is desirable but often causes degraded image quality or long reconstruction time. We propose the Reconstruct Paired Conditional Generative Adversarial Network (Re-Con-GAN) to shorten the 4D MRI reconstruction time while maintaining the reconstruction quality. Methods. Patients who underwent free-breathing liver 4D MRI were included in the study. Fully- and retrospectively under-sampled data at 3, 6 and 10 times (3×, 6× and 10×) were first reconstructed using the nuFFT algorithm. Re-Con-GAN then trained input and output in pairs. Three types of networks, ResNet9, UNet and reconstruction swin transformer (RST), were explored as generators. PatchGAN was selected as the discriminator. Re-Con-GAN processed the data (3D + t) as temporal slices (2D + t). A total of 48 patients with 12 332 temporal slices were split into training (37 patients with 10 721 slices) and test (11 patients with 1611 slices). Compressed sensing (CS) reconstruction with spatiotemporal sparsity constraint was used as a benchmark. Reconstructed image quality was further evaluated with a liver gross tumor volume (GTV) localization task using Mask-RCNN trained from a separate 3D static liver MRI dataset (70 patients; 103 GTV contours). Results. Re-Con-GAN consistently achieved comparable/better PSNR, SSIM, and RMSE scores compared to CS/UNet models. The inference time of Re-Con-GAN, UNet and CS are 0.15, 0.16, and 120 s. The GTV detection task showed that Re-Con-GAN and CS, compared to UNet, better improved the dice score (3× Re-Con-GAN 80.98%; 3× CS 80.74%; 3× UNet 79.88%) of unprocessed under-sampled images (3× 69.61%). Conclusion. A generative network with adversarial training is proposed with promising and efficient reconstruction results demonstrated on an in-house dataset. The rapid and qualitative reconstruction of 4D liver MR has the potential to facilitate online adaptive MR-guided radiotherapy for liver cancer.

Deep Learning for Automatic Gross Tumor Volumes Contouring in Esophageal Cancer Based on Contrast-Enhanced Computed Tomography Images: A Multi-Institutional Study

PurposeTo develop and externally validate an automatic Artificial Intelligence (AI) tool for delineating gross tumor volume (GTV) in esophageal squamous cell carcinoma (ESCC) patients, which can assist in the neo-adjuvant or radical radiation therapy treatment planning. Methods and MaterialsIn this multi-institutional study, contrast-enhanced CT images from 580 eligible ESCC patients were retrospectively collected. The GTV contours delineated by two experts via consensus were used as ground truth. A three-dimensional deep learning model was developed for GTV contouring in the training cohort and internally and externally validated in three validation cohorts. The AI tool was compared against twelve board-certified experts in 25 patients randomly selected from the external validation cohort to evaluate its assistance in improving contouring performance and reducing variation. Contouring performance was measured using dice similarity coefficient (DSC) and average surface distance (ASD). Additionally, our previously established radiomics model for predicting pathological complete response (pCR) was utilized to compare AI-generated and ground truth contours, in order to assess the potential of the AI contouring tool in radiomics analysis. ResultsThe AI tool demonstrated good GTV contouring performance in multi-center validation cohorts, with median DSC values of 0.865, 0.876, and 0.866, and median ASD values of 0.939 mm, 0.789 mm, and 0.875 mm, respectively. Furthermore, the AI tool significantly improved contouring performance for half of twelve board-certified experts (DSC values, 0.794-0.835 vs 0.856-0.881, P = 0.003-0.048), reduced the intra- and inter-observer variations by 37.4% and 55.2%, respectively, and saved contouring time by 77.6%. In the radiomics analysis, 88.7% of radiomic features from ground truth and AI-generated contours demonstrated stable reproducibility, and similar pCR prediction performance for these contours (P = 0.430) were observed. ConclusionsOur AI contouring tool can improve GTV contouring performance and facilitate radiomics analysis in ESCC patients, which indicated its potential on GTV contouring during radiation therapy treatment planning and radiomics studies.

Volumetric MRI with sparse sampling for MR-guided 3D motion tracking via sparse prior-augmented implicit neural representation learning.

Volumetric reconstruction of magnetic resonance imaging (MRI) from sparse samples is desirable for 3D motion tracking and promises to improve magnetic resonance (MR)-guided radiation treatment precision. Data-driven sparse MRI reconstruction, however, requires large-scale training datasets for prior learning, which is time-consuming and challenging to acquire in clinical settings. To investigate volumetric reconstruction of MRI from sparse samples of two orthogonal slices aided by sparse priors of two static 3D MRI through implicit neural representation (NeRP) learning, in support of 3D motion tracking during MR-guided radiotherapy. A multi-layer perceptron network was trained to parameterize the NeRP model of a patient-specific MRI dataset, where the network takes 4D data coordinates of voxel locations and motion states as inputs and outputs corresponding voxel intensities. By first training the network to learn the NeRP of two static 3D MRI with different breathing motion states, prior information of patient breathing motion was embedded into network weights through optimization. The prior information was then augmented from two motion states to 31 motion states by querying the optimized network at interpolated and extrapolated motion state coordinates. Starting from the prior-augmented NeRP model as an initialization point, we further trained the network to fit sparse samples of two orthogonal MRI slices and the final volumetric reconstruction was obtained by querying the trained network at 3D spatial locations. We evaluated the proposed method using 5-min volumetric MRI time series with 340ms temporal resolution for seven abdominal patients with hepatocellular carcinoma, acquired using golden-angle radial MRI sequence and reconstructed through retrospective sorting. Two volumetric MRI with inhale and exhale states respectively were selected from the first 30s of the time series for prior embedding and augmentation. The remaining 4.5-min time series was used for volumetric reconstruction evaluation, where we retrospectively subsampled each MRI to two orthogonal slices and compared model-reconstructed images to ground truth images in terms of image quality and the capability of supporting 3D target motion tracking. Across the seven patients evaluated, the peak signal-to-noise-ratio between model-reconstructed and ground truth MR images was 38.02±2.60dB and the structure similarity index measure was 0.98±0.01. Throughout the 4.5-min time period, gross tumor volume (GTV) motion estimated by deforming a reference state MRI to model-reconstructed and ground truth MRI showed good consistency. The 95-percentile Hausdorff distance between GTV contours was 2.41±0.77mm, which is less than the voxel dimension. The mean GTV centroid position difference between ground truth and model estimation was less than 1mm in all three orthogonal directions. A prior-augmented NeRP model has been developed to reconstruct volumetric MRI from sparse samples of orthogonal cine slices. Only one exhale and one inhale 3D MRI were needed to train the model to learn prior information of patient breathing motion for sparse image reconstruction. The proposed model has the potential of supporting 3D motion tracking during MR-guided radiotherapy for improved treatment precision and promises a major simplification of the workflow by eliminating the need for large-scale training datasets.

Automated Contouring of Cervical Lymph Nodes and Clinical Target Volumes for Nasopharyngeal Carcinoma Based on Deep Learning and Experience Constraints

Application of artificial intelligence (AI) for automated contouring of tumor volumes and organs at risk (OARs) for radiotherapy of nasopharyngeal carcinoma (NPC) leads to improved contouring accuracy and efficiency. However, few studies have involved the automated contouring of gross tumor volume of cervical lymph nodes (GTVn) and clinical target volumes (CTVs). In this work, we proposed an AI automated contouring tool for GTVn and CTVs for radiotherapy of NPC on the plain scans of planning compute tomography (CT). In this retrospective study, plain scan datasets of planning CT covering the nasopharynx and neck from 139 patients with NPC between March 2022 and December 2022 were collected and divided into training, validation, and testing cohorts of 95, 24, and 20 patients, respectively. Ground truth contours of primary gross tumor volume (GTVp), GTVn (divided into GTVn_L in left neck and GTVn_R in right neck), CTVs (including high risk CTV1 contains GTVp and low risk CTV2 contains GTVp and cervical nodal levels) and OARs were delineated and were defined by consensus of two experts. We first proposed a three-dimensional (3D) U-net using GTVp and OARs as experience constrains to guide the automated delineation of GTVn and CTVs. The average Dice similarity coefficient (DSC) and average surface distance (ASD) were used to quantify the performance of the AI tool. Next, five prospective patients were enrolled for clinical evaluation of our AI tool. DSC between automated contours and radiation oncologist-revised contours and time consuming of the revision were record. Clinical characteristics of 139 retrospective and 5 prospective patients are list in Table 1. In the independent testing set of 20 patients, our AI tool showed high performance in GTVn and CTVs contouring when compared with the ground truth contours. The mean DSC were 0.73 ± 0.07, 0.74 ± 0.05, 0.93 ± 0.03, and 0.88 ± 0.03, and the mean ASD were 1.01 ± 0.43 mm, 1.14 ± 0.61 mm, 0.51 ± 0.13 mm, 1.17 ± 0.43 mm for GTVn_L, GTVn_R, CTV1 and CTV2, respectively. In the five prospective patients, mean DSC were 0.74 ± 0.07, 0.74 ± 0.10, 0.95 ± 0.01 and 0.89 ± 0.04, respectively. The median time consuming for GTVn and CTVs revision was 2minutes and 10 seconds (range, 1 minutes to 3 minutes). The proposed AI tool integrating clinical experience as constrains showed high accuracy for contouring GTVn and CTVs of NPC. With the assistance of AI contours, contouring efficiency could be probably increased, which is promising in online adaptive radiotherapy of NPC.

Real Time Volumetric MRI for MR-Guided 3D Motion Tracking via Sparse Prior-Augmented Neural Representation Learning

To reconstruct volumetric MRI from orthogonal cine acquisition aided by sparse priors of 2 static 3D MRI through implicit neural representation (NeRP) learning, with the goal of eliminating large-scale training datasets for data-driven sparse MRI reconstruction and supporting clinical workflow of real time 3D motion tracking during MR-guided radiotherapy. A multi-layer perceptron network was trained to learn the NeRP of a patient-specific MRI dataset, where the network takes 4D data coordinates of voxel locations and motion states as inputs and outputs corresponding voxel intensities. By first learning the NeRP of 2 static 3D MRI with different breathing motion states, prior knowledge of patient breathing motion was embedded into network weights through optimization. The prior knowledge was then augmented from 2 to 31 motion states by querying the optimized network at interpolated/extrapolated motion state coordinates. Starting from the prior-augmented network as an initialization point, the network was further trained using sparse samples of 2 orthogonal cine slices. The final volumetric reconstruction was obtained by querying the trained network at desired 3D spatial locations. We evaluated the proposed method using 5-minute volumetric MRI time series with 340 ms temporal resolution collected from 7 liver carcinoma patients. The time series was acquired using golden-angle radial MRI sequence and reconstructed through retrospective sorting. Two MRI with inhale and exhale states respectively were selected from the first 30 sec of the time series for prior embedding and augmentation. The remaining 4.5-min time series was used for volumetric reconstruction evaluation, where we retrospectively subsampled each MRI to 2 orthogonal slices and compared network-reconstructed images to ground truth images in terms of image quality and the capability of supporting 3D target motion tracking. Across the 7 patients evaluated, the peak signal to noise ratio between model reconstruction and ground truth was 54.66 ± 6.16 dB and the structural similarity index measure was 0.99 ± 0.01. Gross tumor volume (GTV) contours estimated by deforming a reference state MRI to model-reconstructed and ground truth MRI showed good consistency. The 95-percentile Hausdorff distance between GTV contours was 1.89 ± 1.13 mm, which is less than the voxel dimension. The mean GTV centroid position difference between ground truth and model estimation was less than 1 mm in all 3 orthogonal directions. Volumetric MRI from orthogonal cine acquisition with sparse priors is feasible by modeling prior knowledge through implicit neural representation learning. The model-reconstructed images showed sufficient accuracy in supporting 3D motion tracking of abdominal targets. By eliminating the need for large scale training datasets, the method promises to enable clinical implementation of 3D motion tracking for precision radiation therapy.

Failure Mode and Effects Analysis Prior to the Introduction of AI Generated GTVs for Brain Metastases in the Clinical Workflow

AI autosegmentation of organs-at-risk (OARs) is common practice at many radiotherapy clinics. Despite the abundance of gross tumor volume (GTV) autosegmentation algorithms, adoption in clinical care has been slow due to the high risk associated with errors in GTV delineation. Here we present a failure mode and effects analysis (FMEA) to evaluate the risk associated with introducing AI derived GTVs in patients treated with stereotactic radiosurgery (SRS). An AI GTV autosegmentation algorithm for brain metastases was developed in-house based on a V-Net 3D CNN. Registered CT and MR images and a contour of the brain are input into the software and all identified lesions are returned in a DICOM-RT structure set. Following algorithm evaluation, a workflow was developed to enable AI GTV autosegmentation to be introduced clinically for every SRS patient. The following steps were added to existing procedures: 1) workflow to send CT/MR and brain structure to external server, 2) autosegmentation run on the server, 3) AI GTV structures with a standard nomenclature added to existing OAR structure set, and 4) MD review, editing, and approval of AI GTVs. After successfully completing the physics evaluation testing of the new process, we formed a team of 10 faculty and staff including physicists, residents, physicians, and planners to perform the FMEA prior to clinical implementation. The team met to map the process, identify potential failure modes, and score their frequency of occurrence, severity, and detectability. A 3-point scale (1, 3, or 5) was used to simplify the scoring process. Occurrence was defined as rare, sometimes, or often; severity as low, medium, or high; and detectability as obvious, possible, or challenging. The risk probability numbers (RPNs) were calculated and the steps in the process with the highest RPNs were flagged for further discussion. The FMEA team completed their process map and analysis primarily in 4 meetings. The process map began with acquisition of the patients CT simulation scan and ended with physician approval of final volumes for treatment planning. We identified 17 process steps and 72 possible failure modes, of which 26 were associated with the new workflow. Eighteen failure modes had an RPN greater than 30 (highest risk score in at least one category) and were flagged to assess mitigation strategies. Five were unique to the new AI GTV workflow and mitigation strategies will be designed prior to clinical use. Those involved risks related to inaccurate AI GTV contours, false positives, and an incomplete review stemming from over-reliance by team members on AI. AI is increasingly being employed at every step of radiotherapy to automate and streamline processes. The FMEA analysis resulted in the identification of the riskiest parts of using AI GTV autosegmentation. This can be an effective tool in the development of checks to ensure that GTV autosegmentation methods can be safely introduced in support of patient care.

Deep Learning Architecture to Improve Edge Accuracy of Auto-Contouring for Head and Neck Radiotherapy.

Deep learning (DL) methods have shown great promise in auto-segmentation problems. However, for head and neck cancer, we show that DL methods fail at the axial edges of the gross tumor volume (GTV) where the segmentation is dependent on information closer to the center of the tumor. These failures may decrease trust and usage of proposed auto-contouring systems. To increase performance at the axial edges, we propose the spatially adjusted recurrent convolution U-Net (SARC U-Net). Our method uses convolutional recurrent neural networks and spatial transformer networks to push information from salient regions out to the axial edges. On average, our model increased the Sørensen-Dice coefficient (DSC) at the axial edges of the GTV by 11% inferiorly and 19.3% superiorly over a baseline 2D U-Net, which has no inherent way to capture information between adjacent slices. Over all slices, our proposed architecture achieved a DSC of 0.613, whereas a 3D and 2D U-Net achieved a DSC of 0.586 and 0.540, respectively. SARC U-Net can increase accuracy at the axial edges of GTV contours while also increasing accuracy over baseline models, creating a more robust contour.

Use of CT and MR imaging in radiation therapy planning of imaging-diagnosed canine intracranial meningioma achieves better tumor coverage than CT alone.

The aim of this retrospective, secondary analysis study was to quantify the dosimetric impact of the lack of interobserver agreement on gross tumor volume (GTV) delineation for canine meningioma. This study used a previously reported population of 13 dogs with GTVs contoured on CT alone and on registered CT-MR by 18 radiation oncologists. The "true" GTV was generated for each dog using a simultaneous truth and performance-level estimation algorithm, and "true" brain was defined as the whole brain minus true GTV. Treatment plans were generated for each dog and observer combination, using criteria applied to the observer's GTV and brain contours. Plans were then categorized as a pass (met all planning criteria for true GTV and true brain) or fail. A mixed-effects linear regression was performed to examine differences in metrics between CT and CT-MR plans and mixed-effects logistic regression was performed to examine differences in percentages of pass/fail between CT and CT-MRI plans. The mean percent coverage of true GTV by prescribed dose was higher for CT-MR plans than for CT plans (mean difference 5.9%; 95% CI, 3.7-8.0; P<0.001). There was no difference in the mean volume of true brain receiving ≥24Gy and in maximum true brain dose between CT plans and CT-MR plans (P≥0.198). CT-MR plans were significantly more likely to pass the criteria for true GTV and true brain than CT plans (OR 1.75; 95% CI, 1.02-3.01; P=0.044). This study demonstrated significant dosimetric impact when GTV contouring was performed on CT alone compared with CT-MR.

Accuracy of deformable image registration-based intra-fraction motion management in Magnetic Resonance-guided radiotherapy

Intra-fraction motion management is key in Stereotactic Ablative Radiotherapy (SABR) gated delivery. This study assessed the accuracy of automatic tumor segmentation in the delivery of MR-guided radiotherapy (MRgRT) by comparing it to manual delineations performed by experienced observers. Twenty patients previously treated with MR-guided SABR for thoracic and abdominal tumors were included. Five observers with at least two years of experience in MRgRT manually delineated the gross tumor volume (GTV) for 20 patients on 240 frames of a cine MRI on a sagittal plane. Deformable Image Registration (DIR) based GTV contours were propagated using four different algorithms from a reference frame to subsequent frames.Geometrical analysis based on the Dice Similarity Coefficient (DSC), centroid distance and Hausdorff Distance (HDD) were performed to assess the inter-observer variability and the accuracy of automatic segmentation. A Confidence Value (CV) metric for the reliability of the tumor auto-contouring was also calculated. Inter-observer delineation variability resulted in mean DSC of 0.89, HDD of 5.8mm and centroid distance of 1.7mm. Tumor auto-contouring by the four DIR algorithms resulted in an excellent agreement with the manual delineations by the experienced observers. Mean DSC for each algorithm across all patients wasgreater than0.90, whereas the HDD and centroid distances were below 4.0mm and 1.5mm, respectively. The CV showed a strong correlation with the DSC. DIR-based auto-contouring in MRgRT exhibited a high level of agreement with the manual contouring performed by experts, allowing accurate gated delivery.

Automatic gross tumor segmentation of canine head and neck cancer using deep learning and cross-species transfer learning.

Radiotherapy (RT) is increasingly being used on dogs with spontaneous head and neck cancer (HNC), which account for a large percentage of veterinary patients treated with RT. Accurate definition of the gross tumor volume (GTV) is a vital part of RT planning, ensuring adequate dose coverage of the tumor while limiting the radiation dose to surrounding tissues. Currently the GTV is contoured manually in medical images, which is a time-consuming and challenging task. The purpose of this study was to evaluate the applicability of deep learning-based automatic segmentation of the GTV in canine patients with HNC. Contrast-enhanced computed tomography (CT) images and corresponding manual GTV contours of 36 canine HNC patients and 197 human HNC patients were included. A 3D U-Net convolutional neural network (CNN) was trained to automatically segment the GTV in canine patients using two main approaches: (i) training models from scratch based solely on canine CT images, and (ii) using cross-species transfer learning where models were pretrained on CT images of human patients and then fine-tuned on CT images of canine patients. For the canine patients, automatic segmentations were assessed using the Dice similarity coefficient (Dice), the positive predictive value, the true positive rate, and surface distance metrics, calculated from a four-fold cross-validation strategy where each fold was used as a validation set and test set once in independent model runs. CNN models trained from scratch on canine data or by using transfer learning obtained mean test set Dice scores of 0.55 and 0.52, respectively, indicating acceptable auto-segmentations, similar to the mean Dice performances reported for CT-based automatic segmentation in human HNC studies. Automatic segmentation of nasal cavity tumors appeared particularly promising, resulting in mean test set Dice scores of 0.69 for both approaches. In conclusion, deep learning-based automatic segmentation of the GTV using CNN models based on canine data only or a cross-species transfer learning approach shows promise for future application in RT of canine HNC patients.

Clinical target volume segmentation based on gross tumor volume using deep learning for head and neck cancer treatment.

Accurate clinical target volume (CTV) delineation is important for head and neck intensity-modulated radiation therapy. However, delineation is time-consuming and susceptible to interobserver variability (IOV). Based on a manual contouring process commonly used in clinical practice, we developed a deep learning (DL)-based method to delineate a low-risk CTV with computed tomography (CT) and gross tumor volume (GTV) input and compared it with a CT-only input. A total of 310 patients with oropharynx cancer were randomly divided into the training set (250) and test set (60). The low-risk CTV and primary GTV contours were used to generate label data for the input and ground truth. A 3D U-Net with a two-channel input of CT and GTV (U-NetGTV) was proposed and its performance was compared with a U-Net with only CT input (U-NetCT). The Dice similarity coefficient (DSC) and average Hausdorff distance (AHD) were evaluated. The time required to predict the CTV was 0.86 s per patient. U-NetGTV showed a significantly higher mean DSC value than U-NetCT (0.80 ± 0.03 and 0.76 ± 0.05) and a significantly lower mean AHD value (3.0 ± 0.5 mm vs 3.5 ± 0.7 mm). Compared to the existing DL method with only CT input, the proposed GTV-based segmentation using DL showed a more precise low-risk CTV segmentation for head and neck cancer. Our findings suggest that the proposed method could reduce the contouring time of a low-risk CTV, allowing the standardization of target delineations for head and neck cancer.

Feasibility study of deep learning-based markerless real-time lung tumor tracking with orthogonal X-ray projection images.

The feasibility of a deep learning-based markerless real-time tumor tracking (RTTT) method was retrospectively studied with orthogonal kV X-ray images and clinical tracking records acquired during lung cancer treatment. Ten patients with lung cancer treated with marker-implanted RTTT were included. The prescription dose was 50Gy in four fractions, using seven- to nine-port non-coplanar static beams. This corresponds to 14-18 X-ray tube angles for an orthogonal X-ray imaging system rotating with the gantry. All patients underwent 10 respiratory phases four-dimensional computed tomography. After a data augmentation approach, for each X-ray tube angle of a patient, 2250 digitally reconstructed radiograph (DRR) images with gross tumor volume (GTV) contour labeled were obtained. These images were adopted to train the patient and X-ray tube angle-specific GTV contour prediction model. During the testing, the model trained with DRR images predicted GTV contour on X-ray projection images acquired during treatment. The predicted three-dimensional (3D) positions of the GTV were calculated based on the centroids of the contours in the orthogonal images. The 3D positions of GTV determined by the marker-implanted RTTT during the treatment were considered as the ground truth. The 3D deviations between the prediction and the ground truth were calculated to evaluate the performance of the model. The median GTV volume and motion range were 7.42 (range, 1.18-25.74) cm3 and 22 (range, 11-28) mm, respectively. In total, 8993 3D position comparisons were included. The mean calculation time was 85ms per image. The overall median value of the 3D deviation was 2.27 (interquartile range: 1.66-2.95) mm. The probability of the 3D deviation smaller than 5mm was 93.6%. The evaluation results and calculation efficiency show the proposed deep learning-based markerless RTTT method may be feasible for patients with lung cancer.

Real time volumetric MRI for 3D motion tracking via geometry-informed deep learning.

To develop a geometry-informed deep learning framework for volumetric MRI with sub-second acquisition time in support of 3D motion tracking, which is highly desirable for improved radiotherapy precision but hindered by the long image acquisition time. A 2D-3D deep learning network with an explicitly defined geometry module that embeds geometric priors of the k-space encoding pattern was investigated, where a 2D generation network first augmented the sparsely sampled image dataset by generating new 2D representations of the underlying 3D subject. A geometry module then unfolded the 2D representations to the volumetric space. Finally, a 3D refinement network took the unfolded 3D data and outputted high-resolution volumetric images. Patient-specific models were trained for seven abdominal patients to reconstruct volumetric MRI from both orthogonal cine slices and sparse radial samples. To evaluate the robustness of the proposed method to longitudinal patient anatomy and position changes, we tested the trained model on separate datasets acquired more than one month later and evaluated 3D target motion tracking accuracy using the model-reconstructed images by deforming a reference MRI with gross tumor volume (GTV) contours to a 5-min time series of both ground truth and model-reconstructed volumetric images with a temporal resolution of 340ms. Across the seven patients evaluated, the median distances between model-predicted and ground truth GTV centroids in the superior-inferior direction were 0.4 ± 0.3mm and 0.5 ± 0.4mm for cine and radial acquisitions, respectively. The 95-percentile Hausdorff distances between model-predicted and ground truth GTV contours were 4.7 ± 1.1mm and 3.2 ± 1.5mm for cine and radial acquisitions, which are of the same scale as cross-plane image resolution. Incorporating geometric priors into deep learning model enables volumetric imaging with high spatial and temporal resolution, which is particularly valuable for 3D motion tracking and has the potential of greatly improving MRI-guided radiotherapy precision.

Deep learning based time-to-event analysis with PET, CT and joint PET/CT for head and neck cancer prognosis

ObjectivesRecent studies have shown that deep learning based on pre-treatment positron emission tomography (PET) or computed tomography (CT) is promising for distant metastasis (DM) and overall survival (OS) prognosis in head and neck cancer (HNC). However, lesion segmentation is typically required, resulting in a predictive power susceptible to variations in primary and lymph node gross tumor volume (GTV) segmentation. This study aimed at achieving prognosis without GTV segmentation, and extending single modality prognosis to joint PET/CT to allow investigating the predictive performance of combined- compared to single-modality inputs. MethodsWe employed a 3D-Resnet combined with a time-to-event outcome model to incorporate censoring information. We focused on the prognosis of DM and OS for HNC patients. For each clinical endpoint, five models with PET and/or CT images as input were compared: PET-GTV, PET-only, CT-GTV, CT-only, and PET/CT-GTV models, where -GTV indicates that the corresponding images were masked using the GTV contour. Publicly available delineated CT and PET scans from 4 different Canadian hospitals (293) and the MAASTRO clinic (74) were used for training by 3-fold cross-validation (CV). For independent testing, we used 110 patients from a collaborating institution. The predictive performance was evaluated via Harrell's Concordance Index (HCI) and Kaplan-Meier curves. ResultsIn a 5-year time-to-event analysis, all models could produce CV HCIs with median values around 0.8 for DM and 0.7 for OS. The best performance was obtained with the PET-only model, achieving a median testing HCI of 0.82 for DM and 0.69 for OS. Compared with the PET/CT-GTV model, the PET-only still had advantages of up to 0.07 in terms of testing HCI. The Kaplan-Meier curves and corresponding log-rank test results also demonstrated significant stratification capability of our models for the testing cohort. ConclusionDeep learning-based DM and OS time-to-event models showed predictive capability and could provide indications for personalized RT. The best predictive performance achieved by the PET-only model suggested GTV segmentation might be less relevant for PET-based prognosis.

Conebeam CT‐guided 3D MLC‐based spatially fractionated radiation therapy for bulky masses

For fast, safe, and effective management of large and bulky (≥8 cm) non‐resectable tumors, we have developed a conebeam CT‐guided three‐dimensional (3D)‐conformal MLC‐based spatially fractionated radiation therapy (SFRT) treatment. Using an in‐house MLC‐fitting algorithm, Millennium 120 leaves were fitted to the gross tumor volume (GTV) generating 1‐cm diameter holes at 2‐cm center‐to‐center distance at isocenter. SFRT plans of 15 Gy were generated using four to six coplanar crossfire gantry angles 60° apart with a 90° collimator, differentially weighted with 6‐ or 10‐MV beams. A dose was calculated using AcurosXB algorithm, generating sieve‐like dose channels without post‐processing the physician‐drawn GTV contour within an hour of CT simulation allowing for the same day treatment. In total, 50 extracranial patients have been planned and treated using this method, comprising multiple treatment sites. This novel MLC‐fitting algorithm provided excellent dose parameters with mean GTV (V7.5 Gy) and mean GTV doses of 53.2% and 7.9 Gy, respectively, for 15 Gy plans. Average peak‐to‐valley dose ratio was 3.2. Mean beam‐on time was 3.32 min, and treatment time, including patient setup and CBCT to beam‐off, was within 15 min. Average 3D couch correction from original skin‐markers was <1.0 cm. 3D MLC‐based SFRT plans enhanced target dose for bulky masses, including deep‐seated large tumors while protecting skin and adjacent critical organs. Additionally, it provides the same day, safe, effective, and convenient treatment by eliminating the risk to therapists and patients from heavy gantry‐mounted physical GRID‐block—we recommend other centers to use this simple and clinically useful method. This rapid SFRT planning technique is easily adoptable in any radiation oncology clinic by eliminating the need for plan optimization and patient‐specific quality assurance times while providing dosimetry information in the treatment planning system. This potentially allows for dose‐escalation to deep‐seated masses to debulk unresectable large tumors providing an option for neoadjuvant treatment. An outcome study of clinical trial is underway.

Automatic Delineation of Gross Tumor Volume Based on Magnetic Resonance Imaging by Performing a Novel Semisupervised Learning Framework in Nasopharyngeal Carcinoma

We aimed to validate the accuracy and clinical value of a novel semisupervised learning framework for gross tumor volume (GTV) delineation in nasopharyngeal carcinoma. Two hundred fifty-eight patients with magnetic resonance imaging data sets were divided into training (n = 180), validation (n = 20), and testing (n = 58) cohorts. Ground truth contours of nasopharynx GTV (GTVnx) and node GTV (GTVnd) were manually delineated by 2 experienced radiation oncologists. Twenty percent (n = 36) labeled and 80% (n = 144) unlabeled images were used to train the model, producing model-generated contours for patients from the testing cohort. Nine experienced experts were invited to revise model-generated GTV in 20 randomly selected patients from the testing cohort. Six junior oncologists were asked to delineate GTV in 12 randomly selected patients from the testing cohort without and with the assistance of the model, and revision degrees were compared under these 2 modes. The Dice similarity coefficient (DSC) was used to quantify the accuracy of the model. The model-generated contours showed a high accuracy compared with ground truth contours, with an average DSC score of 0.83 and 0.80 for GTVnx and GTVnd, respectively. There was no significant difference in DSC score between T1-2 and T3-4 patients (0.81 vs 0.83; P = .223), or between N1-2 and N3 patients (0.80 vs 0.79; P = .807). The mean revision degree was lower than 10% in 19 (95%) patients for GTVnx and in 16 (80%) patients for GTVnd. With assistance of the model, the mean revision degree for GTVnx and GTVnd by junior oncologists was reduced from 25.63% to 7.75% and from 21.38% to 14.44%, respectively. Meanwhile, the delineating efficiency was improved by over 60%. The proposed semisupervised learning-based model showed a high accuracy for delineating GTV of nasopharyngeal carcinoma. It was clinically applicable and could assist junior oncologists to improve GTV contouring accuracy and save contouring time.

Deep learning-based GTV contouring modeling inter- and intra- observer variability in sarcomas

Background and purposeThe delineation of the gross tumor volume (GTV) is a critical step for radiation therapy treatment planning. The delineation procedure is typically performed manually which exposes two major issues: cost and reproducibility. Delineation is a time-consuming process that is subject to inter- and intra-observer variability. While methods have been proposed to predict GTV contours, typical approaches ignore variability and therefore fail to utilize the valuable confidence information offered by multiple contours. Materials and methodsIn this work we propose an automatic GTV contouring method for soft-tissue sarcomas from X-ray computed tomography (CT) images, using deep learning by integrating inter- and intra-observer variability in the learned model. Sixty-eight patients with soft tissue and bone sarcomas were considered in this evaluation, all underwent pre-operative CT imaging used to perform GTV delineation. Four radiation oncologists and radiologists performed three contouring trials each for all patients. We quantify variability by defining confidence levels based on the frequency of inclusion of a given voxel into the GTV and use a deep convolutional neural network to learn GTV confidence maps. ResultsResults were compared to confidence maps from the four readers as well as ground-truth consensus contours established jointly by all readers. The resulting continuous Dice score between predicted and true confidence maps was 87% and the Hausdorff distance was 14 mm. ConclusionResults demonstrate the ability of the proposed method to predict accurate contours while utilizing variability and as such it can be used to improve clinical workflow.

Real Time Volumetric MRI Using a Physics-Aware Deep Neural Network for MR-Guided Radiation Therapy

To investigate the feasibility of real time volumetric MRI using a physics-aware deep neural network, in support of MRI-based 3D motion tracking for MR-guided adaptive radiotherapy.A deep neural network was proposed to generate volumetric MRI from orthogonal Cine MRI, which can be acquired on MR-Linac systems real time. The network consisted of a 2D generation module where oblique slices were synthesized from two orthogonal MRI slices and a 3D refinement module to generate final high quality volumetric images. A physics module that utilized prior knowledge of imaging physics was used to bridge 2D and 3D domains. The proposed method was evaluated on four patients with intrahepatic carcinoma. Each patient received two 4D MRI examinations that were more than 1 month apart, where one examination was used for network training and another for evaluation. Volumetric images of 21 motion phases were reconstructed via retrospective binning and served as the ground truth. Orthogonal Cine MRI were sampled from volumetric MRI and served as network inputs. To evaluate tracking accuracy using the network predicted volumetric MRI, a reference MRI with gross tumor volume (GTV) contours was deformed to match each of the 21 motion phases. The deformed GTV centroid positions and contours were compared between ground truth and network prediction.Across all the 4 patients evaluated, the averaged distance between predicted and ground truth GTV centroids was less than 1 mm in all 3 directions (anterior-posterior, inferior-superior and lateral). The averaged 95-percentile Hausdorff distances between the predicted and ground truth GTV contours was 4.2 mm, which was similar to the cross-plane imaging resolution (4 mm).Physics-aware deep neural network enabled real time volumetric MRI with sufficient accuracy to support 3D motion tracking. The proposed technique is robust to longitudinal patient position and configuration changes, and has the potential of reducing treatment margins and improving treatment delivery precision on a MR-Linac system.

CT-Guided Interstitial Low Dose-RATE Brachytherapy for Recurrent Colorectal Cancer

Patients with pelvic or retroperitoneal recurrences of colorectal cancer (CRC) have poor overall survival, with retrospective studies suggesting a possible survival benefit with local therapy. However, many patients present with disease that is either unresectable, or in which resection will result in significant morbidity or margin positivity. We sought to examine the local control, overall survival (OS), and toxicity of minimally invasive CT-guided low dose-rate (LDR) brachytherapy in this setting, for which there is limited published data.This study was a retrospective analysis of all patients at our institution who have undergone CT-guided LDR brachytherapy for recurrent CRC. Patient, tumor, and treatment characteristics were collected, along with physician-reported symptoms in the medical record. Patients underwent dose calculation based on pre-implantation CT imaging and contouring of gross tumor volume (GTV), followed by CT-guided percutaneous interstitial needle placement of either Iodine-125 or Palladium-103 permanent seed implants prescribed to the GTV. Response Evaluation Criteria in Solid Tumors (RECIST) criteria were used to interpret treatment response, and the Kaplan-Meier method was used to estimate rates of overall survival (OS) and in-field control (IFC). IFC was defined as a lack of progression of disease within the pre-implantation defined target volume.Between December 2016 and August 2020, 13 patients underwent 18 CT-guided LDR seed implantations (6 I-125, 12 Pd-103) for pelvic or retroperitoneal recurrences of CRC at our institution, with a median follow-up of 17 months. 17 treatments were for pelvic recurrences, while one treatment was for a retrocaval nodal recurrence. The median prescribed dose to the target volume was 80 Gy (range 40-110 Gy), with a median target volume of 5.0 cc (range 0.6-25.5 cc). The 1-year estimated rates of OS and IFC were 94.4% (CI 84.4-100%) and 82.4% (CI 66.1-100%), respectively. All in-field failures occurred in targets above the median target volume of 5 cc. All patients previously received definitive treatment for either initially locally advanced (8 patients) or limited metastatic disease (5 patients). Of the 18 implantations, four were performed for isolated pelvic recurrences and 13 for either oligo-progressive or oligo-recurrent metastatic disease, with a median of two active sites of disease at the time of seed implantation. One patient received palliative brachytherapy after a poly-metastatic recurrence. No patients experienced grade 2 or higher toxicities. Three patients experienced a grade 1 toxicity (proctitis, pain, and fatigue), and one patient developed an asymptomatic recto-vaginal fistula detected on imaging, which was outside of the brachytherapy field.CT-guided interstitial LDR brachytherapy achieved high rates of control and overall survival with minimal long-term toxicity in patients with pelvic and retroperitoneal recurrences of CRC.

A Generalizable Guided Deep Learning Auto-Segmentation Method of Pancreatic GTV on Multi-Protocol Daily MRIs for MR-Guided Adaptive Radiotherapy

Fast and consistent delineation of gross-tumor volume (GTV) on non-contrast enhanced daily MRI is critical but can be challenging in MR-guided Online Adaptive Radiation Therapy (MRgOART). Our previous works have demonstrated the potential of using deep learning (DL) method for pancreatic GTV segmentation on MRI of a single sequence. However, the application of the obtained model is limited by the variations of MRIs from different scan protocols. To overcome the protocol dependence, we propose a generalizable prior knowledge guided DL method for pancreatic GTV auto-segmentation using the guidance of patient-specific contour from the verification MRI (obtained in the treatment room before the first fraction) on subsequent multi-protocol daily MRIs.The daily MRIs acquired from 21 pancreatic cancer patients on a 1.5T MR-Linac during MRgOART were used to demonstrate the method. Two different scan protocols were used: T2 TSE sequence for nine patients and 4D Vane sequence for the remaining twelve patients. Manual contours carefully delineated on each MRI by treating physicians were utilized as ground truth. All the MRIs were pre-processed (e.g., bias corrected, normalized, cropped, and resampled). Each daily MRI was rigidly registered to the verification MRI and a total of 99 image pairs were generated. The DL network backbone used was UNet with 3 encoders interconnected with 3 decoders and a classification block. Model input was the composite of verification MRI, verification GTV mask, and daily MRI. Data augmentation methods such as translation, rotation, flip, guidance-target inversion, and rigid registration error introduction were applied for robust model training. Patient-based leave-one-out-cross-validation was employed and the model performance was evaluated using Dice similarity coefficient (DSC) compared with the ground truth.The average DSC was 0.63 ± 0.18 for all the cases studied. For the cases with GTV > 5cc, the average DSC was 0.76 ± 0.13 and 0.69 ± 0.08 for T2 TSE and 4D Vane MRIs, respectively. The results are reasonably comparable to the inter-observer variation of DSC = 0.71 estimated in a previous study. The time required to generate GTV contour on a daily MRI set was less than 1s on a hardware with an I7 CPU and a GTX 1060 GPU.The proposed prior-knowledge guided deep leaning auto-segmentation method for pancreatic GTV on daily multi-protocol non-contrast MRIs performed reasonably well even with the relatively small dataset. This proof-of-principle study suggests that, with patient-specific prior knowledge guidance, it is feasible to implement a single DL auto-segmentation model for multiple MRI sequences in MRgOART. The method can be generalized for other structures or tumors. (First and second authors contributed equally.).