Search

International audience; This paper presents an overview of the third edition of the HEad and neCK TumOR segmentation and outcome prediction (HECKTOR) challenge, organized as a satellite event of the 25th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI) 2022. The challenge comprises two tasks related to the automatic analysis of FDG-PET/CT images for patients with Head and Neck cancer (H &N), focusing on the oropharynx region. Task 1 is the fully automatic segmentation of H &N primary Gross Tumor Volume (GTVp) and metastatic lymph nodes (GTVn) from FDG-PET/CT images. Task 2 is the fully automatic prediction of Recurrence-Free Survival (RFS) from the same FDG-PET/CT and clinical data. The data were collected from nine centers for a total of 883 cases consisting of FDG-PET/CT images and clinical information, split into 524 training and 359 test cases. The best methods obtained an aggregated Dice Similarity Coefficient (DSCagg) of 0.788 in Task 1, and a Concordance index (C-index) of 0.682 in Task 2.

Background and Purpose: Conventional magnetic resonance imaging (MRI) poses challenges in quantitative analysis because voxel intensity values lack physical meaning. While intensity standardization methods exist, their effects on head and neck MRI have not been investigated. We developed a workflow based on healthy tissue region of interest (ROI) analysis to determine intensity consistency within a patient cohort. Through this workflow, we systematically evaluated intensity standardization methods for MRI of head and neck cancer (HNC) patients. Materials and Methods: Two HNC cohorts (30 patients total) were retrospectively analyzed. One cohort was imaged with heterogenous acquisition parameters (HET cohort), whereas the other was imaged with homogenous acquisition parameters (HOM cohort). The standard deviation of cohort-level normalized mean intensity (SD NMIc), a metric of intensity consistency, was calculated across ROIs to determine the effect of five intensity standardization methods on T2-weighted images. For each cohort, a Friedman test followed by a post-hoc Bonferroni-corrected Wilcoxon signed-rank test was conducted to compare SD NMIc among methods. Results: Consistency (SD NMIc across ROIs) between unstandardized images was substantially more impaired in the HET cohort (0.29 ± 0.08) than in the HOM cohort (0.15 ± 0.03). Consequently, corrected p-values for intensity standardization methods with lower SD NMIc compared to unstandardized images were significant in the HET cohort (p 0.05). In both cohorts, differences between methods were often minimal and nonsignificant. Conclusions: Our findings stress the importance of intensity standardization, either through the utilization of uniform acquisition parameters or specific intensity standardization methods, and the need for testing intensity consistency before performing quantitative analysis of HNC MRI.

While there has been an overall decline of tobacco and alcohol-related head and neck cancer in recent decades, there has been an increased incidence of HPV-associated oropharyngeal cancer (OPC). Recent research studies and clinical trials have revealed that the cancer biology and clinical progression of HPV-positive OPC is unique relative to its HPV-negative counterparts. HPV-positive OPC is associated with higher rates of disease control following definitive treatment when compared to HPV-negative OPC. Thus, these conditions should be considered unique diseases with regards to treatment strategies and survival. In order to sufficiently characterize HPV-positive OPC and guide treatment strategies, there has been a considerable effort to diagnose, prognose, and track the treatment response of HPV-associated OPC through advanced imaging research. Furthermore, HPV-positive OPC patients are prime candidates for radiation de-escalation protocols, which will ideally reduce toxicities associated with radiation therapy and has prompted additional imaging research to detect radiation-induced changes in organs at risk. This manuscript reviews the various imaging modalities and current strategies for tackling these challenges as well as provides commentary on the potential successes and suggested improvements for the optimal treatment of these tumors.

PurposeParotid whole-gland magnetic resonance (MR) T1 intensity, thresholded at the 90th percentile (T1 P90), has been previously reported to be a candidate MR imaging biomarker (MR-IBM) for improved prediction of xerostomia development after radiotherapy. Although P90 was previously derived from the parotid glands of T1-weighted MRI, in this study, we aim to validate P90 in an external cohort using fat only images reconstructed from a T1 Dixon MRI sequence, as well as determining alternative T1 intensity thresholds for potential qualification as predictive FDA BEST biomarkers of xerostomia development 6 months after radiotherapy (Xero6m).MethodsMR-IBMs derived from T1 Dixon intensity-normalized scans from 76 head and neck cancer (HNC) patients were extracted from pre-treatment MR images. Scans were normalized to fat tissue, and imaging characteristics were quantified. A reference model and MR-IBM models were created using multivariable logistic regression to predict Xero6m. External validation was performed using the model coefficients described in a previous study. The area under the curve (AUC) of the resulting models were compared. Stepwise forward feature selection was performed to discover additional MR-IBMs for improved predictions of xerostomia.ResultsThe external validation of a previous model coefficients against our cohort showed decreased performance of the P90 MR-IBM model (AUC of 0.73 (CI 0.61-0.85)). The reference model exhibited improved performance when P90 was incorporated (AUC of 0.78 (CI 0.67-0.89)). Feature selection demonstrated the P10 MR-IBM provided performance improvements (AUC of 0.79 (CI: 0.69-0.90)).ConclusionOur findings validated P90 as predictive biomarker for radiation-induced xerostomia and showed MR-IBMs derived from Dixon sequences can improve Xero6m prediction when compared to the reference model. Formal biomarker qualification should be considered for T1 sequences/relaxometry via formalized approaches.

Deep learning is becoming increasingly popular and available to new users, particularly in the medical field. Deep learning image segmentation, outcome analysis, and generators rely on presentation of Digital Imaging and Communications in Medicine (DICOM) images and often radiation therapy (RT) structures as masks. Although the technology to convert DICOM images and RT structures into other data types exists, no purpose-built Python module for converting NumPy arrays into RT structures exists. The 2 most popular deep learning libraries, Tensorflow and PyTorch, are both implemented within Python, and we believe a set of tools built in Python for manipulating DICOM images and RT structures would be useful and could save medical researchers large amounts of time and effort during the preprocessing and prediction steps. Our module provides intuitive methods for rapid data curation of RT-structure files by identifying unique region of interest (ROI) names and ROI structure locations and allowing multiple ROI names to represent the same structure. It is also capable of converting DICOM images and RT structures into NumPy arrays and SimpleITK Images, the most commonly used formats for image analysis and inputs into deep learning architectures and radiomic feature calculations. Furthermore, the tool provides a simple method for creating a DICOM RT-structure from predicted NumPy arrays, which are commonly the output of semantic segmentation deep learning models. Accessing DicomRTTool via the public Github project invites open collaboration, and the deployment of our module in PyPi ensures painless distribution and installation. We believe our tool will be increasingly useful as deep learning in medicine progresses.

Highlights • Accurate biomarkers of immunotherapy treatment response are needed. • Radiomic features are emerging biomarkers of tumor immune biology and immunotherapy response. • Delta radiomics may be critical for predicting immunotherapy response. • Lower-dimensional feature associations are more robust than for textural features. • Composite models features and other variables show greater predictive power., Immunotherapies are leading to improved outcomes for many cancers, including those with devastating prognoses. As therapies like immune checkpoint inhibitors (ICI) become a mainstay in treatment regimens, many concurrent challenges have arisen – for instance, delineating clinical responders from non-responders. Predicting response has proven to be difficult given a lack of consistent and accurate biomarkers, heterogeneity of the tumor microenvironment (TME), and a poor understanding of resistance mechanisms. For the most part, imaging data have remained an untapped, yet abundant, resource to address these challenges. In recent years, quantitative image analyses have highlighted the utility of medical imaging in predicting tumor phenotypes, prognosis, and therapeutic response. These studies have been fueled by an explosion of resources in high-throughput mining of image features (i.e. radiomics) and artificial intelligence. In this review, we highlight current progress in radiomics to understand tumor immune biology and predict clinical responses to immunotherapies. We also discuss limitations in these studies and future directions for the field, particularly if high-dimensional imaging data are to play a larger role in precision medicine.

BackgroundPain is a common, debilitating symptom experienced by patients with oral cavity and oropharyngeal cancer (OC/OPC) treated with radiotherapy (RT). Managing acute pain (AP) over 6 - 7 weeks of RT remains a significant challenge, warranting further investigation. Using a modern prospective registry, the objective of this study was to characterize longitudinal AP profiles and temporal changes in vital signs (VS), radiation toxicities, and analgesic prescribing patterns during RT.MethodsA total of 351 patients with OC (n=120) and OPC (n=228) treated with curative RT from 2013-2021 were included. Baseline cohort characteristics, weekly patient-reported pain descriptors, physician-graded toxicities (CTCAE v5), and analgesic orders during RT were extracted. Temporal changes in AP scores and VS were analyzed using linear mixed effect models. AP trajectories were reduced to single metric area under the curve calculations (AUCpain). Correlations were assessed using Spearman correlation coefficients.ResultsMedian age was 60 years, and 70% and 42% received chemotherapy and surgery, respectively. A significant increase in pain, mucositis, dermatitis, and overall treatment toxicity severity were observed by the end of RT. AUCpain was significantly different based on gender, primary tumor site, surgery, drug use history and pre-RT pain. There was a temporal mean weight loss of 7.1% bodyweight (95%CI, 10-8.2; PPPP.001) while weight and MAP were negatively associated over time (PConclusionThis study characterizes longitudinal treatment-related toxicity kinetics using a prospective OC/OPC registry and demonstrates an ongoing need for optimized, timely pain control. Pain AUC metrics preserve temporal information and may be useful for developing algorithmic pain prediction and management models.

Purpose/Objective(s)To determine diffusion-weighted imaging (DWI) MRI parameters associated with tumor response and oncologic outcomes in head and neck (HNC) patients treated with definitive radiation therapy (RT).Materials/MethodsEighty-six HNC patients enrolled in an active prospective imaging study at The University of Texas MD Anderson Cancer Center were included in the analysis. Patients had MRIs pre-, mid-, and post-RT completion. Inclusion criteria included adults with histologic evidence of malignant head and neck neoplasm indicated for curative-intent treatment with RT with/without chemotherapy, good performance status (ECOG score 0-2), and with no contraindications to MRI. Patients were scanned using a MAGNETOM Aera 1.5T MR scanner (Siemens Healthcare, Germany) with two large four-channel flex phased-array coils. We used fat-suppressed T2-weighted turbo spin echo sequences for tumor segmentation which were co-registered to respective DWIs for extraction of apparent diffusion coefficient (ADC) measurements. Treatment response was assessed at mid-RT and at 8-12 weeks post-RT using RECIST 1.1 criteria and was defined as: complete response (CR) vs. non-complete response (non-CR). Pre-RT ADC was correlated with RT response (CR vs. non-CR) at mid- and post-RT. The Mann-Whitney U test was used to compare ADC values between the mid-treatment CR group and the non-CR group. Recursive partitioning analysis (RPA) was performed to identify ADC threshold associated with relapse. Cox proportional hazards models were done for clinical vs. clinical and imaging parameters and internal validation was done using bootstrapping technique.ResultsEighty-one patients were included in this analysis. Median follow-up was 31 months. Pre-treatment ADC was not correlated with tumor response or oncologic outcomes (P>0.05). For patients with post-RT CR, there was a significant increase in mean ADC at mid-RT compared to baseline ((1.8 ± 0.29) × 10−3 mm2/s versus (1.37 ± 0.22) × 10− 3 mm2/s, p < 0.0001), while patients with non-CR had no statistically significant increase (p >0.05). RPA identified GTV-P delta (Δ) ADCmean p=0.01). Univariable and multivariable analysis of prognostic outcomes showed that GTV-P ΔADCmean at mid-RT ≥7% was significantly associated with better LC and RFS. The addition of ΔADCmean significantly improved the c-indices of LC and RFS models compared with standard clinical variables (0.85 vs. 0.77 and 0.74 vs. 0.68 for LC and RFS, respectively, pConclusionADC change at mid-RT is a strong predictor of oncologic outcomes in HNC patients. Patients with no significant increase of primary tumor site ADC at mid-RT relative to baseline values are at high risk of disease relapse. Multi-institutional data are needed for validation of our results.

Auto-segmentation of primary tumors in oropharyngeal cancer using PET/CT images is an unmet need that has the potential to improve radiation oncology workflows. In this study, we develop a series of deep learning models based on a 3D Residual Unet (ResUnet) architecture that can segment oropharyngeal tumors with high performance as demonstrated through internal and external validation of large-scale datasets (training size = 224 patients, testing size = 101 patients) as part of the 2021 HECKTOR Challenge. Specifically, we leverage ResUNet models with either 256 or 512 bottleneck layer channels that demonstrate internal validation (10-fold cross-validation) mean Dice similarity coefficient (DSC) up to 0.771 and median 95% Hausdorff distance (95% HD) as low as 2.919 mm. We employ label fusion ensemble approaches, including Simultaneous Truth and Performance Level Estimation (STAPLE) and a voxel-level threshold approach based on majority voting (AVERAGE), to generate consensus segmentations on the test data by combining the segmentations produced through different trained cross-validation models. We demonstrate that our best performing ensembling approach (256 channels AVERAGE) achieves a mean DSC of 0.770 and median 95% HD of 3.143 mm through independent external validation on the test set. Our DSC and 95% HD test results are within 0.01 and 0.06 mm of the top ranked model in the competition, respectively. Concordance of internal and external validation results suggests our models are robust and can generalize well to unseen PET/CT data. We advocate that ResUNet models coupled to label fusion ensembling approaches are promising candidates for PET/CT oropharyngeal primary tumors auto-segmentation. Future investigations should target the ideal combination of channel combinations and label fusion strategies to maximize segmentation performance.

PET/CT images provide a rich data source for clinical prediction models in head and neck squamous cell carcinoma (HNSCC). Deep learning models often use images in an end-to-end fashion with clinical data or no additional input for predictions. However, in the context of HNSCC, the tumor region of interest may be an informative prior in the generation of improved prediction performance. In this study, we utilize a deep learning framework based on a DenseNet architecture to combine PET images, CT images, primary tumor segmentation masks, and clinical data as separate channels to predict progression-free survival (PFS) in days for HNSCC patients. Through internal validation (10-fold cross-validation) based on a large set of training data provided by the 2021 HECKTOR Challenge, we achieve a mean C-index of 0.855 ± 0.060 and 0.650 ± 0.074 when observed events are and are not included in the C-index calculation, respectively. Ensemble approaches applied to cross-validation folds yield C-index values up to 0.698 in the independent test set (external validation), leading to a 1st place ranking on the competition leaderboard. Importantly, the value of the added segmentation mask is underscored in both internal and external validation by an improvement of the C-index when compared to models that do not utilize the segmentation mask. These promising results highlight the utility of including segmentation masks as additional input channels in deep learning pipelines for clinical outcome prediction in HNSCC.

PET/CT images provide a rich data source for clinical prediction models in head and neck squamous cell carcinoma (HNSCC). Deep learning models often use images in an end-to-end fashion with clinical data or no additional input for predictions. However, in the context of HNSCC, the tumor region of interest may be an informative prior in the generation of improved prediction performance. In this study, we utilize a deep learning framework based on a DenseNet architecture to combine PET images, CT images, primary tumor segmentation masks, and clinical data as separate channels to predict progression-free survival (PFS) in days for HNSCC patients. Through internal validation (10-fold cross-validation) based on a large set of training data provided by the 2021 HECKTOR Challenge, we achieve a mean C-index of 0.855 ± 0.060 and 0.650 ± 0.074 when observed events are and are not included in the C-index calculation, respectively. Ensemble approaches applied to cross-validation folds yield C-index values up to 0.698 in the independent test set (external validation), leading to a 1(st) place ranking on the competition leaderboard. Importantly, the value of the added segmentation mask is underscored in both internal and external validation by an improvement of the C-index when compared to models that do not utilize the segmentation mask. These promising results highlight the utility of including segmentation masks as additional input channels in deep learning pipelines for clinical outcome prediction in HNSCC.

Highlights • mpMRI coupled to deep learning can generate reasonable OPC tumor segmentations. • Using multiple input channels may positively impact segmentation performance. • Deep learning segmentations were non-inferior to ground truth as per Turing test., Background/Purpose Oropharyngeal cancer (OPC) primary gross tumor volume (GTVp) segmentation is crucial for radiotherapy. Multiparametric MRI (mpMRI) is increasingly used for OPC adaptive radiotherapy but relies on manual segmentation. Therefore, we constructed mpMRI deep learning (DL) OPC GTVp auto-segmentation models and determined the impact of input channels on segmentation performance. Materials/Methods GTVp ground truth segmentations were manually generated for 30 OPC patients from a clinical trial. We evaluated five mpMRI input channels (T2, T1, ADC, Ktrans, Ve). 3D Residual U-net models were developed and assessed using leave-one-out cross-validation. A baseline T2 model was compared to mpMRI models (T2 + T1, T2 + ADC, T2 + Ktrans, T2 + Ve, all five channels [ALL]) primarily using the Dice similarity coefficient (DSC). False-negative DSC (FND), false-positive DSC, sensitivity, positive predictive value, surface DSC, Hausdorff distance (HD), 95% HD, and mean surface distance were also assessed. For the best model, ground truth and DL-generated segmentations were compared through a blinded Turing test using three physician observers. Results Models yielded mean DSCs from 0.71 ± 0.12 (ALL) to 0.73 ± 0.12 (T2 + T1). Compared to the T2 model, performance was significantly improved for FND, sensitivity, surface DSC, HD, and 95% HD for the T2 + T1 model (p 0.05). Most models demonstrated significant correlations between tumor size and HD or Surface DSC (p 0.05). On average, there were no significant differences between ground truth and DL-generated segmentations for all observers (p > 0.05). Conclusion DL using mpMRI provides reasonably accurate segmentations of OPC GTVp that may be comparable to ground truth segmentations generated by clinical experts. Incorporating additional mpMRI channels may increase the performance of FND, sensitivity, surface DSC, HD, and 95% HD, and improve model robustness to tumor size.

Determining progression-free survival (PFS) for head and neck squamous cell carcinoma (HNSCC) patients is a challenging but pertinent task that could help stratify patients for improved overall outcomes. PET/CT images provide a rich source of anatomical and metabolic data for potential clinical biomarkers that would inform treatment decisions and could help improve PFS. In this study, we participate in the 2021 HECKTOR Challenge to predict PFS in a large dataset of HNSCC PET/CT images using deep learning approaches. We develop a series of deep learning models based on the DenseNet architecture using a negative log-likelihood loss function that utilizes PET/CT images and clinical data as separate input channels to predict PFS in days. Internal model validation based on 10-fold cross-validation using the training data (N = 224) yielded C-index values up to 0.622 (without) and 0.842 (with) censoring status considered in C-index computation, respectively. We then implemented model ensembling approaches based on the training data cross-validation folds to predict the PFS of the test set patients (N = 101). External validation on the test set for the best ensembling method yielded a C-index value of 0.694, placing 2(nd) in the competition. Our results are a promising example of how deep learning approaches can effectively utilize imaging and clinical data for medical outcome prediction in HNSCC, but further work in optimizing these processes is needed.

Background: Personalised radiotherapy can improve treatment outcomes of patients with head and neck cancer (HNC), where currently a ‘one-dose-fits-all’ approach is the standard. The aim was to establish individualised outcome prediction based on multi-institutional international ‘big-data’ to facilitate risk-based stratification of patients with HNC. Methods: The data of 4611 HNC radiotherapy patients from three academic cancer centres were split into four cohorts: a training (n = 2241), independent test (n = 786), and external validation cohorts 1 (n = 1087) and 2 (n = 497). Tumour- and patient-related clinical variables were considered in a machine learning pipeline to predict overall survival (primary end-point) and local and regional tumour control (secondary end-points); serially, imaging features were considered for optional model improvement. Finally, patients were stratified into high-, intermediate-, and low-risk groups. Results: Performance score, AJCC8th stage, pack-years, and Age were identified as predictors for overall survival, demonstrating good performance in both the training cohort (c-index = 0.72 [95% CI, 0.66–0.77]) and in all three validation cohorts (c-indices: 0.76 [0.69–0.83], 0.73 [0.68–0.77], and 0.75 [0.68–0.80]). Excellent stratification of patients with HNC into high, intermediate, and low mortality risk was achieved; with 5-year overall survival rates of 17–46% for the high-risk group compared to 92–98% for the low-risk group. The addition of morphological image feature further improved the performance (c-index = 0.73 [0.64–0.81]). These models are integrated in a clinic-ready interactive web interface: https://uic-evl.github.io/hnc-predictor/ Conclusions: Robust model-based prediction was able to stratify patients with HNC in distinct high, intermediate, and low mortality risk groups. This can effectively be capitalised for personalised radiotherapy, e.g., for tumour radiation dose escalation/de-escalation.

Auto-segmentation of primary tumors in oropharyngeal cancer using PET/CT images is an unmet need that has the potential to improve radiation oncology workflows. In this study, we develop a series of deep learning models based on a 3D Residual Unet (ResUnet) architecture that can segment oropharyngeal tumors with high performance as demonstrated through internal and external validation of large-scale datasets (training size = 224 patients, testing size = 101 patients) as part of the 2021 HECKTOR Challenge. Specifically, we leverage ResUNet models with either 256 or 512 bottleneck layer channels that are able to demonstrate internal validation (10-fold cross-validation) mean Dice similarity coefficient (DSC) up to 0.771 and median 95% Hausdorff distance (95% HD) as low as 2.919 mm. We employ label fusion ensemble approaches, including Simultaneous Truth and Performance Level Estimation (STAPLE) and a voxel-level threshold approach based on majority voting (AVERAGE), to generate consensus segmentations on the test data by combining the segmentations produced through different trained cross-validation models. We demonstrate that our best performing ensembling approach (256 channels AVERAGE) achieves a mean DSC of 0.770 and median 95% HD of 3.143 mm through independent external validation on the test set. Concordance of internal and external validation results suggests our models are robust and can generalize well to unseen PET/CT data. We advocate that ResUNet models coupled to label fusion ensembling approaches are promising candidates for PET/CT oropharyngeal primary tumors auto-segmentation, with future investigations targeting the ideal combination of channel combinations and label fusion strategies to maximize segmentation performance.

Determining progression-free survival (PFS) for head and neck squamous cell carcinoma (HNSCC) patients is a challenging but pertinent task that could help stratify patients for improved overall outcomes. PET/CT images provide a rich source of anatomical and metabolic data for potential clinical biomarkers that would inform treatment decisions and could help improve PFS. In this study, we participate in the 2021 HECKTOR Challenge to predict PFS in a large dataset of HNSCC PET/CT images using deep learning approaches. We develop a series of deep learning models based on the DenseNet architecture using a negative log-likelihood loss function that utilizes PET/CT images and clinical data as separate input channels to predict PFS in days. Internal model validation based on 10-fold cross-validation using the training data (N=224) yielded C-index values up to 0.622 (without) and 0.842 (with) censoring status considered in C-index computation, respectively. We then implemented model ensembling approaches based on the training data cross-validation folds to predict the PFS of the test set patients (N=101). External validation on the test set for the best ensembling method yielded a C-index value of 0.694. Our results are a promising example of how deep learning approaches can effectively utilize imaging and clinical data for medical outcome prediction in HNSCC, but further work in optimizing these processes is needed.

PurposeIn order to accurately accumulate delivered dose for head and neck cancer patients treated with the Adapt to Position workflow on the 1.5T magnetic resonance imaging (MRI)-linear accelerator (MR-linac), the low-resolution T2-weighted MRIs used for daily setup must be segmented to enable reconstruction of the delivered dose at each fraction. In this study, our goal is to evaluate various autosegmentation methods for head and neck organs at risk (OARs) on on-board setup MRIs from the MR-linac for off-line reconstruction of delivered dose.MethodsSeven OARs (parotid glands, submandibular glands, mandible, spinal cord, and brainstem) were contoured on 43 images by seven observers each. Ground truth contours were generated using a simultaneous truth and performance level estimation (STAPLE) algorithm. 20 autosegmentation methods were evaluated in ADMIRE: 1-9) atlas-based autosegmentation using a population atlas library (PAL) of 5/10/15 patients with STAPLE, patch fusion (PF), random forest (RF) for label fusion; 10-19) autosegmentation using images from a patient’s 1-4 prior fractions (individualized patient prior (IPP)) using STAPLE/PF/RF; 20) deep learning (DL) (3D ResUNet trained on 43 ground truth structure sets plus 45 contoured by one observer). Execution time was measured for each method. Autosegmented structures were compared to ground truth structures using the Dice similarity coefficient, mean surface distance, Hausdorff distance, and Jaccard index. For each metric and OAR, performance was compared to the inter-observer variability using Dunn’s test with control. Methods were compared pairwise using the Steel-Dwass test for each metric pooled across all OARs. Further dosimetric analysis was performed on three high-performing autosegmentation methods (DL, IPP with RF and 4 fractions (IPP_RF_4), IPP with 1 fraction (IPP_1)), and one low-performing (PAL with STAPLE and 5 atlases (PAL_ST_5)). For five patients, delivered doses from clinical plans were recalculated on setup images with ground truth and autosegmented structure sets. Differences in maximum and mean dose to each structure between the ground truth and autosegmented structures were calculated and correlated with geometric metrics.ResultsDL and IPP methods performed best overall, all significantly outperforming inter-observer variability and with no significant difference between methods in pairwise comparison. PAL methods performed worst overall; most were not significantly different from the inter-observer variability or from each other. DL was the fastest method (33 seconds per case) and PAL methods the slowest (3.7 – 13.8 minutes per case). Execution time increased with number of prior fractions/atlases for IPP and PAL. For DL, IPP_1, and IPP_RF_4, the majority (95%) of dose differences were within ±250 cGy from ground truth, but outlier differences up to 785 cGy occurred. Dose differences were much higher for PAL_ST_5, with outlier differences up to 1920 cGy. Dose differences showed weak but significant correlations with all geometric metrics (R2 between 0.030 and 0.314).ConclusionsThe autosegmentation methods offering the best combination of performance and execution time are DL and IPP_1. Dose reconstruction on on-board T2-weighted MRIs is feasible with autosegmented structures with minimal dosimetric variation from ground truth, but contours should be visually inspected prior to dose reconstruction in an end-to-end dose accumulation workflow.

Background and PurposeOropharyngeal cancer (OPC) primary gross tumor volume (GTVp) segmentation is crucial for radiotherapy. Multiparametric MRI (mpMRI) is increasingly used for OPC adaptive radiotherapy but relies on manual segmentation. Therefore, we constructed mpMRI deep learning (DL) OPC GTVp auto-segmentation models and determined the impact of input channels on segmentation performance.Materials and MethodsGTVp ground truth segmentations were manually generated for 30 OPC patients from a clinical trial. We evaluated five mpMRI input channels (T2, T1, ADC, Ktrans, Ve). 3D Residual U-net models were developed and assessed using leave-one-out cross-validation. A baseline T2 model was compared to mpMRI models (T2+T1, T2+ADC, T2+Ktrans, T2+Ve, all 5 channels [ALL]) primarily using the Dice similarity coefficient (DSC). Sensitivity, positive predictive value, Hausdorff distance (HD), false-negative DSC (FND), false-positive DSC, surface DSC, 95% HD, and mean surface distance were also assessed. For the best model, ground truth and DL-generated segmentations were compared through a Turing test using physician observers.ResultsModels yielded mean DSCs from 0.71 (ALL) to 0.73 (T2+T1). Compared to the T2 model, performance was significantly improved for HD, FND, sensitivity, surface DSC, and 95% HD for the T2+T1 model (p0.05).ConclusionDL using mpMRI provides high-quality segmentations of OPC GTVp. Incorporating additional mpMRI channels may increase the performance of certain evaluation metrics. This pilot study is a promising step towards fully automated MR-guided OPC radiotherapy.

Purpose To introduce a contouring guideline for the taste bud bearing tongue mucosa for head and neck cancer patients receiving radiotherapy. Methods and Materials CT simulation images of oropharyngeal cancer patients were used to delineate both the whole tongue (extrinsic/intrinsic tongue muscles, floor of mouth) and the taste bud bearing tongue mucosa (method A: adaptation of the whole tongue structure; method B: axial adaptation of a mid-sagittal contour). Volumetric and dosimetric parameters of the whole tongue and the two methods of mucosal delineation, spatial overlap between methods A and B, and inter-observer variability for method B were calculated. Results The study cohort was comprised of 70 patients with T1–4 N0–1 tonsillar (83%) and base of tongue (17%) cancers. Most of the comparative parameters between the whole tongue and mucosa (method A) significantly differed (mean, minimum, and maximum dose, V5–V70, D40–D90). The mean dose calculated for the whole tongue deviated on average 3.77 Gy compared to method A. No significant differences were found between methods A and B of the taste bud bearing tongue mucosa structure, and none of the dosimetric parameters differed more than 1.03 Gy on average. The mean Dice similarity coefficient for both mucosal structures was 0.79 ± 0.05, and 0.63 ± 0.12 for the inter-observer analysis of method B. Conclusions We defined two methods for delineating the taste bud bearing mucosa and both are equally satisfactory procedures. Either method is preferable over delineation of the whole tongue as organ at risk for taste impairment.

Earlier research indicated that nearly 20% of patients diagnosed with either bipolar disorder (BD) or borderline personality disorder (BPD) also met criteria for the other diagnosis. Yet limited data are available concerning the potential impact of co-occurring BPD and/or BPD features on the course or outcome in patients with BD. Therefore, this study examined this comorbidity utilizing the standardized Borderline Personality Questionnaire (BPQ).This study involved 714 adult patients with a primary diagnosis of BD per DSM-IV criteria who were admitted to the psychiatric unit at an academic hospital in Houston, TX between July 2013 and July 2018. All patients completed the BPQ within 72 hours of admission. Statistical analysis was used to detect correlations between severity of BD, length of stay (LOS), and scores on the BPQ. A machine learning model was constructed to predict the parameters affecting patients' readmission rates within 30 days.Analysis revealed that the severity of certain BPD traits at baseline was associated with mood state and outcome measured by LOS. Inpatients with BD who were admitted during acute depressive episodes had significantly higher mean scores on 7 of the 9 BPQ subscales (P0.05) compared with those admitted during acute manic episodes. Inpatients with BD with greater BPQ scores on 4 of the 9 BPQ subscales had significantly shorter LOS than those with lower BPQ scores (P0.05). The machine learning model identified 6 variables as predictors for likelihood of 30-day readmission with a high sensitivity (83%), specificity (77%), and area under the receiver operating characteristic curve of 86%.Although preliminary, these results suggest that inpatients with BD who have higher levels of BPD features were more likely to have depressive rather than manic symptoms, fewer psychotic symptoms, and a shorter LOS. Moreover, machine learning models may be particularly valuable in identifying patients with BD who are at the highest risk for adverse consequences including rapid readmission.

Complex oxides of the RE2Hf2O7 series are functional materials that exist in the fluorite or pyrochlore phase depending on synthesis method and calcination temperature. In this study, we investigate the process of synthesis, crystal structure stabilization, and phase transition in a series of RE hafnate compounds, synthesized by the coprecipitation process of a single-source complex hydroxide precursor followed with direct calcination or molten-salt synthesis (MSS) method. Phase pure RE2Hf2O7 (RE = Y, La, Pr, Gd, Er, and Lu) ultrafine nanostructured powders were obtained after calcinating the precursor in a molten salt at 650 °C for 6 h. Moreover, we demonstrate that the MSS method can successfully stabilize ideal pyrochlore structures for La2Hf2O7 and Pr2Hf2O7 in the nanodomain, which is not possible to achieve by direct calcination of the coprecipitated precursor at 650 °C. We propose mechanisms to elucidate the differences in these two synthesis methods and highlight the superiority of the MSS method for the production of RE hafnate nanoparticles.

Segmentation of head and neck cancer (HNC) primary tumors on medical images is an essential, yet labor-intensive, aspect of radiotherapy. PET/CT imaging offers a unique ability to capture metabolic and anatomic information, which is invaluable for tumor detection and border definition. An automatic segmentation tool that could leverage the dual streams of information from PET and CT imaging simultaneously, could substantially propel HNC radiotherapy workflows forward. Herein, we leverage a multi-institutional PET/CT dataset of 201 HNC patients, as part of the MICCAI segmentation challenge, to develop novel deep learning architectures for primary tumor auto-segmentation for HNC patients. We preprocess PET/CT images by normalizing intensities and applying data augmentation to mitigate overfitting. Both 2D and 3D convolutional neural networks based on the U-net architecture, which were optimized with a model loss function based on a combination of dice similarity coefficient (DSC) and binary cross entropy, were implemented. The median and mean DSC values comparing the predicted tumor segmentation with the ground truth achieved by the models through 5-fold cross validation are 0.79 and 0.69 for the 3D model, respectively, and 0.79 and 0.67 for the 2D model, respectively. These promising results show potential to provide an automatic, accurate, and efficient approach for primary tumor auto-segmentation to improve the clinical practice of HNC treatment.

ObjectivesTo determine the utility of low-flip angle “black bone” magnetic resonance imaging (MRI) for cortical mandibular bone assessment by comparing interdentium cortical measurements and inter-observer morphometric variability in relation to computed tomography (CT).MethodsQuantification of cortical mandible bone width was performed as per Hamada et al. at 15 cross-sectional interdentium locations on pre-treatment black bone MRI and CT for 15 oropharyngeal cancer patients, with inter-observer analyses on a subset of 3 patients by 11 observers. Bland-Altman limits of agreement and bias estimation, Lin’s concordance correlation (LCC), and Deming orthogonal regression were used to compare CT and MRI measurements. The absolute variance and intraclass correlation coefficient (ICC) were implemented for the inter-observer error quantification.ResultsBoth the Bland Altman and Deming regression analyses showed CT and black bone MRI measurements were comparable within ±0.85 mm limits of agreement, and systematically smaller for MRI. LCC (0.60[0.52;0.67]) showed moderate equivalence between modalities. The average absolute variance between the observers was similar on CT (1.13±0.06 mm) and MRI (1.15 ±0.06 mm). The ICC analyses showed that measurement consistency was significantly higher (pConclusionsBlack bone MR sequence is usable as an alternative to CT for cortical mandible bone measurements, allowing use for early detection of cortical alteration (e.g. osteonecrosis). The cortical bone measurements showed substantive but equivalent inter-observer variation on both CT and black bone MRI. (Semi)automated measurement may mitigate this in future work.

This study presents the structural, optical, and radioluminescent characterization of newly synthesized europium-doped lanthanum hafnate (La 2 Hf 2 O 7 :xmol%Eu 3+ , x=0 to 35) nanoparticles (NPs) for use as phosphors and scintillation materials. Samples prepared through a combined co-precipitation and molten salt synthetic process were found to crystalize in the pyrochlore phase, a radiation tolerant structure related to the fluorite structure. These samples exhibit red luminescence under ultraviolet and X-ray excitation. Under these excitations, the optical intensity and quantum yield of the La 2 Hf 2 O 7 :xmol%Eu 3+ NPs depend on the Eu 3+ concentration and are maximized at 5%. It is proposed that there is a trade-off between the quenching due to defect states/cross-relaxation and dopant concentration. An optimal dopant concentration allows the La 2 Hf 2 O 7 :5 mol%Eu 3+ NPs to show the best luminescent properties of all the samples.

Complex oxides of the RE

IMPACT: This work will standardize necessary image pre-processing for diagnostic and prognostic clinical workflows dependent on quantitative analysis of conventional magnetic resonance imaging. OBJECTIVES/GOALS: Conventional magnetic resonance imaging (MRI) poses challenges for quantitative analysis due to a lack of uniform inter-scanner voxel intensity values. Head and neck cancer (HNC) applications in particular have not been well investigated. This project aims to systematically evaluate voxel intensity standardization (VIS) methods for HNC MRI. METHODS/STUDY POPULATION: We utilize two separate cohorts of HNC patients, where T2-weighted (T2-w) MRI sequences were acquired before beginning radiotherapy for five patients in each cohort. The first cohort corresponds to patients with images taken at various institutions with a variety of non-uniform acquisition scanners and parameters. The second cohort corresponds to patients from a prospective clinical trial with uniformity in both scanner and acquisition parameters. Regions of interest from a variety of healthy tissues assumed to have minimal interpatient variation were manually contoured for each image and used to compare differences between a variety of VIS methods for each cohort. Towards this end, we implement a new metric for cohort intensity distributional overlap to compare region of interest similarity in a given cohort. RESULTS/ANTICIPATED RESULTS: Using a simple and interpretable metric, we have systematically investigated the effects of various commonly implementable VIS methods on T2-w sequences for two independent cohorts of HNC patients based on region of interest intensity similarity. We demonstrate VIS has a substantial effect on T2-w images where non-uniform acquisition parameters and scanners are utilized. Oppositely, it has a modest to minimal impact on T2-w images generated from the same scanner with the same acquisition parameters. Moreover, with a few notable exceptions, there does not seem to be a clear advantage or disadvantage to using one VIS method over another for T2-w images with non-uniform acquisition parameters. DISCUSSION/SIGNIFICANCE OF FINDINGS: Our results inform which VIS methods should be favored in HNC MRI and may indicate VIS is not a critical factor to consider in circumstances where similar acquisition parameters can be utilized. Moreover, our results can help guide downstream quantitative imaging tasks that may one day be implemented in clinical workflows.

A series of 5 mol % Eu3+-doped rare earth (RE) hafnium oxide RE2Hf2O7 (RE = Y, La, Pr, Gd, Er, and Lu) nanoparticles (NPs) have been synthesized, calcinated, and systematically investigated using X-ray diffraction (XRD), Raman spectroscopy, X-ray photoelectron spectroscopy, scanning electron microscopy, Fourier transform infrared spectroscopy, and optically excited luminescence. Effects of the A-site RE3+ cation and calcination on the crystal structure of the RE2Hf2O7:5%Eu NPs were distinguished using XRD and Raman studies. Spectroscopic analysis showed that the La2Hf2O7:5%Eu3+ and Pr2Hf2O7:5%Eu3+ possessed ordered pyrochlore structures while the RE2Hf2O7:5%Eu3+ compositions (RE = Y, Er, and Lu) possessed disordered fluorite structure and were thermodynamically stable up to the highest calcination temperature employed in this study (1500 °C); however, a disordered–ordered transition observed in the Gd2Hf2O7:5%Eu3+ composition indicated that it was not thermodynamically stable. Detailed photoluminescence (...

Residents in the city of Mission, Texas have been noted to complain of health related problems possibly associated with low dose radiation exposure. The purpose of this study was to quantify radioactivity levels in the entire Mission area by measuring natural and anthropogenic radionuclide concentrations (226Ra, 228Ra, 137Cs) in 30 representative surface soil samples through high-resolution gamma-ray spectroscopy. The activity concentrations of these radionuclides followed an approximately normal distribution across the samples and were found to be comparable to other similar studies. In addition, radiological impact assessment factors were calculated and found to be within the recommended values.

2031Background: Radiomics derives quantitative features from medical images to reveal novel information about imaging phenotypes. Although significant research has been conducted on the application...

It has been previously shown that the free surface of molecular glasses has enhanced surface diffusion compared to the bulk. However, the degree by which the glass dynamics are affected by the free surface remains unexplored. Here, we measure enhanced dynamics in ultra-thin molecular glass films as a function of film thickness. We demonstrate that these films exhibit a sharp transition from glassy solid to liquid-like behavior when the thickness is reduced below 30 nm. This liquid-like behavior persists even at temperatures well below the glass transition temperature, Tg. The enhanced dynamics in these films can produce large scale morphological features during physical vapor deposition and lead to a dewetting instability in films held at temperatures as low as Tg − 35 K. The effective viscosity of these films are measured by monitoring the dewetting kinetics. These measurements combined with cooling-rate dependent Tg measurements show that the apparent activation barrier for rearrangement decreases sharply in films thinner than 30 nm. This sharp transition in the dynamics suggests that long-range correlated dynamics exists in these films such that the enhancement induced by the free surface can strongly affect the dynamics of the film over a length scale that is ten times larger than the size of the molecules.