Edge roughness quantifies impact of physician variation on training and performance of deep learning auto-segmentation models for the esophagus.

Manual segmentation of tumors and organs-at-risk (OAR) in 3D imaging for radiation-therapy planning is time-consuming and subject to variation between different observers. Artificial intelligence (AI) can assist with segmentation, but challenges exist in ensuring high-quality segmentation, especially for small, variable structures, such as the esophagus. We investigated the effect of variation in segmentation quality and style of physicians for training deep-learning models for esophagus segmentation and proposed a new metric, edge roughness, for evaluating/quantifying slice-to-slice inconsistency. This study includes a real-world cohort of 394 patients who each received radiation therapy (mainly for lung cancer). Segmentation of the esophagus was performed by 8 physicians as part of routine clinical care. We evaluated manual segmentation by comparing the length and edge roughness of segmentations among physicians to analyze inconsistencies. We trained eight multiple- and individual-physician segmentation models in total, based on U-Net architectures and residual backbones. We used the volumetric Dice coefficient to measure the performance for each model. We proposed a metric, edge roughness, to quantify the shift of segmentation among adjacent slices by calculating the curvature of edges of the 2D sagittal- and coronal-view projections. The auto-segmentation model trained on multiple physicians (MD1-7) achieved the highest mean Dice of 73.7 ± 14.8%. The individual-physician model (MD7) with the highest edge roughness (mean ± SD: 0.106 ± 0.016) demonstrated significantly lower volumetric Dice for test cases compared with other individual models (MD7: 58.5 ± 15.8%, MD6: 67.1 ± 16.8%, p < 0.001). A multiple-physician model trained after removing the MD7 data resulted in fewer outliers (e.g., Dice ≤ 40%: 4 cases for MD1-6, 7 cases for MD1-7, Ntotal = 394). While we initially detected this pattern in a single clinician, we validated the edge roughness metric across the entire dataset. The model trained with the lowest-quantile edge roughness (MDER-Q1, Ntrain = 62) achieved significantly higher Dice (Ntest = 270) than the model trained with the highest-quantile ones (MDER-Q4, Ntrain = 62) (MDER-Q1: 67.8 ± 14.8%, MDER-Q4: 62.8 ± 15.7%, p < 0.001). This study demonstrates that there is significant variation in style and quality in manual segmentations in clinical care, and that training AI auto-segmentation algorithms from real-world, clinical datasets may result in unexpectedly under-performing algorithms with the inclusion of outliers. Importantly, this study provides a novel evaluation metric, edge roughness, to quantify physician variation in segmentation which will allow developers to filter clinical training data to optimize model performance.

Observer preference of artificial intelligence-generated versus clinical prostate contours for ultrasound-based high dose rate brachytherapy.

BACKGROUND: For trans-rectal ultrasound (TRUS)-based high dose rate (HDR) prostate brachytherapy, prostate contouring can be challenging due to artifacts from implanted needles, bleeding, and calcifications. PURPOSE: To evaluate the geometric accuracy and observer preference of an artificial intelligence (AI) algorithm for generating prostate contours on TRUS images with implanted needles. METHODS: We conducted a retrospective study of 150 patients, who underwent HDR brachytherapy. These patients were randomly divided into training (104), validation (26) and testing (20) sets. An AI algorithm was trained/validated utilizing the TRUS image and reference (clinical) contours. The algorithm then provided contours for the test set. For evaluation, we calculated the Dice coefficient between AI and reference prostate contours. We then presented AI and reference contours to eight clinician observers, and asked observers to select their preference. Observers were blinded to the source of contours. We calculated the percentage of cases in which observers preferred AI contours. Lastly, we evaluate whether the presence of AI contours improved the geometric accuracy of prostate contours provided by five resident observers for a 10-patient subset. RESULTS: The median Dice coefficient between AI and reference contours was 0.92 (IQR: 0.90-0.94). Observers preferred AI contours for a median of 57.5% (IQR: 47.5, 65.0) of the test cases. For resident observers, the presence of AI contours was associated with a 0.107 (95% CI: 0.086, 0.128; p < 0.001) improvement in Dice coefficient for the 10-patient subset. CONCLUSION: The AI algorithm provided high-quality prostate contours on TRUS with implanted needles. Further prospective study is needed to better understand how to incorporate AI prostate contours into the TRUS-based HDR brachytherapy workflow.

Prospective deployment of an automated implementation solution for artificial intelligence translation to clinical radiation oncology.

Introduction: Artificial intelligence (AI)-based technologies embody countless solutions in radiation oncology, yet translation of AI-assisted software tools to actual clinical environments remains unrealized. We present the Deep Learning On-Demand Assistant (DL-ODA), a fully automated, end-to-end clinical platform that enables AI interventions for any disease site featuring an automated model-training pipeline, auto-segmentations, and QA reporting. Materials and methods: We developed, tested, and prospectively deployed the DL-ODA system at a large university affiliated hospital center. Medical professionals activate the DL-ODA via two pathways (1): On-Demand, used for immediate AI decision support for a patient-specific treatment plan, and (2) Ambient, in which QA is provided for all daily radiotherapy (RT) plans by comparing DL segmentations with manual delineations and calculating the dosimetric impact. To demonstrate the implementation of a new anatomy segmentation, we used the model-training pipeline to generate a breast segmentation model based on a large clinical dataset. Additionally, the contour QA functionality of existing models was assessed using a retrospective cohort of 3,399 lung and 885 spine RT cases. Ambient QA was performed for various disease sites including spine RT and heart for dosimetric sparing. Results: Successful training of the breast model was completed in less than a day and resulted in clinically viable whole breast contours. For the retrospective analysis, we evaluated manual-versus-AI similarity for the ten most common structures. The DL-ODA detected high similarities in heart, lung, liver, and kidney delineations but lower for esophagus, trachea, stomach, and small bowel due largely to incomplete manual contouring. The deployed Ambient QAs for heart and spine sites have prospectively processed over 2,500 cases and 230 cases over 9 months and 5 months, respectively, automatically alerting the RT personnel. Discussion: The DL-ODA capabilities in providing universal AI interventions were demonstrated for On-Demand contour QA, DL segmentations, and automated model training, and confirmed successful integration of the system into a large academic radiotherapy department. The novelty of deploying the DL-ODA as a multi-modal, fully automated end-to-end AI clinical implementation solution marks a significant step towards a generalizable framework that leverages AI to improve the efficiency and reliability of RT systems.

Synthesis and Physical Properties of Phase-Engineered Transition Metal Dichalcogenide Monolayer Heterostructures.

Heterostructures of transition metal dichalcogenides (TMDs) offer the attractive prospect of combining distinct physical properties derived from different TMD structures. Here, we report direct chemical vapor deposition of in-plane monolayer heterostructures based on 1H-MoS2 and 1T'-MoTe2. The large lattice mismatch between these materials led to intriguing phenomena at their interface. Atomic force microscopy indicated buckling in the 1H region. Tip-enhanced Raman spectroscopy showed mode structure consistent with Te substitution in the 1H region during 1T'-MoTe2 growth. This was confirmed by atomic resolution transmission electron microscopy, which also revealed an atomically stitched, dislocation-free 1H/1T' interface. Theoretical modeling revealed that both the buckling and absence of interfacial misfit dislocations were explained by lateral gradients in Te substitution levels within the 1H region and elastic coupling between 1H and 1T' domains. Phase field simulations predicted 1T' morphologies with spike-shaped islands at specific orientations consistent with experiments. Electrical measurements across the heterostructure confirmed its electrical continuity. This work demonstrates the feasibility of dislocation-free stitching of two different atomic configurations and a pathway toward direct synthesis of monolayer TMD heterostructures of different phases.

Large-area synthesis of high-quality monolayer 1T'-WTe2 flakes.

Large-area growth of monolayer films of the transition metal dichalcogenides is of the utmost importance in this rapidly advancing research area. The mechanical exfoliation method offers high quality monolayer material but it is a problematic approach when applied to materials that are not air stable. One important example is 1T'-WTe2, which in multilayer form is reported to possess a large non saturating magnetoresistance, pressure induced superconductivity, and a weak antilocalization effect, but electrical data for the monolayer is yet to be reported due to its rapid degradation in air. Here we report a reliable and reproducible large-area growth process for obtaining many monolayer 1T'-WTe2 flakes. We confirmed the composition and structure of monolayer 1T'-WTe2 flakes using x-ray photoelectron spectroscopy, energy-dispersive x-ray spectroscopy, atomic force microscopy, Raman spectroscopy and aberration corrected transmission electron microscopy. We studied the time dependent degradation of monolayer 1T'-WTe2 under ambient conditions, and we used first-principles calculations to identify reaction with oxygen as the degradation mechanism. Finally we investigated the electrical properties of monolayer 1T'-WTe2 and found metallic conduction at low temperature along with a weak antilocalization effect that is evidence for strong spin-orbit coupling.

Kelvin probe microscopy and electronic transport measurements in reduced graphene oxide chemical sensors.

Reduced graphene oxide (RGO) is an electronically hybrid material that displays remarkable chemical sensing properties. Here, we present a quantitative analysis of the chemical gating effects in RGO-based chemical sensors. The gas sensing devices are patterned in a field-effect transistor geometry, by dielectrophoretic assembly of RGO platelets between gold electrodes deposited on SiO2/Si substrates. We show that these sensors display highly selective and reversible responses to the measured analytes, as well as fast response and recovery times (tens of seconds). We use combined electronic transport/Kelvin probe microscopy measurements to quantify the amount of charge transferred to RGO due to chemical doping when the device is exposed to electron-acceptor (acetone) and electron-donor (ammonia) analytes. We demonstrate that this method allows us to obtain high-resolution maps of the surface potential and local charge distribution both before and after chemical doping, to identify local gate-susceptible areas on the RGO surface, and to directly extract the contact resistance between the RGO and the metallic electrodes. The method presented is general, suggesting that these results have important implications for building graphene and other nanomaterial-based chemical sensors.