26800 research outputs found
Sort by
3D lymphoma segmentation on PET/CT images via multi-scale information fusion with cross-attention
No full textBackground: Accurate segmentation of diffuse large B-cell lymphoma (DLBCL) lesions is challenging due to their complex patterns in medical imaging. Traditional methods often struggle to delineate these lesions accurately.
Objective: This study aims to develop a precise segmentation method for DLBCL using 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography (PET) and computed tomography (CT) images.
Methods: We propose a 3D segmentation method based on an encoder-decoder architecture. The encoder incorporates a dual-branch design based on the shifted window transformer to extract features from both PET and CT modalities. To enhance feature integration, we introduce a multi-scale information fusion (MSIF) module that performs multi-scale feature fusion using cross-attention mechanisms with a shifted window framework. A gated neural network within the MSIF module dynamically adjusts feature weights to balance the contributions from each modality. The model is optimized using the dice similarity coefficient (DSC) loss function, minimizing discrepancies between the model prediction and ground truth. Additionally, we computed the total metabolic tumor volume (TMTV) and performed statistical analyses on the results.
Results: The model was trained and validated on a private dataset of 165 DLBCL patients and a publicly available dataset (autoPET) containing 145 PET/CT scans of lymphoma patients. Both datasets were analyzed using five-fold cross-validation. On the private dataset, our model achieved a DSC of 0.7512, sensitivity of 0.7548, precision of 0.7611, an average surface distance (ASD) of 3.61 mm, and a Hausdorff distance at the 95th percentile (HD95) of 15.25 mm. On the autoPET dataset, the model achieved a DSC of 0.7441, sensitivity of 0.7573, precision of 0.7427, ASD of 5.83 mm, and HD95 of 21.27 mm, outperforming state-of-the-art methods (p \u3c 0.05, t-test). For TMTV quantification, Pearson correlation coefficients of 0.91 (private dataset) and 0.86 (autoPET) were observed, with R2 values of 0.89 and 0.75, respectively. Extensive ablation studies demonstrated the MSIF module\u27s contribution to enhanced segmentation accuracy.
Conclusion: This study presents an effective automatic segmentation method for DLBCL that leverages the complementary strengths of PET and CT imaging. The method demonstrates robust performance on both private and publicly available datasets, ensuring its reliability and generalizability. Our method provides clinicians with more precise tumor delineation, which can improve the accuracy of diagnostic interpretations and assist in treatment planning for DLBCL patients. The code for the proposed method is available at https://github.com/chenzhao2023/lymphoma_seg
Long Trip Charging Planning of Battery Electric Vehicle Considering Vehicle Waiting Time Forecast at Fast Charging Stations: A Mixed-Integer Dynamic Programming Approach
Get PDFBattery electric vehicles have shared concerns such as range anxiety and long charging times compared to conventional vehicles. To consider an electric vehicle as a replacement for a conventional vehicle, it is essential to reduce the cost of ownership and encourage widespread adoption without range concerns. It is appropriate that charging planning should be utilized to tackle such problems. It can also be helpful to have information from charging planning services to apprise incoming vehicles for charging to prepare their trip itinerary. For long-distance road trips, choosing appropriate charging stations can be instrumental in achieving the optimization goal. Selecting the right charging station can affect the total trip time. To optimize charging planning, we utilize mixed-integer dynamic programming to achieve the global optimal solution. The information on vehicle waiting time is used and parameterized from open-source US household travel survey information. Probability distributions of daily trips, daily mileage, and probability distribution of arrival vehicle SOC can be derived. Monte Carlo simulation generates the waiting time at the charging station based on the probability distributions and station configuration. The presented optimal charging plan method is validated by multiple test cases. The test cases are designed based on US Alternate Fuel Corridors utilizing interstate highways. For the selected test scenarios up to 9.6% of time saving is observed when there is no waiting time at the charging stations. The validation tests with waiting time encountered at the charging station, time saving of up to 15.1% is achieved with energy level closer to the minimum limit at the end of the trip. This charging planning method significantly enhances the viability of EVs for long-distance travel by addressing core concerns like range anxiety and lengthy trip time
Dual-modality Encoder-decoder Framework for Urban Real-time Rainfall-runoff Prediction
No full textUrban real-time rainfall-runoff prediction (URRP) is a potential nonstructural measure for urban flood management. Due to significant influence of human activities and the built environment, it is still a challenging issue to obtain satisfactory results that support effective emergency response. To obtain accurate and stable prediction results, this investigation employs rainfall and runoff datasets that mutually complementary to construct URRP model, providing more comprehensive features of runoff process. To extract features from such two kinds of datasets, a dual-modality encoder-decoder (DM-ED) model is proposed. DM-ED model employs encoder-decoder (ED) framework to enhance multi-steps prediction performance, and LSTM is embedded in the encoder and decoder layers, respectively, thus capturing high time-dependence and non-linearity of rainfall and runoff features. Then interactive dual cross (IDC) attention module is designed to capture global cross feature between rainfall and runoff. Additionally, different from common used pre-fusion approach, we propose a post-fusion (PF) module to efficiently fuse rainfall and runoff features, which can capture more comprehensive information and enhance model robustness. The DM-ED with IDC and PF model is trained and tested on urban rainfall-runoff events (January 2018 - December 2019) over a 3.52 km2 terrain in Chongqing, China. Several experiments have been conducted on this terrain, and the experimental results show that NSE, RMSE,, MAE coefficients outperform other traditional models. The results indicate that DM-ED with IDC and PF model is expected to offer a reliable and effective method for URRP tasks
Microstructural and Mechanical Evaluation of Rapidly Solidified Mg–Zn–Ca–Mn and Mg–Zn–Ca–Zr Melt-Spun Alloys
No full textThis work presents the preparation of Mg–2.0Zn–0.5Ca–0.4Mn (CZM) and Mg–1.8Zn–0.3Ca–0.2Zr (CZZ) alloys using typical casting or melt-spinning, and each followed by hot extrusion. The as-cast CZM and CZZ alloys comprise well-formed α-Mg grains, characterized by numerous coarse Mg2Ca and Mg6Zn3Ca2 precipitates, a low solute concentration, and a coarse network of precipitates at the grain boundaries (GBs), together with α-Mg + Mg6Zn3Ca2 and α-Mg + Mg2Ca eutectic structures at the GBs triple junctions. The as-cast alloy demonstrates a lower tensile strength (224 MPa) compared to the as-spun alloy (265 MPa) because of these structural characteristics. In comparison, the microstructure of the as-spun CZM and CZZ samples is distinguished by coarse grains encircled by relatively fine grains, indicative of dynamic recrystallization. The elevated strength of the as-spun CZM and CZZ alloys is ascribed to robust refinement of grains strengthening, solution strengthening, and a greater density of smaller precipitates contributing to precipitation strengthening. Fracture study indicates that the as-cast CZM and CZZ alloys exhibit unique ductile dimples, with quasicleavage and cleavage fractures. The fracture characteristics of the as-spun CZM and CZZ alloys continue to display cleavage planes and ductile dimples
A Review of Efforts To Improve Dynamic Environment Testing Practices
No full textWhen performing in-lab vibration or shock testing, the test article is attached to the desired excitation equipment through a test fixture. Historically, these fixtures are designed to be rigid, with the fixture natural frequencies much higher than the band of interest, therefore not inducing any additional dynamics on the test article. However, this approach results in an unrealistic representation of the dynamic field environments experienced by test articles which can lead to over-testing or under-testing. Several methods aimed to improve dynamic environment testing have been developed and studied over the years. This paper contains a review of the history of vibration testing and methods explored to improve common test practices with a focus on dynamic fixture design and optimization
Highly Efficient Visible Light Thermo-Photo Catalytic Water Splitting over Ru/Black TiO2
No full textPhotocatalytic hydrogen production using solar energy represents a sustainable approach to addressing the dual challenges of fossil fuel dependence and global energy demand. Despite extensive research, the efficiency of visible-light-driven water splitting remains limited, hindering practical application. In this work, we demonstrate a high-performance thermo-photo catalytic system using Ru/black TiO2 which achieves an impressive hydrogen production rate of 307 mmol·g–1·h–1 under visible light at 300 °C. An apparent quantum efficiency (QE) of 82% was measured across the full visible spectrum, surpassing most previously reported values
From Vulnerability to Robustness: A Survey of Patch Attacks and Defenses in Computer Vision
No full textAdversarial patch attacks have emerged as a powerful and practical threat to machine learning models in vision-based tasks. Unlike traditional perturbation-based adversarial attacks, which often require imperceptible changes to the entire input, patch attacks introduce localized and visible modifications that can consistently mislead deep neural networks across varying conditions. Their physical realizability makes them particularly concerning for real-world security-critical applications. In response, a growing body of research has proposed diverse defense strategies, including input preprocessing, robust model training, detection-based approaches, and certified defense mechanisms. In this paper, we provide a comprehensive review of patch-based adversarial attacks and corresponding defense techniques. First, we introduce a new task-oriented taxonomy that systematically categorizes patch attack methods according to their downstream vision applications (e.g., classification, detection, segmentation), and then we summarize defense mechanisms based on three major strategies: Patch Localization and Removal-based Defenses, Input Transformation and Reconstruction-based Defenses, Model Modification and Training-based Defenses. This unified framework provides an integrated perspective that bridges attack and defense research. Furthermore, we highlight open challenges, such as balancing robustness and model utility, addressing adaptive attackers, and ensuring physical-world resilience. Finally, we outline promising research directions to inspire future work toward building trustworthy and robust vision systems against patch-based adversarial threats
Disentangling complex relationships and disjunctions in western Camassia: Integrating multiple criteria to resolve taxonomic boundaries
No full textPremise: Understanding genetic and morphological variability helps efforts to sustain landscapes and develop effective species concepts for resolving difficult groups. To unravel puzzling relationships and range disjunctions, we applied morphometrics, phenology, phylogenetics and population genetics in Camassia species with cultural, ecological, and conservation value, asking: Do the unusual Camassia populations in northeastern California represent previously unrecognized, disjunct C. howellii? Do C. howellii, C. leichtlinii, and C. quamash maintain diagnostic features in allopatry or sympatry? Are C. quamash subsp. breviflora and subsp. linearis taxonomically distinct?. Methods: We evaluated 34 Camassia populations in situ for morphometric traits, phenology, and habitat type, collecting tissue for population microsatellite and phylogenetic analyses (rpl16, trnD-trnT). Fieldwork and genetic analyses of Camassia species allowed hypothesis testing of all criteria. Results: Oregon and California populations of C. howellii shared 94–95% morphospace but differed significantly from C. leichtlinii and C. quamash, primarily in having more basal leaves, subglobose fruits, and smaller flowers that open in mid-late afternoon, closing at sunset without reopening. Both microsatellite and phylogenetic data indicated separation of the three species, with slight genetic differentiation between the disjunct populations of C. howellii. Subspecies of C. quamash differed morphologically and genetically, with clear phylogenetic separation. Conclusions: Integrative approaches proved effective, affirming disputed species identities and upholding subspecific status for C. quamash subsp. linearis. For C. howellii, population genetic differentiation between disjunct regions appears congruent with phylogenetic analyses. Slight morphological differentiation of Oregon and California populations is consistent with geographic isolation, implying subspecific genetic divergence worthy of future study
Validation of directed energy laser simulation and evaluation of HEL weapon thermal impacts on UAV with MuSES
No full textDirected energy laser weapons are becoming more common on the global battlefield, with an increasing prevalence of both Earth-based and space-based platforms expected in the future. The thermal impact of these HEL weapons can be significant and understanding the potential for thermally-induced physical damage as a function of laser power, beam focus and dwell time is critical to mission planning. Testing is understood to be the standard for ground truth in this regard, and perhaps cannot be avoided completely, but laboratory and (especially) field testing can be difficult to depend on entirely. This is particularly true when the object of interest is an uncooperative or adversarial target in a challenging environment (such as a fast-moving airborne asset in a turbulent atmosphere). The use of transient thermal prediction software is motivated by the need to perform scientific studies where variables such as laser power, beam profile, dwell time on target and atmospheric effects can be controlled. Additionally, the effectiveness of possible survivability countermeasures can be evaluated virtually by computing the reduction in thermal impact due to a particular design/scenario change. In this paper we validate transient thermal simulations of several metals undergoing heating from a high power laser source against previously-published laboratory test data. We report thermal predictions that closely match the published test data based on the published material properties and boundary conditions. Subsequently, we demonstrate a methodology for simulating the thermal impact of a HEL weapon on an unmanned aerial vehicle (UAV) in flight. We detail how critical factors such as laser power level, beam profile and dwell time can be included in such a study. We report DE-induced, scenario-dependent temperature rise and explore a representative countermeasure design to demonstrate efficacy evaluation. Finally, we suggest how the incorporation of atmospheric effects into the transient simulation could be accomplished
Cyber-Informed Transmission Planning Incorporating Security Violations and Cascading
No full textThe North American Electric Reliability Corporation (NERC) has recently promoted the concept of cyber-informed transmission planning, through integration of components including substations, generation, and transmission and distribution interfaces. While unmanned substations utilize camera surveillance to deter physical intrusions, vulnerability through electronic breaches is still a relevant concern. This paper enhances grid characterization by employing physics-based modeling through power flow analysis. Hypothetical simulated combinations in conjunction with both established and innovative models makeup the approach. Integration of data and models through simulation produce a platform capable of comprehensive analysis, monitoring, and optimization. The paper underscores the crucial role of cyber-related contingencies in power systems and their respective substations, highlighting proof-of-concept studies conducted by simulation