Elimination Of False Positives Research Articles

A One-Step Methodology for Identifying Concrete Pathologies Using Neural Networks—Using YOLO v8 and Dataset Review

Pathologies in concrete structures can be visually evidenced on the concrete surface, such as by fissures or cracks, fragmentation of part of the concrete, concrete efflorescence, corrosion stains on the concrete surface, or exposed steel bars, the latter two occurring in reinforced concrete. Therefore, these pathologies can be analyzed via the images of concrete structures. This article proposes a methodology for visually inspecting concrete structures using deep neural networks. This method makes it possible to speed up the detection task and increase its effectiveness by saving time in preparing the identifications to be analyzed and eliminating or reducing errors, such as those resulting from human errors caused by the execution of tedious, repetitive analysis tasks. The methodology was tested to analyze its accuracy. The neural network architecture used for detection was YOLO, versions 4 and 8, which was tested to analyze the gain with migration to a more recent version. The dataset for classification was Ozgnel, which was trained with YOLO version 8, and the detection dataset was CODEBRIM. The use of a dedicated classification dataset allows for a better-trained network for this function and results in the elimination of false positives in the detection stage. The classification achieved 99.65% accuracy.

Minimizing false positives for CTC identification

BackgroundCancer is a leading cause of death worldwide, with metastasis playing a significant role. Circulating Tumour Cells (CTCs) can provide important real-time insights into tumour heterogeneity and clonal evolution, making them an important tool for early diagnosis and patient monitoring. Isolated CTCs are typically identified by immunocytochemistry using positive biomarkers (cytokeratin) and exclusion biomarkers (CD45). However, some white blood cell (WBC) populations can express low levels of CD45 and stain non-specifically for cytokeratin, increasing their risk of misclassification as CTCs. There is a clear need to improve CTC detection and enumeration criteria to unequivocally eliminate interfering WBC populations. ResultsThis study showed that, indeed, some granulocyte subpopulations expressed low levels of CD45 and stained non-specifically for cytokeratin, misidentifying them as CTCs. These same cells, however, strongly expressed CD15, allowing them to be identified as WBCs and excluded from CTC classification. Flow cytometry confirmed the specificity of the CD15 antibody for the granulocyte subpopulation. False positives were considerably reduced from 25 % to 0.2 % by double exclusion, combining a CD15 antibody with a highly specific CD45 antibody. Furthermore, complete elimination of potential false positives was achieved using double exclusion in combination with improved selection of cytokeratin antibody. The study emphasises the importance of a robust exclusion criteria and high antibody specificity in CTC immuno-assays for accurate identification of CTC candidates and thorough exclusion of interfering WBC subpopulations. SignificanceThis study demonstrated how misidentifying a granulocyte subpopulation can lead to inaccurate CTC evaluation. However, sensitivity and specificity of CTC identification may be improved by using high-performing antibodies and by including a second exclusion biomarker, in turn, allowing for a more comprehensive clinical application of CTCs.

Benchmarking Heterogeneous Photocatalysts for the Nitrogen Reduction Reaction

Ammonia is the key ingredient for living organisms, and artificial catalytic synthesis of ammonia (known as the Haber-Bosch process) is undoubtedly a milestone in scientific research in the 20th century and the cornerstone of today's society.[1] However, the Haber-Bosch process is energy-intensive (2% of the global energy consumption) and is accompanied by large CO2 emissions (1% of global CO2 emission).[2] To catch up with the goal of “Net Zero CO2 Emissions” by 2050, it is imperative to develop a carbon-free and sustainable ammonia production process. The photocatalytic ammonia synthesis method relies only on renewable feedstocks (water and air) and sustainable energy input (light) and can be operated under ambient conditions. The current progress in photocatalytic ammonia synthesis suggests, however, there exists a large gap in performance before commercial use is viable. To replace the Haber-Bosch process and produce commercial fertilizer, the solar-to-ammonia (STA) efficiency needs to be at least 0.1% and the highest reported efficiency carried out in a lab setting is approaching 0.04%[2][3] With advances in measurement techniques, recent studies in the field of electrocatalytic ammonia synthesis have shown that many reported highly active catalysts have later been shown to be inactive toward nitrogen reduction.[4] This false positive result is mainly due to the adventitious ammonia in the environment. A series of discussions were triggered, including identifying and eliminating contaminations, developing robust measurements for low-concentration ammonia detection, and establishing rigorous testing protocols.[4-7] At present, the ammonia yield of electrocatalysis is at least an order of magnitude higher than that of photocatalysis, so the problem of nitrogen contamination will only be more serious in the field of photocatalysis.[7] Unfortunately, such discussions have not received equal attention in this field, and few if any rigorous testing protocols have been employed in existing reports.It is imperative to benchmark and develop a robust testing protocol for the growing field of photocatalytic ammonia synthesis. Here, we developed a rigorous setup to suppress all known ammonia contaminations (e.g., ammonia and NOx), systematically discussed the effect of various gas pretreatment methods, and established the background contamination level in the system. We synthesized three highly active photocatalytic materials (Fe-BiOBr, OVs-TiO2, and Fe2O3/g-C3N4) from the literature and measured their ammonia yield using the setup with strict controls.[8-10] To avoid potential interference effects during the measurements, we cross-compared different ammonia-detecting methods (Indophenol blue method and ion chromatography).[6] In addition, we applied isotope labeling with quantitative NMR to confirm the genuine production of photo-fixed ammonia. Going a step further, we methodically compare the effects of two experimental parameters (temperature and gas flow rate) on the performance of photocatalytic ammonia synthesis. These parameters vary greatly across reports, making it difficult to objectively analyze their impacts on the production of photo-fixed ammonia. Based on our findings, we aim to provide a solid benchmark for this burgeoning field and draw more attention to the importance of rigorous testing. Reference: Smil, Vaclav. Enriching the earth: Fritz Haber, Carl Bosch, and the transformation of world food production. MIT press, 2004.Medford, Andrew J., and Marta C. Hatzell. "Photon-driven nitrogen fixation: current progress, thermodynamic considerations, and future outlook." Acs Catalysis4 (2017): 2624-2643.Yuan, Jili, et al. "Efficient Photocatalytic Nitrogen Fixation: Enhanced Polarization, Activation, and Cleavage by Asymmetrical Electron Donation to N N Bond." Advanced Functional Materials4 (2020): 1906983.Andersen, Suzanne Z., et al. "A rigorous electrochemical ammonia synthesis protocol with quantitative isotope measurements." Nature7762 (2019): 504-508.Choi, Jaecheol, et al. "Identification and elimination of false positives in electrochemical nitrogen reduction studies." Nature communications1 (2020): 1-10.Zhao, Yunxuan, et al. "Ammonia detection methods in photocatalytic and electrocatalytic experiments: how to improve the reliability of NH3 production rates?." Advanced Science8 (2019): 1802109.Iriawan, Haldrian, et al. "Methods for nitrogen activation by reduction and oxidation." Nature Reviews Methods Primers1 (2021): 1-26.Liu, Yang, Zhuofeng Hu, and Jimmy C. Yu. "Fe enhanced visible-light-driven nitrogen fixation on BiOBr nanosheets." Chemistry of Materials4 (2020): 1488-1494.Zhang, Guoqiang, et al. "Constructing a tunable defect structure in TiO 2 for photocatalytic nitrogen fixation." Journal of Materials Chemistry A1 (2020): 334-341.Mou, Hongyu, et al. "A one-step deep eutectic solvent assisted synthesis of carbon nitride/metal oxide composites for photocatalytic nitrogen fixation." Journal of materials chemistry A10 (2019): 5719-5725.

Enhancement of Lithium-Mediated Ammonia Synthesis By Addition of Oxygen

Ammonia is one of the most produced chemicals worldwide and is currently synthesized by the Haber-Bosch process, which is a thermally catalyzed method that requires high pressures and temperatures. These harsh conditions, in addition to the prerequisite steam reforming process, leads to about 1 % of the annual energy consumption and 1.4 % of the global CO2 emission. One way to mitigate some of the Haber-Bosch process is to produce ammonia electrochemically, utilizing renewable energy sources.The electrochemical synthesis of ammonia faces several big issues. One being the selectivity, since the more facile hydrogen evolution reaction (HER) will always dominate over the nitrogen reduction reaction (NRR), and another issue being the activity, given that the nitrogen triple bond is very stable and therefore hard to split. Hence, most often the reported ammonia contents are in the low ppm regime, which makes it very susceptible to contaminations both from the gas stream (NH3 and NOx impurities) and the system itself (catalyst, cell, chemicals, nitrile gloves, etc.). To avoid misleading results, several protocols have been published on how to correctly perform NRR experiments [1, 2]. One of these confirm that only the Li-mediated ammonia synthesis (LiMeAS) is currently able to produce ammonia electrochemically [3]. The actual mechanism is not fully understood, but it is generally believed that the first step is Li plating from a Li salt containing non-aqueous electrolyte. The very reactive Li will then react with N2 solvated in the electrolyte to form Li3N, which is believed to hydrolyze to ammonia when in contact with a proton source. The currently highest archived faradaic efficiency (FE) is at 69 % at 20 bar N2 pressure when applying an ionic liquid as a proton shuttle [4].In this work, we achieved up to 79 % FE at 20 bar N2 by the addition of 0.8 mol. % O2 in the reaction atmosphere. The positive effect of O2 is a very counterintuitive observation, since the original work by Tsuneto et al. [5] showed that the use of synthetic air significantly hindered the reaction, as it was postulated that O2 inhibits the reaction due to LiO2 formation and/or leading primarily to the oxygen reduction reaction (ORR). We will present experimental results obtained at 10 and 20 bar with varying O2 contents, which were measured accurately by a mass spectrometer probing the atmosphere just above the electrolyte inside the pressure vessel. By combining experimental observations with theoretical modelling, we conclude that the unexpectedly beneficial role of small O2 concentrations has a positive influences the solid electrolyte interface (SEI), which is of great importance in our system. Additional ex-situ X-Ray diffraction (XRD) and X-Ray photoelectron spectroscopy (XPS) measurements were conducted without exposure to air and moisture, to analyze the SEI layer and deposition after electrochemistry. We believe that this study will not only be beneficial for industrializing the LiMeAS, but will also bring us a step further in understanding the complex mechanism behind this process.[1] S. Z. Andersen et al., "A rigorous electrochemical ammonia synthesis protocol with quantitative isotope measurements," Nature, vol. 570, pp. 504-508, 2019.[2] H. Iriawan et al., "Methods for nitrogen activation by reduction and oxidation," Nature Reviews Methods Primers, vol. 1, no. 1, pp. 1-26, 2021.[3] J. Choi et al., "Identification and elimination of false positives in electrochemical nitrogen reduction studies," Nature communications, vol. 11, no. 1, pp. 1-10, 2020.[4] B. H. Suryanto et al., "Nitrogen reduction to ammonia at high efficiency and rates based on a phosphonium proton shuttle," Science, vol. 372, no. 6547, pp. 1187-1191, 2021.[5] A. Tsuneto, A. Kudo, and T. Sakata, "Lithium-mediated electrochemical reduction of high pressure N2 to NH3," Journal of Electroanalytical Chemistry, vol. 367, no. 1-2, pp. 183-188, 1994.

Automatic pulmonary ground-glass opacity nodules detection and classification based on 3D neural network.

Pulmonary ground-glass opacity (GGO) nodules are more likely to be malignant compared with solid solitary nodules. Due to indistinct boundaries of GGO nodules, the detection and diagnosis are challenging for doctors. Therefore, designing an automatic GGO nodule detection and classification scheme is significantlyessential. In this paper, we proposed a two-stage 3D GGO nodule detection and classification framework. First, we used a pretrained 3D U-Net to extract lung parenchyma. Second, we adapted the architecture of Mask region-based convolutional neural networks (RCNN) to handle 3D medical images. The 3D model was then applied to detect the locations of GGO nodules and classify lesions (benign or malignant). The class-balanced loss function was also used to balance the number of benign and malignant lesions. Finally, we employed a novel false positive elimination scheme called the feature-based weighted clustering (FWC) to promote the detection accuracyfurther. The experiments were conducted based on fivefold cross-validation with the imbalanced data set. Experimental results showed that the mean average precision could keep a high level (0.5182) in the phase of detection. Meanwhile, the false positive rate was effectively controlled, and the competition performance metric (CPM) reached 0.817 benefited from the FWC algorithm. The comparative statistical analyses with other deep learning methods also proved the effectiveness of our proposedmethod. We put forward an automatic pulmonary GGO nodules detection and classification framework based on deep learning. The proposed method locate and classify nodules accurately, which could be an effective tool to help doctors in clinical diagnoses.

Survey of Approaches for Postprocessing of Static Analysis Alarms

Static analysis tools have showcased their importance and usefulness in automated detection of defects. However, the tools are known to generate a large number of alarms which are warning messages to the user. The large number of alarms and cost incurred by their manual inspection have been identified as two major reasons for underuse of the tools in practice. To address these concerns plentitude of studies propose postprocessing of alarms: processing the alarms after they are generated. These studies differ greatly in their approaches to postprocess alarms. A comprehensive overview of the postprocessing approaches is, however, missing. In this article, we review 130 primary studies that propose postprocessing of alarms. The studies are collected by combining keywords-based database search and snowballing. We categorize approaches proposed by the collected studies into six main categories: clustering, ranking, pruning, automated elimination of false positives, combination of static and dynamic analyses, and simplification of manual inspection. We provide overview of the categories and sub-categories identified for them, their merits and shortcomings, and different techniques used to implement the approaches. Furthermore, we provide (1) guidelines for selection of the postprocessing techniques by the users/designers of static analysis tools; and (2) directions that can be explored by the researchers.

Patch-based 3D U-Net and transfer learning for longitudinal piglet brain segmentation on MRI

Background and Objectives:In order to study neural plasticity in immature brain following early brain lesion, large animal model are needed. Because of its morphological similarities with the human developmental brain, piglet is a suitable but little used one. Its study from Magnetic Resonance Imaging (MRI) requires the development of automatic algorithms for the segmentation of the different structures and tissues. A crucial preliminary step consists in automatically segmenting the brain.Methods: We propose a fully automatic brain segmentation method applied to piglets by combining a 3D patch-based U-Net and a post-processing pipeline for spatial regularization and elimination of false positives. Our approach also integrates a transfer-learning strategy for managing an automated longitudinal monitoring evaluated for four developmental stages (2, 6, 10 and 18 weeks), facing the issue of MRI changes resulting from the rapid brain development. It is compared to a 2D approach and the Brain Extraction Tool (BET) as well as techniques adapted to other animals (rodents, macaques). The influence of training patches size and distribution is studied as well as the benefits of spatial regularization.Results: Results show that our approach is efficient in terms of average Dice score (0.952) and Hausdorff distance (8.51), outperforming the use of a 2D U-Net (Dice: 0.919, Hausdorff distance: 11.06) and BET (Dice: 0.764, Hausdorff distance: 25.91). The transfer-learning strategy achieves a good performance on older piglets (Dice of 0.934 at 6 weeks, 0.956 at 10 weeks and 0.958 at 18 weeks) compared to a standard training strategy with few data (Dice of 0.636 at 6 weeks, 0.907 at 10 weeks, not calculable at 18 weeks because of too few training piglets).Conclusions: In conclusion, we provide a method for longitudinal MRI piglet brain segmentation based on 3D U-Net and transfer learning which can be used for future morphometric studies and applied to other animals.

Using Atomically Dispersed Bimetallic Active Sites to Probe the Nitrate Reduction Mechanism to Ammonia

The production of synthetic ammonia via nitrogen reduction is a critical step toward producing fertilizers, central the supporting global food security. To date, the energy-intensive and greenhouse gas-emitting Haber-Bosch process remains the only way to produce ammonia on industrial scale. Recently, significant research efforts have focused on identifying carbon- neutral and distributed pathways to synthesize ammonia, mainly via the electrochemical reduction of dinitrogen (N2).1 This approach, however, is limited by the low N2 solubility, competing hydrogen evolution reactions, and the stability of the N≡N triple bond. A promising alternative pathway to ammonia production is through the reduction of nitrate (NO3 -), a more kinetically and thermodynamically favorable nitrogen precursor. The reduction of nitrate to ammonia provides dual benefits as both a carbon-neutral pathway to ammonia and as a way of denitrification to remove all-too-prevalent and environmentally harmful nitrate contamination, while creating a useful product.2 Although there exists in the literature several works using metal catalysts for the nitrate reduction reaction (NTRR) to ammonia, only very recently have single atom catalysts (SAC) been applied to nitrate reduction.3 , 4 The NTRR is an 8e- transfer reaction, in which the exact electron transfer pathways, either a direct 8e- transfer or by 2 + 6e- pathway, in which a nitrite (NO2 -) intermediate is desorbed and reabsorbed. These processes are not well-studied on SACs. In this talk, we will present a series of atomically dispersed iron and molybdenum, mono and bi-metallic nitrogen doped carbon (M-N-C, M = Fe, Mo, FeMo) electrocatalysts for the NTRR. Our catalysts show potential for faradic efficiencies over 90%, while exhibiting outstanding stability over 48 hours of operation. Furthermore, our results suggest distinct NTRR mechanisms over Fe and Mo active sites, revealing new insights into preferred NTRR pathways over varying single atom sites. Such a detailed understanding of the NTRR electron transfer pathway will not only advance development of high performance, selective SACs, but stands to provide a new route to environmentally responsible ammonia synthesis. References (1) Choi, J.; Suryanto, B. H. R.; Wang, D.; Du, H. L.; Hodgetts, R. Y.; Ferrero Vallana, F. M.; MacFarlane, D. R.; Simonov, A. N. Identification and Elimination of False Positives in Electrochemical Nitrogen Reduction Studies. Nat. Commun. 2020, 11 (1).(2) van Langevelde, P. H.; Katsounaros, I.; Koper, M. T. M. Electrocatalytic Nitrate Reduction for Sustainable Ammonia Production. Joule 2021, 5 (2),(3) Zhu, T.; Chen, Q.; Liao, P.; Duan, W.; Liang, S.; Yan, Z.; Feng, C. Single-Atom Cu Catalysts for Enhanced Electrocatalytic Nitrate Reduction with Significant Alleviation of Nitrite Production. Small 2020, 16 (49).(4) Niu, H.; Zhang, Z.; Wang, X.; Wan, X.; Shao, C.; Guo, Y. Theoretical Insights into the Mechanism of Selective Nitrate-to-Ammonia Electroreduction on Single-Atom Catalysts. Adv. Funct. Mater. 2020, 2008533 (3), 1–8.. Figure 1

Multilevel Intrusion Alert Post-processing for the Elimination of False Positives

Abstract: Intrusion detection systems are the last line of defence in the network security domain. Improving the performance of intrusion detection systems always increase false positives. This is a serious problem in the field of intrusion detection. In order to overcome this issue to a great extend, we propose a multi level post processing of intrusion alerts eliminating false positives produced by various intrusion detection systems in the network. For this purpose, the alerts are normalized first. Then, a preliminary alert filtration phase prioritize the alerts and removes irrelevant alerts. The higher priority alerts are then aggregated to fewer numbers of hyper alerts. In the final phase, alert correlation is done and alert correlation graph is constructed for finding the causal relationship among the alerts which further eliminates false positives. Experiments were conducted on LLDOS 1.0 dataset for verifying the approach and measuring the accuracy. Keywords: Intrusion detection system, alert prioritization, alert aggregation, alert correlation, LLDOS 1.0 dataset, alert correlation graph.

Does the Incremental Value of 123I-Metaiodobenzylguanidine SPECT/CT over Planar Imaging Justify the Increase in Radiation Exposure?

Planar scintigraphy with 123I-radiolabeled metaiodobenzylguanidine (123I-mIBG) is an important imaging modality to evaluate neuroblastoma. In recent years, Single Photon Emission Computed Tomography combined with Computed Tomography (SPECT/CT) has revolutionized nuclear medicine. Nevertheless, the addition of the CT has increased the patients' irradiation. We aimed to evaluate the incremental benefits of 123I-mIBG SPECT/CT over conventional planar imaging and to estimate the relative increase of radiation dose. We retrospectively evaluated the added value of 56 SPECT/CT performed in 40 children in terms of better characterization of the lesion and its locoregional extension, better lymph node staging, detection of new lesions, and elimination of false positives by a paired comparison between the planar images and the SPECT/CT ones. Then, we calculated the percentage contribution of the additional radiation of the CT in this hybrid imagery. In 88% (49 out of 56) of the examinations, SPECT/CT provided additional information, which was crucial in 20% of the cases. It allowed a better characterization of the lesion and its locoregional extension in 44 cases, a better lymph node staging in 28 cases, the detection of 33 new lesions, and the elimination of 9 false positives. The CT effective dose was significantly lower than the SPECT one. The average additional radiation exposure due to CT was 12% (4-23%). 123I-mIBG SPECT/CT has an undeniable added value that improves planar imaging interpretation and impacts patient management. These potential benefits would justify the low additional radiation induced by the CT.

FAIXID: A Framework for Enhancing AI Explainability of Intrusion Detection Results Using Data Cleaning Techniques

Organizations depend on heavy use of various cyber defense technologies, including intrusion detection and prevention systems, to monitor and protect networks and devices from malicious activities. However, large volumes of false alerts from such technologies challenge cybersecurity analysts in isolating credible alerts from false positives for further investigations. In this article, we propose a framework named FAIXID that leverages Explainable Artificial Intelligence (XAI) and data cleaning methods for improving the explainability and understandability of intrusion detection alerts, which in turn assist cyber analysts in making more informed decisions fueled by the quick elimination of false positives. We identified five functional modules in FAIXID: (1) the pre-modeling explainability module that improves the quality of network traffic’s data through data cleaning; (2) the modeling module that provides explanations of the AI models to help analysts make sense of the model internals; (3) the post-modeling explainability module that provides additional explanations to enhance the understandability of the results produced by the AI models; (4) the attribution module that selects the appropriate explanations for the analysts according to their needs; and (5) the evaluation module that evaluates the explanations and collects feedback from analysts. FAIXID has been implemented and evaluated using experiments with real-world datasets. Evaluation of results demonstrates that the utilization of data cleaning and AI explainability techniques provide quality explanations to analysts depending on their expertise and backgrounds.

Functionalized Co2FeAl Nanoparticles for Detection of SARS CoV-2 Based on Reverse Transcriptase Loop-Mediated Isothermal Amplification.

Loop-mediated isothermal amplification (LAMP) is a sensitive, efficient, and rapid nucleic acid amplification technique resulting in a large number of amplicons; however, it suffers from a high incidence of false positives due to carry-over and aerosol. Herein, we report a 10 min nano-capture system that is used in conjunction with a modified reverse transcriptase-LAMP (RT-LAMP) assay for the accurate detection of SARS CoV-2 virus. The nano-capture system employs in-house-designed probe-functionalized magnetic nanoparticles Co2FeAl (cobalt-based Heusler alloy) for efficient capture of contaminating amplicons from the reaction mixture preceding RT-LAMP. The nano-cleaned RT-LAMP assay along with engineered primers successfully detected the presence of 10 copies of SARS CoV-2 virus while completely eliminating the incidence of false positives. The presented contaminant-capture method has been compared with other approaches for elimination of contaminants and was found to be more effective. The insight brought in this work is the design of a rapid nano-capture system that hybridizes with contaminating amplicons (carry-over) with high specificity to enable easy removal from the assay for elimination of false positives. The method has been proven to be successful for RT-LAMP assays in the rapid and highly specific detection of SARS CoV-2, which is currently a major challenge for global health. To the best of our knowledge, this is the first work involving a nano-based cleaning strategy for reliable and rapid diagnosis using isothermal amplification approaches.

Efficient collection of a large number of mutations by mutagenesis of DNA damage response defective animals

With the development of massive parallel sequencing technology, it has become easier to establish new model organisms that are ideally suited to the specific biological phenomena of interest. Considering the history of research using classical model organisms, we believe that the efficient construction and sharing of gene mutation libraries will facilitate the progress of studies using these new model organisms. Using C. elegans, we applied the TMP/UV mutagenesis method to animals lacking function in the DNA damage response genes atm-1 and xpc-1. This method produces genetic mutations three times more efficiently than mutagenesis of wild-type animals. Furthermore, we confirmed that the use of next-generation sequencing and the elimination of false positives through machine learning could automate the process of mutation identification with an accuracy of over 95%. Eventually, we sequenced the whole genomes of 488 strains and isolated 981 novel mutations generated by the present method; these strains have been made available to anyone who wants to use them. Since the targeted DNA damage response genes are well conserved and the mutagens used in this study are also effective in a variety of species, we believe that our method is generally applicable to a wide range of animal species.

Learning Efficient Binarized Object Detectors With Information Compression.

In this paper, we propose a binarized neural network learning method (BiDet) for efficient object detection. Conventional network binarization methods directly quantize the weights and activations in one-stage or two-stage detectors with constrained representational capacity, so that the information redundancy in the networks causes numerous false positives and degrades the performance significantly. On the contrary, our BiDet fully utilizes the representational capacity of the binary neural networks by redundancy removal, through which the detection precision is enhanced with alleviated false positives. Specifically, we generalize the information bottleneck (IB) principle to object detection, where the amount of information in the high-level feature maps is constrained and the mutual information between the feature maps and object detection is maximized. Meanwhile, we learn sparse object priors so that the posteriors are concentrated on informative detection prediction with false positive elimination. Since BiDet employs a fixed IB trade-off to balance the total and relative information contained in the high-level feature maps, the information compression leads to ineffective utilization of the network capacity or insufficient redundancy removal for input in different complexity. To address this, we further present binary neural networks with automatic information compression (AutoBiDet) to automatically adjust the IB trade-off for each input according to the complexity. Moreover, we further propose the class-aware sparse object priors by assigning different sparsity to objects in various classes, so that the false positives are alleviated more effectively without recall decrease. Extensive experiments on the PASCAL VOC and COCO datasets show that our BiDet and AutoBiDet outperform the state-of-the-art binarized object detectors by a sizable margin.

Abnormal Event Correlation and Detection Based on Network Big Data Analysis

With the continuous development of network technology, various large-scale cyber-attacks continue to emerge. These attacks pose a severe threat to the security of systems, networks, and data. Therefore, how to mine attack patterns from massive data and detect attacks are urgent problems. In this paper, an approach for attack mining and detection is proposed that performs tasks of alarm correlation, false-positive elimination, attack mining, and attack prediction. Based on the idea of CluStream, the proposed approach implements a flow clustering method and a two-step algorithm that guarantees efficient streaming and clustering. The context of an alarm in the attack chain is analyzed and the LightGBM method is used to perform false-positive recognition with high accuracy. To accelerate the search for the filtered alarm sequence data to mine attack patterns, the PrefixSpan algorithm is also updated in the store strategy. The updated PrefixSpan increases the processing efficiency and achieves a better result than the original one in experiments. With Bayesian theory, the transition probability for the sequence pattern string is calculated and the alarm transition probability table constructed to draw the attack graph. Finally, a long-short-term memory network and embedding word-vector method are used to perform online prediction. Results of numerical experiments show that the method proposed in this paper has a strong practical value for attack detection and prediction.

Establishing a Reference Dose-Response Calibration Curve for Dicentric Chromosome Aberrations to Assess Accidental Radiation Exposure in Saudi Arabia.

In cases of nuclear and radiological accidents, public health and emergency response need to assess the magnitude of radiation exposure regardless of whether they arise from disaster, negligence, or deliberate act. Here we report the establishment of a national reference dose–response calibration curve (DRCC) for dicentric chromosome (DC), prerequisite to assess radiation doses received in accidental exposures. Peripheral blood samples were collected from 10 volunteers (aged 20–40 years, median = 29 years) of both sexes (three females and seven males). Blood samples, cytogenetic preparation, and analysis followed the International Atomic Energy Agency EPR-Biodosimetry 2011 report. Irradiations were performed using 320 kVp X-rays. Metafer system was used for automated and assisted (elimination of false-positives and inclusion of true-positives) metaphases findings and DC scoring. DC yields were fit to a linear–quadratic model. Results of the assisted DRCC showed some variations among individuals that were not statistically significant (homogeneity test, P = 0.66). There was no effect of age or sex (P > 0.05). To obtain representative national DRCC, data of all volunteers were pooled together and analyzed. The fitted parameters of the radiation-induced DC curve were as follows: Y = 0.0020 (±0.0002) + 0.0369 (±0.0019) * D + 0.0689 (±0.0009) * D2. The high significance of the fitted coefficients (z-test, P < 0.0001), along with the close to 1.0 p-value of the Poisson-based goodness of fit (χ2 = 3.51, degrees of freedom = 7, P = 0.83), indicated excellent fitting with no trend toward lack of fit. The curve was in the middle range of DRCCs published in other populations. The automated DRCC over and under estimated DCs at low (<1 Gy) and high (>2 Gy) doses, respectively, with a significant lack of goodness of fit (P < 0.0001). In conclusion, we have established the reference DRCC for DCs induced by 320 kVp X-rays. There was no effect of age or sex in this cohort of 10 young adults. Although the calibration curve obtained by the automated (unsupervised) scoring misrepresented dicentric yields at low and high doses, it can potentially be useful for triage mode to segregate between false-positive and near 2-Gy exposures from seriously irradiated individuals who require hospitalization.

Tumor segmentation in automated whole breast ultrasound using bidirectional LSTM neural network and attention mechanism

Accurate breast mass segmentation of automated breast ultrasound (ABUS) is a great help to breast cancer diagnosis and treatment. However, the lack of clear boundary and significant variation in mass shapes make the automatic segmentation very challenging. In this paper, a novel automatic tumor segmentation method SC-FCN-BLSTM is proposed by incorporating bi-directional long short-term memory (BLSTM) and spatial-channel attention (SC-attention) module into fully convolutional network (FCN). In order to decrease performance degradation caused by ambiguous boundaries and varying tumor sizes, an SC-attention module is designed to integrate both finer-grained spatial information and rich semantic information. Since ABUS is three-dimensional data, utilizing inter-slice context can improve segmentation performance. A BLSTM module with SC-attention is constructed to model the correlation between slices, which employs inter-slice context to assist segmentation for false positive elimination. The proposed method is verified on our private ABUS dataset of 124 patients with 170 volumes, including 3636 2D labeled slices. The Dice similarity coefficient (DSC), Recall, Precision and Hausdorff distance (HD) of the proposed method are 0.8178, 0.8067, 0.8292 and 11.1367. Experimental results demonstrate that the proposed method offered improved segmentation results compared with existing deep learning-based methods.

The Blood Vasculature of the Retina Using Routine Segmentation

ABSTRACTBiomedical imaging has helped the medical practitioners to facilitate treatments faster and more accurate than conventional methods. Automated technologies suffice the default preprocessing and identification from the images of the retina and their vessel vascular network. Human retinal diseases, leading to blindness, will be identified by the segmentation of blood vessels to enhance the statistical accuracy of intra-retinal layer thickness. This proposal overcomes the drawbacks of traditional angiogram techniques by coloring the retinal paths by fluorescent dyes. From the obtained results, automated segmentation by using the proposed technique has improved then compared methodologies. Employing Gabor filters in the preprocessing functions, elimination of false positives is greatly achieved with some modifications. The followed processes divide the retinal images into linear and curvature segments after implying Gabor constraints to obtain a predicted trajectory of vessels. Limiting the major chances of false positives and better vessel construction in this automated proposal are proved and discussed.

Automated Segmentation of Cervical Intervertebral Disks from Videofluorography Using a Convolutional Neural Network and its Performance Evaluation

Dysphagia has become an important issue in many countries, and there is a strong need for elucidating the causes of dysphagia. One of the promising ways is to analyze the static and dynamic mechanisms of cervical structures, such as epiglottises, hyoid bones, and cervical vertebral bodies, based on medical images. In this study, we propose an automated segmentation method of cervical intervertebral disks (IDs) from videofluorography (VF) by use of a convolutional neural network (CNN). First, cervical masks are extracted from the frame images of VF, and then a patch-based CNN is applied to the cervical masks to obtain the probability images of ID regions. The sizes of patches are changed in a certain range, and pixel values in the patches are normalized. Morphological image filters are applied to the probability images to eliminate false-positive pixels. The proposed method is applied to VF of 58 participants, consisting of 39 healthy people and 19 patients. The segmentation results are compared with the ground truth as determined by a medical doctor and are evaluated with the pixel-wise F-measure. The F-measure is highest (0.880) when the patch size is 21 × 21 pixels and the both of the pixel value normalization (PVN) and the false positive elimination (FPE) are applied. On the other hand, the F-measure is lowest (0.443) when the patch size is 15 × 15 pixels and neither PVN nor FPE is applied.

Optimization and validation of automated dicentric chromosome analysis for radiological/nuclear triage applications

Dicentric Chromosome Assay (DCA) is the most preferred cytogenetic technique for absorbed radiation dose assessment in exposed humans. However, DCA is somewhat impractical for triage application owing to its labor intensive and time consuming nature. Although lymphocyte culture for 48 h in vitro is inevitable for DCA, manual scoring of dicentric chromosomes (DCs) requires an additional time of 24–48 h, making the overall turnaround time of 72–96 h for dose estimation. To accelerate the speed of DC analysis for dose estimation, an automated tool was optimized and validated for triage mode of scoring. Several image training files were created to improve the specificity of automated DC analysis algorithm. Accuracy and efficiency of the automated (unsupervised) DC scoring was compared with the semi-automated scoring that involved human verification and correction of DCs (elimination of false positives and inclusion of true positives). DC scoring was performed by both automated and semi-automated modes for different doses of X-rays and γ-rays (0 Gy–5 Gy). Biodoses estimated from the frequencies of DCs detected by both automated (unsupervised) and semi-automated (supervised) scoring modes were grossly similar to the actual delivered doses in the range of 0.5 to 3 Gy of low LET radiation. We suggest that the automated DC tool can be effectively used for large scale radiological/nuclear incidents where a rapid segregation is essential for prioritizing moderately or severely exposed humans to receive appropriate medical countermeasures.