Search

Although the development of computer-aided algorithms for sleep staging is integrated into automatic detection of sleep disorders, most supervised deep learning-based models might suffer from insufficient labeled data. While the adoption of semi-supervised learning (SSL) can mitigate the issue, the SSL models are still limited to the lack of discriminative feature extraction for diverse obstructive sleep apnea (OSA) severity. This model deterioration might be exacerbated during the domain adaptation. Such exploration on the alleviation of domain-shift of SSL model between different OSA conditions has attracted more and more attentions from the clinic. In this work, a co-attention meta sleep staging network (CMS2-net) is proposed to simultaneously deal with two issues: the inter-class disparity problem and the intra-class selection problem. Within CMS2-net, a co-attention module and a triple-classifier are designed to explicitly refine the coarse feature representations by identifying the class boundary inconsistency. Moreover, the mutual information with meta contrastive variance is introduced to supervise the gradient stream from a multi-scale view. The performance of the proposed framework is demonstrated on both public and local datasets. Furthermore, our approach achieves the state-of-the-art SSL results on both datasets. ispartof: Ieee Journal Of Biomedical And Health Informatics vol:26 issue:7 pages:3447-3453 ispartof: location:United States status: published

Analysis of electroencephalogram (EEG) is a crucial diagnostic criterion for many sleep disorders, of which sleep staging is an important component. Manual stage classification is a labor-intensive process and usually suffered from many subjective factors. Recently, more and more computer-aided techniques have been applied to this task, among which deep convolutional neural network has been performing well as an effective automatic classification model. Despite some comprehensive models have been developed to improve classification results, the accuracy for clinical applications has not been reached due to the lack of sufficient labeled data and the limitation of extracting latent discriminative EEG features. Therefore, we propose a novel hybrid manifold-deep convolutional neural network with hyperbolic attention. To overcome the shortage of labeled data, we update the semi-supervised training scheme as an optimal solution. In order to extract the latent feature representation, we introduce the manifold learning module and the hyperbolic module to extract more discriminative information. Eight subjects from the public dataset are utilized to evaluate our pipeline, and the model achieved 89% accuracy, 70% precision, 80% sensitivity, 72% f1-score and kappa coefficient of 78%, respectively. The proposed model demonstrates powerful ability in extracting feature representation and achieves promising results by using semi-supervised training scheme. Therefore, our approach shows strong potential for future clinical development.

With the development of modern cameras, more physiological signals can be obtained from portable devices like smartphone. Some hemodynamically based non-invasive video processing applications have been applied for blood pressure classification and blood glucose prediction objectives for unobtrusive physiological monitoring at home. However, this approach is still under development with very few publications. In this paper, we propose an end-to-end framework, entitled cocktail causal container, to fuse multiple physiological representations and to reconstruct the correlation between frequency and temporal information during multi-task learning. Cocktail causal container processes hematologic reflex information to classify blood pressure and blood glucose. Since the learning of discriminative features from video physiological representations is quite challenging, we propose a token feature fusion block to fuse the multi-view fine-grained representations to a union discrete frequency space. A causal net is used to analyze the fused higher-order information, so that the framework can be enforced to disentangle the latent factors into the related endogenous association that corresponds to down-stream fusion information to improve the semantic interpretation. Moreover, a pair-wise temporal frequency map is developed to provide valuable insights into extraction of salient photoplethysmograph (PPG) information from fingertip videos obtained by a standard smartphone camera. Extensive comparisons have been implemented for the validation of cocktail causal container using a Clinical dataset and PPG-BP benchmark. The root mean square error of 1.329±0.167 for blood glucose prediction and precision of 0.89±0.03 for blood pressure classification are achieved in Clinical dataset.

Modern wireless communication systems are facing an increasingly complex electromagnetic environment, which is affected by a variety of noise and interference signals. In the communication process, if the communication party can effectively identify the type of interference signal, it can take corresponding anti-interference measures to avoid or suppress interference to the maximum extent and reduce the damage of interference to communication quality. In recent years, with the continuous development of deep learning and its outstanding performance in image and speech processing, it has been verified that deep learning has extremely strong nonlinear mapping and data expression capabilities, inspiring researchers to apply deep learning to the field of communication anti-jamming, and improving the antijamming capability of the system. In this paper, an algorithm of interference signal recognition based on a complex convolution neural network is proposed. It also introduces the network architecture. Next, six typical interference signals at each level are identified, and the results are analyzed. The last chapter of the paper summarizes the full text. The experimental results show that the recognition ability of the interference signal recognition algorithm proposed in this paper is better than that of the traditional convolutional neural network.

PURPOSE Leukemia is a lethal disease that is harmful to bone marrow and overall blood health. The classification of white blood cell images is crucial for leukemia diagnosis. The purpose of this study is to classify white blood cells by extracting discriminative information from cell segmentation and combining it with the fine-grained features. We propose a hybrid adversarial residual network with support vector machine (SVM), which utilizes the extracted features to improve the classification accuracy for human peripheral white cells. METHODS Firstly, we segment the cell and nucleus by utilizing an adversarial residual network, which contains a segmentation network and a discriminator network. To extract features that can handle the inter-class consistency problem effectively, we introduce the adversarial residual network. Then, we utilize convolutional neural network (CNN) features and histogram of oriented gradient (HOG) features, which can extract discriminative features from images of segmented cell nuclei. To utilize the representative features fully, a discriminative network is introduced to deal with neighboring information at different scales. Finally, we combine the vectors of HOG features with those of CNN features and feed them into a linear SVM to classify white blood cells into six types. RESULTS We used three methods to evaluate the effect of leukocyte classification based on 5000 leukocyte images acquired from a local hospital. The first approach is to use the CNN features as the input of SVM to classify leukocytes, which achieved 94.23% specificity, 95.10% sensitivity, and 94.41% accuracy. The use of the HOG features for SVM achieved 83.50% specificity, 87.50% sensitivity, and 85.00% accuracy. The use of combined CNN and HOG features achieved 94.57% specificity, 96.11% sensitivity, and 95.93% accuracy. CONCLUSIONS We propose a novel hybrid adversarial-discriminative network for the classification of microscopic leukocyte images. It improves the accuracy of cell classification, reduces the difficulty and time pressure of doctors' work, and economizes the valuable time of doctors in daily clinical diagnosis.

Immunotherapy has become a promising form of treatment for cancers. There is a need to predict response to immunotherapy accurately. In the UCSC Xena, pan-cancer analysis revealed a positive relationship between APOBEC3B (A3B) and tumor mutational burden ( R = 0.28 , P < 0.001 ) and microsatellite instability ( R = 0.12 , P < 0.05 ). Naturally, the A3B high expression group had higher tumor mutational burden and microsatellite instability than the low expression group. The bladder cancer (BLCA) cohort in The Cancer Genome Atlas (TCGA) revealed tumor mutational signatures of A3B high and low expression groups. Compared to the low expression group, the high expression group had a higher number of SNPs and mutations. Subsequently, A3B was profiled for immune cell infiltration and immune checkpoints in bladder cancer. The results showed that A3B was positively correlated with most immune cells. Compared with the A3B low expression group, the A3B high expression group had higher expression of immune checkpoints. A3B was positively correlated with CD274 ( R = 0.12 , P = 0.016 ). This indicated that the high expression of A3B may have a better response to immunotherapy. Furthermore, data from the IMvigor210 immunotherapy clinical trial was used to confirm the findings of this study. The combined survival analysis of A3B and CD274 showed that the group of patients with high expression of CD274 and A3B was found to have a significantly higher survival rate than the rest of the patient group ( P < 0.047 ). The results demonstrated that A3B has a significant role in immunotherapy. Moreover, the combined biomarkers of A3B and CD274 were more effective in predicting response to immunotherapy in bladder urothelial carcinoma. The findings of this study provide valuable insights for precision medicine.

Wound dressings can be applied over the wound sites to provide long-lasting wound management and improve wound healing. Biological wound dressings are superior to synthetic materials due to biodegradability and biocompatibility. These biomaterials have demonstrated huge potential in the field of wound dressings. Applying bibliometric analysis combined with results-based descriptions to characterize the research status, hotspots, and cutting-edge topics, this study is the first in-depth qualitative, quantitative, data-driven overview of biological wound dressings research in recent decades. Filtered data were used to construct co-citation, heatmaps, bi-clustering, strategy maps, and other analyses and visualization. The results show that research on biological wound dressings has progressed considerably in the last 5 years with extensive global collaboration. A clear knowledge base has been developed. Chitosan hydrogels, bacterial cellulose, active agents (silver nanoparticles, growth factors, curcumin, etc.), and electrospinning fibers stand out as research hotspots. The research frontiers include novel starting materials, precise and controlled release systems, and clinical and regenerative medicine applications. We interpreted an overview of the excavated topics and expected the findings here to provide a guide and inspire innovations for developing the next generation wound dressings.

Two Cd(II) coordination compounds, i.e. [Cd(HL)(phen)2(H2O)]·HL·3H2O (1) and {[Cd(HMeL)2(4,4′-bipy)(H2O)]·H2O}n (2) and (where HL = N-carboxymethyl-N-phenyliminoacetato, HMeL = N-carboxymethyl-N-(m-methylphenyl)iminoacetato, 4,4′-bipy = 4,4′-bipyridine and phen = phenanthroline) have been synthesized and characterized by elemental analysis, IR and single-crystal X-ray diffraction methods. The photoluminescence properties and mechanisms, studied by means of solid state absorption, excitation/emission spectra, time-resolved lifetime techniques, and density functional theory (DFT) calculations, indicate that the emissions of both compounds originate from a combination of exciplex and charge-transfer complex. Interestingly, both compounds exhibit excitation-dependent emissions, which can be used to fine-tune their emission wavelengths.

Router security defense technologies emerging in recent years could hardly detect and block the new booming threats with unknown signatures such as hardware Trojan, zero-day attacks, etc. We present a novel router defense technology, distributed router shadow, which builds a closed execution environment to deceive attacks entering into the router, thereby misleading the attackers into regarding it as the real attack target and executing the suspicious code to maximize the chances of detonating the system exploit; thus the original router is prevented from attacking and the suspicious code can be detected. Our experiment and analysis show that the router shadow can defend not only attacks with signature but also some new attacks without signature.

Ir/H-SSZ-13 catalysts were used for methylcyclopentane (MCP) ring-opening (RO) reaction, and exhibit higher yields to RO products than H-ZSM-5, H-USY and Al2O3 supported Ir catalysts. Ir species supported on H-SSZ-13-A (about 200 nm) exhibits higher activity than those on H-SSZ-13-B (about 8 μm) due to its more accessible acid sites. Ir/H-SSZ-13-A gives higher reactivity than Ir/Al2O3, especially in the case of low Ir loading because of the promotion of the strong acid sites located on the external surface of H-SSZ-13-A. This work demonstrates that SSZ-13 supported metal is a potential catalyst for RO reaction.

Building a service function chain (SFC) based on Virtual Network Functions (VNFs) greatly improves the flexibility of network service deployment. However, compared to highly reliable carrier-grade proprietary hardware, VNF is facing a series of failure risks. Although the existing backup methods can solve the problem of VNF failure to some extent, malicious attacks can't be stopped totally because there is no consideration of the homogeneity of the original node and the backup nod. In this paper, we propose a heterogeneous backup deployment scheme. It ensures the heterogeneity of backup server nodes and VNF nodes with the original one during backup. We designed and implemented the corresponding deployment algorithm. Simulation results show that compared with the backup method without considering heterogeneity, our method increases attacker's attack time cost by 13.2% under the cost that the request acceptance rate decreases by 4.81 %.

The SDN (Software Defined Networking) paradigm rings flexibility to the network management and is an enabler to offer huge opportunities for network programmability. And, to solve the scalability issue raised by the centralized architecture of SDN, multi-controllers deployment (or distributed controllers system) is envisioned. In this paper, we focus on increasing the diversity of SDN control plane so as to enhance the network security. Our goal is to limit the ability of a malicious controller to compromise its neighboring controllers, and by extension, the rest of the controllers. We investigate a heterogeneous Susceptible-Infectious-Susceptible (SIS) epidemic model to evaluate the security performance and propose a coloring algorithm to increase the diversity based on community detection. And the simulation results demonstrate that our algorithm can reduce infection rate in control plane and our work shows that diversity must be introduced in network design for network security.

Sniffer attack has been a severe threat to network communication security. Traditional network usually uses static network configuration, which provides convenience to sniffer attack. In this paper, an SDN-based double hopping communication (DHC) approach is proposed to solve this problem. In DHC, ends in communication packets as well as the routing paths are changed dynamically. Therefore, the traffic will be distributed to multiple flows and transmitted along different paths. Moreover, the data from multiple users will be mixed, bringing difficulty for attackers in obtaining and recovering the communication data, so that sniffer attack will be prevented effectively. It is concluded that DHC is able to increase the overhead of sniffer attack, as well as the difficulty of communication data recovery.

To improve the limitations of Ethernet over Coax (EoC). technology in the bidirectional reformation of the hybrid fiber-coaxial (HFC) network, this paper proposes the concept of Ethernet passive electronic network (EPEN) based on multi-point control protocol (MPCP) and the hybrid access scheme of Ethernet passive optical network (EPON) and EPEN. This paper describes the network architecture of EPEN, and introduces the implementation scheme and design of the EPEN system. At present, the project group has been working on the core technology research and key software and hardware design of the EPEN system, on which we have accomplished the hardware design and debugging, the embedded software coding and debugging, the EPEN network management protocol design and software coding. However, there is still a lot to do to achieve networking capability, management, security, quality of service (QoS) and some other functions.

At present, the maximal transmission distance between the optical line terminal (OLT) and optical network unit (ONU) of an Ethernet passive optical network (EPON) system is 20 km. However, this distance should be extended to 50 km or even longer in some applications such as fiber to the village (FTTV). A method for extending the transmission distance of an EPON system is proposed in this paper, in which the optical power amplifier and the adjustment of dynamic bandwidth allocation (DBA) are the key technologies.

The character of 1 Gbps bandwidth and treebased structure make Ethernet passive optical network (EPON) very suitable for broadcast or multicast services such as Internet protocol television video (IPTV). The document proposed a novel scheme, based on the former research for controllable multicast over EPON system, mainly considering system security and maintainability. It can both control the IPTV program source’s and the receiver’s validity, improving the efficiency and precision.

Point clouds registration is one of the key parts in 3D model reconstruction. Random Sample Consensus (RANSAC) is a typical algorithm for coarse registration, which can provide initial values for accurate registration methods such as ICP. In this paper, we propose an improved RANSAC algorithm based on 3D region covariance descriptor(RC RANSAC). Region covariance descriptor of each point in a down sampled point cloud is established. The region covariance descriptor extracts the statistical information of neighborhood for each point. Region covariance descriptor has more significant distinction than Euclidean distance adopted by classical RANSAC and another improved RANSAC algorithm, called M RANSAC. The accuracy can be improved by significant distinction. Additionally, the region covariance descriptor with compact structure is efficient in time. Experimental results on 3D point clouds registration show that the RCRANSAC has superior performance to RANSAC and MRANSAC in computational complexity and accuracy.

This paper describes the design of a burst-mode ultra-long haul PON trunk fiber protector based on SOA. It's transparent to protocols and thus compatible with various PON systems. With this technology, the robustness and transmission distance of PON network can be greatly improved and extended. The fiber protector supports a maximum transmission distance of 50km and a 1:128 branching ratio. It supports automatic protection switching of trunk fiber whose longest protection switching time is less than 20ms. In addition, in the design of the ultra-long haul PON system, by EPON as an example, the ranging of EPON system, RTT compensation and other parameters also need to be amended. As a result, this article also makes research and description on synchronization, ranging, time delay compensation, and other key parameters of the EPON system.

To address the disadvantages of the EPON+EOC technical scheme used in the bidirectional reformation of HFC network and broadband access application, this paper proposes the concept of Ethernet Passive Electronic Network(EPEN) based on MPCP protocol. All kinds of modulated EOC technologies are based on CSMA/CA half-duplex mode MAC protocol, showing low efficiency in access network application. Learning from EPON point to multi-point (P2MP) control protocol, this paper originally proposes the concept of full-duplex EPEN based on MPCP protocol for upstream. This paper describes the network architecture of EPEN, and discusses the technical implementation scheme of the EPEN system. At present the core technologies and implementation of software and hardware of EPEN are being studied.

The character of 1Gbps bandwidth and tree-based structure make EPON very suitable for broadcast or multicast services such as IPTV. The document proposed a novel scheme, based on the former research for controllable multicast over EPON system, mainly considering system security and maintainability. It can both control the IPTV program source's and the receiver's validity, improving the efficiency and precision. The processing of the two configurations mode is given in detail.

The character of 1Gbps bandwidth and tree-based structure make EPON very suitable for broadcast or multicast services such as IPTV. In this paper we focus on security mechanisms that should be available to groups that specifically request it, and therefore are willing to pay the extra cost of implementing those mechanisms. We concentrate on two issues: how unauthorized multicast traffic can be prevented by means of multicast transit traffic control; how a group's susceptibility is likely to be reduced by means of multicast group access control. We have proposed a novel scheme that should be available to multicast groups that desire precautionary measures be taken to avert the threats of Invalid source and unauthorized access.

Performance management is a very important part of network management system, it monitors network usage and troubleshoot network-related problems, captures and stores significant network traffic data to analysis network QoS. A methodology of performance management for EPON is proposed in this paper, which consists of three parts: statistics collector, embedded performance database, data analyzer and processor.

A main issue to be resoled urgently in multi-service supported EPON system is to find a reasonable solution for TDM service access. This paper introduces a suitable TDM access scheme which is the combination of VoIP and CESoP technology. In our EPON system design, VoIP and CESoP technology are applied respectively to deal with two main kind of TDM service - Voice and Leased line(such as E1/T1). The scheme has been implemented in our EPON system and has achieved good results in some triple-play pilot projects.