2021 Vol. 30, No. 5

Hippocampal Segmentation in Brain MRI Images Using Machine Learning Methods: A Survey
PAN Yi, LIU Jin, TIAN Xu, LAN Wei, GUO Rui
2021, 30(5): 793-814. doi: 10.1049/cje.2021.06.002
Abstract(1192) PDF(776)
The hippocampus is closely related to many brain diseases, such as Alzheimer's disease. Accurate measurement of the hippocampus is helpful for clinicians in identifying lesions and then diagnosing and treating the related brain diseases. Therefore, accurate segmentation of the hippocampus is of vital significance for the in-depth study of many brain diseases. However, the accurate measurement of the hippocampus depends on its accurate segmentation, and hippocampal segmentation has always been a challenging problem due to the small size, irregular shape, and fuzzy boundaries with surrounding tissues of the hippocampus. With the development of machine learning, many innovative methods have been proposed to segment the hippocampus. The purpose of this survey is to provide a comprehensive overview of hippocampal segmentation in brain MRI images using machine learning methods. First, a brief introduction to hippocampal segmentation in brain MRI images is given. Then, common evaluation metrics of hippocampal segmentation are introduced. Next, brain hippocampal segmentation methods based on traditional machine learning and deep learning are described. Subsequently, some common open datasets and toolkits applied to brain hippocampal segmentation are presented. Finally, objective conclusions regarding hippocampal segmentation in brain MRI images using machine learning methods are drawn, and future developments and trends are identified for brain hippocampal segmentation.
A Machine Learning Method for Differentiating and Predicting Human-Infective Coronavirus Based on Physicochemical Features and Composition of the Spike Protein
WANG Chao, ZOU Quan
2021, 30(5): 815-823. doi: 10.1049/cje.2021.06.003
Abstract(745) PDF(47)
Several Coronaviruses (CoVs) are epidemic pathogens that cause severe respiratory syndrome and are associated with significant morbidity and mortality. In this paper, a machine learning method was developed for predicting the risk of human infection posed by CoVs as an early warning system. The proposed Spike-SVM (Support vector machine) model achieved an accuracy of 97.36% for Human-infective CoV (HCoV) and Nonhuman-infective CoV (Non-HCoV) classification. The top informative features that discriminate HCoVs and Non-HCoVs were identified. Spike-SVM is anticipated to be a useful bioinformatics tool for predicting the infection risk posed by CoVs to humans.
Geometric Mean Maximum FSVMI Model and Its Application in Carotid Artery Stenosis Risk Prediction
ZHANG Xueying, GUO Yuling, LI Fenglian, WEI Xin, HU Fengyun, HUI Haisheng, JIA Wenhui
2021, 30(5): 824-832. doi: 10.1049/cje.2021.06.004
Abstract(712) PDF(37)
Carotid artery stenosis is a serious medical condition that can lead to stroke. Using machine learning method to construct classifier model, carotid artery stenosis can be diagnosed with transcranial doppler data. We propose an improved fuzzy support vector machine model to predict carotid artery stenosis, with the maximum geometric mean as the optimization target. The fuzzy membership function is obtained by combining information entropy with the normalized class-center distance. Experimental results showed that the proposed model was superior to the benchmark models in sensitivity and geometric mean criteria.
Myocardial Infarction Detection and Localization with Electrocardiogram Based on Convolutional Neural Network
LIU Jikui, WANG Ruxin, WEN Bo, LIU Zengding, MIAO Fen, LI Ye
2021, 30(5): 833-842. doi: 10.1049/cje.2021.06.005
Abstract(646) PDF(34)
Electrocardiogram (ECG) is widely used in Myocardial infarction (MI) diagnosis. The automatic diagnosis of MI based on the 12-lead ECG needs to consider not only the waveform change features in multi-resolution time series, but also the spatial correlation information between the leads. To this end, this work proposed multiscale spatiotemporal feature extraction method based on Convolutional neural network (CNN) for MI automatic diagnosis. First, the 12-lead ECG is first transformed into an ECG image through wavelet decomposition and 3-dimensional space reconstruction. The MI-CNN model is then constructed to identify MI using 41368 ECG images. Finally, we develop the LL-CNN model, which is utilized only after the ECG signal is identified as an MI event by the MI-CNN model, to localize MI by employing transfer learning to overcome the limited data problem. The proposed method has achieved an accuracy of 99.51% on MI detection, and a macro-F1 of 99.14% on MI localization. Moreover, the features visualization shows that U-wave has significant diagnostic value for MI. The proposed method significantly improves the performance of MI detection and localization compared with other methods. It is promising to be used for MI monitoring and diagnosis.
Diagnostic Classification of Lung Cancer Using Deep Transfer Learning Technology and Multi-Omics Data
ZHU Rong, DAI Lingyun, LIU Jinxing, GUO Ying
2021, 30(5): 843-852. doi: 10.1049/cje.2021.06.006
Abstract(764) PDF(55)
In recent years, with the increasing application of highthroughput sequencing technology, researchers have obtained and accumulated a large amount of multi-omics data, making it possible to diagnose cancer at the gene expression level. The proliferation of various omics data can provide a large amount of biological information, which brings new opportunities and great challenges as well to cancer classification and diagnosis. Machine learning algorithms for early diagnosis of lung cancer have emerged that distinguish cancers of the early and late stages by using genomic features. Omics data are generally characterized with low sample size, high dimensionality and high noise. Therefore, simple direct application of common classification methods cannot achieve better performance and must be improved in a targeted manner. This paper puts forward a combined convolutional neural network and convolutional autoencoders approach to construct a deep migratory learning classification model for early lung cancer diagnosis. First, the convolutional auto-encoders algorithm is used to reduce the dimensionality of the dataset in order to make it better meet the requirements of migration learning. Second, a neural network model is constructed with the original dataset and the existing labeled dataset, and the model migration rules are set as well. Finally, a small number of labeled target datasets are used in the training to complete the construction of the classification model. The proposed convolutional neural network method based on model migration and five other popular machine learning models are used to classify and predict the three lung cancer gene datasets and the integrated dataset. The experimental results show that such four evaluation metrics as accuracy, precision, recall, and f1-score with our proposed method have obtained better prediction performance, and the average area under curve result also shows our proposed method is optimal.
A 1.8-V 240-MHz 2.19-mW Four-Stage CMOS OTA with a Segmenting Frequency Compensation Technique
LIANG Yuhua, ZHENG Zirui, LIU Shubin, LI Dengquan, DING Ruixue, ZHU Zhangming
2021, 30(5): 853-860. doi: 10.1049/cje.2021.06.007
Abstract(775) PDF(52)
A four-stage Operational transconductance amplifier (OTA) used in an infrared temperature sensor adopting the proposed Feed-forward Gm-stage and segmenting nested Miller compensation technique is presented. The purpose of the proposed segment compensation is primarily to make more amplifier stages concatenated. The circuit linked several transconductance stages to form a segment, and linked several segments to form a large multistage amplifier. For example, a two-stage amplifier linked with a three-stage amplifier can realize a five-stage amplifier, two three-stage amplifiers linked can realize a six-stage amplifier. A four-stage amplifier in the form of 2+2 (two stages and two stages) was used as an example to verify this compensation method. The proposed OTA is designed in the 180nm complementary metal oxide semiconductor process. It consists of two parts which ensure the stability and improve the bandwidth performance. The first part is a feed-forward transconductance stage, and the second part is the two-segment transconductance stages with the Miller compensation techniques employed within and between the two segments. Based on the small-signal model, stability analysis and theoretical derivation are performed in theory. On condition of a 2-pF load capacitance, a direct current gain of 109dB and a gain-bandwidth of 240MHz with a phase margin of 50° can be achieved. The proposed design consumes 2.19mW in a 1.8-V supply voltage. The transient simulation indicates that the settling time of the output is 19ns with the settling error being 1%, and the slew rate is 114V/μs.
Analysis on Three-Dimensional Gate Edge Roughness of Gate-All-Around Devices
SUN Shuang, LI Ming, ZHANG Baotong, LI Xiaokang, CAI Qifeng, LI Haixia, BI Ran, XU Xiaoyan, HUANG Ru
2021, 30(5): 861-865. doi: 10.1049/cje.2021.06.008
Abstract(526) PDF(32)
As the physical size of metal-oxide-semiconductor field effect transistor approaches the end of scaling down, the effect of process-induced variations such as gate edge roughness on device performance cannot be neglected. For gate-all-around devices, the three-dimensional gate profiles make the evaluation of gate edge roughness different and more complicated than that in planar metal-oxide-semiconductor field effect transistors. In this work, an evaluation algorithm was proposed to model the three-dimensional gate edge roughness in a real gate-all-around device. The results show that the typical trapezoidal gate is more likely to suffer from gate edge roughness effect than the ideal rectangular gate. The effect of the size of the gate and the correlation coefficient of the edges on the effective channel length variation was also studied.
Targeted Adversarial Examples Generating Method Based on cVAE in Black Box Settings
YU Tingyue, WANG Shen, ZHANG Chunrui, WANG Zhenbang, LI Yetian, YU Xiangzhan
2021, 30(5): 866-875. doi: 10.1049/cje.2021.06.009
Abstract(633) PDF(21)
In recent years, adversarial examples has become one of the most important security threats in deep learning applications. For testing the security of deep learning models in adversarial environment, many researches focus on generating adversarial examples quickly and efficiently. In order to solve the problems of existing generative adversarial networks based methods which can not effectively generate the targeted adversarial examples in black box settings, and to improve the temporal performance of gradient-based generating methods, an adversarial examples generating method based on conditional Variational autoencoder (cVAE) is proposed in this paper, where a cVAE is designed elaborately to generate adversarial examples without most of the detailed information about the attacked deep learning models, of which the output can be controlled arbitrarily by these crafted inputs, used to test the robustness of deep learning models against adversarial examples. The experimental results show that the proposed method can achieve a comparable attack success rate and a better temporal performance than the existing gradient-based generating methods in black box environment.
Online Video Popularity Regression Prediction Model with Multichannel Dynamic Scheduling Based on User Behavior
QIAO Sibo, PANG Shanchen, WANG Min, ZHAI Xue, DAI Feng
2021, 30(5): 876-884. doi: 10.1049/cje.2021.06.010
Abstract(674) PDF(32)
Popularity prediction of online video is widely used in many different scenarios. It can not only help video service providers to schedule video web sites, but also bring considerable profits on investment for both providers and advertisers if popularity of online video is predicted accurately. However, online video popularity prediction still cannot have a satisfactory result, due to the complexity of many crucial factors especially of video distribution network. In this article, we extract seven factors from huge amounts of data about user behavior, establishing a new multiple linear regression model to initially predict online video popularity. After that, a multichannel video popularity dynamic scheduling model is proposed to schedule videos on which channel and what time to be broadcast, according to its popularity predicted by multiple linear regression model, ensuring that maximum the sum value of online video popularity of each channel. Experimental results on dataset obtained from Sohu Video, a video service provider in China, and real-world video flow in Sohu Video demonstrate that the proposed model is robust and has promising performance in predicting online video popularity, which is helpful for video service providers to schedule videos on web sites effectively in the future.
Extraction Security of Sequential Aggregate Signatures
ZHAI Jiaqi, LIU Jian, CHEN Lusheng
2021, 30(5): 885-894. doi: 10.1049/cje.2021.06.011
Abstract(371) PDF(25)
Aggregate signature schemes enable anyone to compress many signatures into one. Besides the unforgeability of aggregate signature, another property called aggregate extraction sometimes is useful. In sequential aggregate signature schemes, the aggregate signature is computed incrementally by the signers. We introduce extraction security in the sequential context which can be seen as an analogy of aggregate extraction introduced by Boneh et al. in CRYPTO 2003. In addition to the practical meaning, it also has benefits in the construction of hierarchical identity-based signatures as we will point out. We study several well known aggregate signature schemes and prove some of them satisfy extraction security.
Secret Sharing Schemes from Linear Codes over $\mathbb{F}_2$+v$\mathbb{F}_2$+v2$\mathbb{F}_2$
WANG Yaru, LI Fulin, ZHU Shixin
2021, 30(5): 895-901. doi: 10.1049/cje.2021.06.012
Abstract(577) PDF(29)
Secret sharing is an important concept in cryptography, however it is a difficult problem to determine the access structure of the secret sharing scheme based on a linear code. In this work, we construct two-weight linear codes over finite field by using linear codes over finite ring. We first study MacDonald codes over the finite ring $\mathbb{F}_2$+v$\mathbb{F}_2$+v2$\mathbb{F}_2$ with v3=v. Then we give torsion codes of MacDonald codes of type α and β, which are two-weight linear codes. Finally we give the access structures of secret sharing schemes based on the dual codes of the two-weight codes.
STATE: A Clustering Algorithm Focusing on Edges Instead of Centers
ZHAO Boxiang, WANG Shuliang, LIU Chuanlu
2021, 30(5): 902-908. doi: 10.1049/cje.2021.07.001
Abstract(684) PDF(34)
With the expansion of data scale and the increase in data complexity, it is particularly important to accurately identify clusters and efficiently save clustering results. To address this, we propose a novel clustering algorithm, Shape clustering based on data field (STATE), which can quickly identify clusters of arbitrary shapes and greatly reduce the storage space of clustering results in any datasets without reducing the accuracy. STATE mainly focuses on finding the edges of clusters and directions of edges instead of clustering centers through the data field. The results of STATE are presented as the edges of clusters without data objects inside clusters and without noise. Extensive experiments show that STATE can recognize complex data distribution in noisy environments without discrimination and greatly save the storage space of clustering results. When it is applied in a real-world scene, facial feature extraction, STATE can recognize eyes, nose, mouth, eyebrows and facial contours automatically without calibrating key features or training. Using the extracted facial features, we achieve facial recognition with high accuracy.
Sitting or Standing Data Acquisition Based Breast Ultrasound Computed Tomography
LI Ruijing, CHEN Houjin, PENG Yahui, LI Jupeng, LI Yanfeng
2021, 30(5): 909-917. doi: 10.1049/cje.2021.07.002
Abstract(452) PDF(15)
Ultrasound computed tomography (UCT) is a promising approach for early breast cancer screening. However, current studies which use prone posture to collect breast ultrasonic data cause four problems, a long non-data-acquisition time, inconvenient, possible chances for cross infection, and a large area occupied by equipment. The purpose of this study is to estimate a complete breast UCT image by using sitting or standing data acquisition, which can obtains a more rapid, convenient and sanitary examination process, and a less space occupied by equipment. Therefore, this study proposes a sitting or standing data acquisition based breast UCT method, which is a more practical data acquisition method for breast UCT. This study places a uniform soft sleeve on the outside of the breast so that it would not be deformed significantly due to the change of body posture. Because the soft sleeve is an influencing factor from outside for breast imaging, this study discusses the considerations for selecting that. Computer simulations are conducted to prove the effectiveness of the proposed method. Results suggest that, by using a soft sleeve whose sound speed is between 1450m/s and 1550m/s, the proposed method is effective; the biases of the reconstructed images are less than 1% under the 5% noise condition.
Hypersonic Vehicle Trajectory Prediction Algorithm Based on Hough Transform
LI Fan, XIONG Jiajun, LAN Xuhui, BI Hongkui, TAN Xiansi
2021, 30(5): 918-930. doi: 10.1049/cje.2021.07.003
Abstract(793) PDF(44)
Trajectory prediction is a prerequisite for missile and high-speed vehicle guidance interception. To address the trajectory problem for the near space hypersonic unpowered gliding vehicle. Firstly, the trajectory prediction mechanism is analyzed. Based on the existing NSHV trajectory prediction method, the key techniques of trajectory prediction are discussed from the perspectives of prediction parameter selection and prediction parameter description. Then, a trajectory prediction method based on Hough transform is proposed. The core of the method is to select the longitude latitude, and high direction positions as prediction parameters, and use the Hough transform to fit the prior basis functions. In the latitude and longitude direction, the rationality of predicting parameters is discussed from two aspects:position extreme point and lateral maneuverability. In the height direction, for the NSHV oscillation drop characteristic, two Hough transforms are used to separate the periodic term from the linear term. In addition, we give the approximate interval of the parameter values of the method, and design a parameter adaptive scheme. Finally, trajectory prediction is performed on two different control patterns. The error of the prediction time of 100s is within 25km, and the results show the effectiveness of the proposed method.
GridNet-3D: A Novel Real-Time 3D Object Detection Algorithm Based on Point Cloud
YUE Yuanchen, CAI Yunfei, WANG Dongsheng
2021, 30(5): 931-939. doi: 10.1049/cje.2021.07.004
Abstract(791) PDF(31)
3D object detection based on point cloud has an important application prospect in automatic driving technology. Aiming at the low precision of 3D object detection based on point cloud and the poor real-time performance caused by large numbers of 3D convolutions, a novel end-to-end real-time object detection algorithm named GridNet-3D is proposed. In the work, 2D gridmapping is used to preprocess the original point clouds. Then a novel structure grid encoding layer is adopted to encode point cloud features and is gotten grid feature maps in bird's eye view which is connected to region proposal network module to generate detections. Despite only using point clouds, the results on the KITTI 3D detection benchmark show that our algorithm has higher detection precision and better real-time performance on the detection of cars, pedestrians and cyclists, which has high practical value.
Dynamic Multi-Task Allocation Method for Passenger Diffusion in Mobile Crowd Sensing
JIANG Weijin, LYU Sijian
2021, 30(5): 940-946. doi: 10.1049/cje.2021.07.005
Abstract(713) PDF(26)
This paper aims to solve the problem of low efficiency, high cost and instability in opportunistic network transmission in the process of mobile group intelligence perception task allocation. Two multi-task dynamic distribution methods based on Lowest cost Participant selection algorithm (LC-PSA) based on user incentive cost and Least number Participant selection algorithm (LN-PSA) based on number of users are proposed respectively. Through these two algorithms, the goal of minimizing the number of people and moving distance required for the task and reducing the system's incentive cost is achieved. Simulation experiments show that compared with similar algorithms, the number of participants in the task distribution scheme selected by the LN-PSA algorithm is reduced by 24.0%, and the system resource consumption is lower, which can provide stable services for the system when users are insufficient in emergencies. Compared with the traditional greedy heuristics algorithm, the LC-PSA algorithm reduces the total system cost by 37.74% and has better overall performance in the comparison experiment.
A Pairing-Free Certificateless Signcryption Scheme for Vehicular Ad Hoc Networks
DU Hongzhen, WEN Qiaoyan, ZHANG Shanshan, GAO Mingchu
2021, 30(5): 947-955. doi: 10.1049/cje.2021.07.006
Abstract(768) PDF(29)
Vehicular ad hoc networks (VANETs) create an vital platform for communication between vehicles, which can realize accident warning, auxiliary driving, road traffic information query, passenger communication and other applications. While providing convenient services for people, VANETs also bring some security risks. Security and privacy are the primary issues in the research of VANETs. Signcryption is an ideal way to transfer messages in a logical step in a secure and authenticated way. We design a Certificateless signcryption (CLSC) scheme to provide confidentiality, authentication, integrity, non-repudiation and user privacy preservation for the information transmitted between vehicle communication units. We demonstrate the confidentiality and unforgeability of the proposed scheme in the random oracle model. Moreover, compared with the existing CLSC schemes, ours realizes the perfect combination of efficiency, security and privacy, and it is particularly well adapted to the secure communication of vehicle networks.
A Novel Synchronization Method in Terahertz Large-Scale Antenna Array System
ZHENG Chen, DING Xuhui, LIU Dekang, BU Xiangyuan, AN Sining
2021, 30(5): 956-968. doi: 10.1049/cje.2021.07.007
Abstract(705) PDF(52)
We focus on the problems of the accurate time delay estimation, the design of training pilots, and hybrid matrix optimization within the large-scale antenna array Terahertz (THz) broadband communication system. In contrast to the existing researches based on narrow-band arrays, we hereby shed light on the time delay estimation of broadband arrays. In THz broadband communication systems, the data symbol duration is relatively short when comparing with the dimension of the antenna array. In large-scale antenna systems, signals received in each antenna are no longer different phase-shifted copies of the same symbol, but completely different symbols in which occasion traditional narrowband structure is no longer suitable. Based on the above conclusion, firstly, we put forward a system model based on large-scale antenna arrays and Time delay line (TDL) structure. Secondly, we deduce the Cramer-Rao lower bound (CRLB) of the time delay estimation, and present a time delay estimation algorithm that could reach the CRLB. Thirdly, by minimizing the CRLB, we address the design of the training pilot and optimized TDL structure under the condition of constant envelope training pilot and modulus TDL structure. Finally, we disclose the numerical simulation results. According to the simulation results, the aforementioned method is workable in reaching the CRLB, the TDL structure can significantly surpass that of the traditional model, and the optimal pilot design method outperforms the pseudo-random pilot structure.
An Asynchronous Quasi-Cloud/Edge/Client Collaborative Federated Learning Mechanism for Fault Diagnosis
MA Xue, WEN Chenglin
2021, 30(5): 969-977. doi: 10.1049/cje.2021.07.008
Abstract(672) PDF(33)
Although the federated learning method has the ability to balance data and protect data privacy by means of model aggregation, while the existing methods are difficult to achieve the effectiveness of centralized learning under data sharing. The existing federated structure only has a certain degree of confidentiality for data privacy, that is to say, each client can reconstruct a part of the information of other clients based on the model parameters shared between the server and the clients under certain conditions. In order to make the federated learning mechanism more confidential, we breaks the existing mechanism that the parameters between the federated model and the client model are completely shared, and establishes a new asynchronous quasi-cloud/edge/client collaborative federated learning mechanism. We construct a hierarchical multi-level confidential communication network, where the network parameters are shared in a way of quasi-cloud/edge/client coordination without data communication. The cloud and the edges respectively use the sequential Kalman filter algorithm to perform an asynchronous fusion of the network parameters uploaded in their respective fusion centers for the next round of updates; The effectiveness of the proposed algorithm is verified on the data of a type of rotating machinery.
Method for Evaluating Digital Video Electromagnetic Information Leakage from Video Cable
WANG Sen, QIU Yang, TIAN Jin
2021, 30(5): 978-985. doi: 10.1049/cje.2021.07.009
Abstract(566) PDF(26)
Electromagnetic emissions from electrical information equipment may contain useful information and lead to information leakage. In order to detect the electromagnetic information leakage of digital video cable, a rapid testing method for computer video information leakage detection without reconstructing the displayed image is proposed. In this method, the electromagnetic radiation spectrum of typical image transmitted by HDMI cable is simulated. Then the electromagnetic radiation spectrum of the computer when transmitting the same image as the typical image is obtained by conduction and radiation EMC test methods. Finally, the correlation between the spectrum of simulation and the test spectrum is analyzed, and the correlation coefficient is used to judge whether the computer has produced the electromagnetic leakage of video information. The analysis and test results show that when the correlation coefficient is greater than 0.5, the electromagnetic leakage of computer video information can be judged.
Total Ionizing Dose Effect and Failure Mechanism of Digital Signal Processor
YU Xin, LU Wu, LI Xiaolong, LIU Mohan, WANG Xin, SUN Jing, GUO Qi
2021, 30(5): 986-990. doi: 10.1049/cje.2021.07.010
Abstract(538) PDF(26)
Ionizing radiation effect and failure mechanism of Digital signal processor (DSP) is studied through test-board and automatic test equipment to find the relationship between system function failure and parameter degradation. Static bias is more sensitive than dynamic bias when DSP is tested on-line during radiation. Core current, high-Z leakage current and timing parameter are sensitive to ionizing radiation. No enhanced low-dose-rate sensitivity is found by comparing experiment results under high and low dose rate radiation. External memory interface and Timer are deduced to be the sensitive module by step radiation and analysis basing full parameter test in Verigy 93000. The timing parameter degradation have a strong correlation to these module functions. And the degeneration mechanism is analysed on inverter through Hspice simulation which indicate the leakage circuit caused by radiation can lead a delay to the digital signal propagating. The parasitical capacitance among long connections make it worse to the data transmission around DSP, field programmable gate array and memory. Then an early function failure occurs in test board than Verigy 93000. This work provide support to systematic radiation hardness design and hardness assurance/lot acceptance testing in space applications.