检索结果-内蒙古大学图书馆

A novel deep learning approach for diabetic retinopathy classification using optical coherence tomography angiography

引用

Multimedia Tools and Applications 2025年 1-39页

作者： A, Abini M. Priya, S. Sridevi Sathya Department of Electronics and Communication Engineering Karunya Institute of Technology and Sciences Tamilnadu Coimbatore India

Diabetic Retinopathy (DR), the leading cause of vision loss in diabetics, requires early and accurate diagnosis to ensure appropriate treatment and prevent irreversible vision damage. This work highlights an automated and noninvasive approach to detect and classify diabetic retinopathy using Optical Coherence Tomography Angiography (OCTA) images. This study addresses several issues including limited dataset size, image quality, and efficient feature extraction through a novel integration of data augmentation, preprocessing, and deep learning techniques. Conditional Generative Adversarial Networks (cGANs) were employed and improved augmentation techniques were implemented to address the problems posed by limited datasets, increasing the dataset size to 3000 high-resolution images. A comprehensive preprocessing pipeline was used to highlight important retinal features, including contrast enhancement via the Regional Dynamic Histogram Equalization (RDHE) algorithm, noise reduction via the Generalized Gauss-Markov Random Field (GGMRF) method, and vessel segmentation using a Markov-Gibbs Random included Field (MGRF) model. The proposed Convolutional Neural Network (CNN) demonstrated exceptional performance with 99.5% accuracy, 99.67% precision, 99.33% recall/sensitivity, 99.67% specificity, 99.5% F1 score and an AUC ROC of 0.997. With a low false negative rate, these measures demonstrate how well the system can differentiate between DR and non-DR cases. By providing early and accurate diagnosis, our work provides a reliable and useful solution for DR screening, improving the diagnosis of diabetic eyes. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2025.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

A Morris-Lecar model with novel fast-slow structure and its FPGA implementation

引用

Science China Technological Sciences 2025年第5期68卷 215-225页

作者： Xihong YU Lilian HUANG Han BAO Bocheng BAO College of Information and Communication Engineering Key Laboratory of Advanced Marine Communication and Information Technology (Ministry of Industry and Information Technology)National Key Laboratory of Underwater Acoustic TechnologyHarbin Engineering University Wang Zheng School of Microelectronics Changzhou University

The Morris-Lecar model is a neuronal model designed to replicate the oscillatory activity of barnacle giant muscle fibers. This work presents a three-dimensional Morris-Lecar model with a novel fast-slow structure. The model introduces a current containing slow variables, thereby forming a new fast-slow structure with the membrane potential. Numerical results reveal the existence of two independent unstable foci and complex dynamical behaviors that include period-doubling bifurcations and spiking/bursting activities. Two types of bifurcation mechanisms for bursting activities are elucidated through theoretical analysis based on four typical bursting activities. Finally, a digital neuronal circuit was developed based on FPGA, and experimental results show good consistency with numerical results.

关键词： Morris-Lecar model fast-slow structure firing activity bifurcation mechanism digital neuronal circuit

来源：评论

学校读者我要写书评

暂无评论

RD-Net: Residual-Dense Network for Glaucoma Prediction Using Structural Features of Optic Nerve Head

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2025年第1期6卷 107-117页

作者： Preity Bhandari, Ashish Kumar Jha, Akanksha Shahnawazuddin, Syed National Institute of Technology Patna Department of Electronics and Communication Engineering Patna800005 India

Glaucoma is called as the silent thief of eyesight. It is related to the internal damage of optical nerve head (ONH). For early screening, the simplest way is to analyze the subtle variations in structural features such as cup to disc ratio (CDR), disc damage likelihood scale (DDLS), rim width of the inferior, superior, nasal, and temporal (ISNT) regions of ONH. This can be done by accurate segmentation of optic disc (OD) and optic cup (OC). In this work, we have introduced a deep learning framework, called residual dense network (RD-NET), for disc and cup segmentation. Based on the segmentation results, the structural features are calculated. The proposed design differs from the traditional U-Net in that it utilizes filters with variable sizes and an alternative optimization method throughout the up- and down-sampling processes. The introduced method is a hybrid deep learning model that incorporates dense residual blocks and squeeze excitation block introduced within the conventional U-Net architecture. Unlike the classical approaches that are primarily based on CDR calculation, in this work, we first segment OD and OC using RD-Net and then analyze ISNT and DDLS. Once a suspicious case is detected, we then go for CDR calculation. In addition to developing an efficient segmentation model, six distinct kinds of data augmentation techniques have been also used in this study to increase the amount of training data. This, in turn, leads to a better estimation of model parameters. The model is rigorously trained and tested on four benchmark datasets namely DRISHTI, RIMONE, ORIGA, and REFUGE. Subsequently, the structural parameters are calculated for glaucoma prediction. The average accuracies are observed to be 0.9940 and 0.9894 for OD and cup segmentation, respectively. The extensive experiments presented in this article show that our method outperforms other existing state-of-the art algorithms. © 2024 IEEE.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

MTBC: Masked Vision Transformer and Brownian Distance Covariance Classifier for Cross-Domain Few-Shot HRRP Recognition

引用

IEEE Sensors Journal 2025年第9期25卷 16440-16454页

作者： Li, Shuai Li, Wentao Huang, Pengjun Zheng, Muhai Tian, Biao Xu, Shiyou Sun Yat-sen University School of Electronics and Communication Engineering Shenzhen518107 China

High-resolution range profile (HRRP), with its concise vector representation and rich structural information, has become the primary focus of research in the field of radar automatic target recognition (RATR) in the sensor community. In actual combat situations, due to limitations in sensing conditions, acquiring data from noncooperative targets is challenging. Consequently, identifying HRRP targets in the absence of sufficient data is a significant and urgent issue that requires immediate attention. In addition, the precise target recognition represents a pressing hurdle. To this end, this article proposes a masked Vision Transformer (ViT) and masked vision transformer and masked vision transformer and Brownian distance covariance classifier (MTBC). ViT is initially pretrained through a self-supervised learning strategy of masked reconstruction, which enables the model to gain a comprehensive understanding of the intricate interconnections between neighboring blocks in HRRP, thereby facilitating the extraction of more nuanced and discriminative features. In addition, within the context of prototype learning in few-shot metric learning, a Brownian distance covariance (BDC) metric-based classifier is devised. This combines a parametric feature extractor and a nonparametric metric classifier, enabling the accurate description of the dependency between support class prototypes and query vectors. Experimental results validated on the measured HRRP dataset demonstrate that the proposed method performs superiorly in few-shot HRRP target recognition task. © 2001-2012 IEEE.

关键词： MIMO radar

来源：评论

学校读者我要写书评

暂无评论

Experimental Validation of Induction Heating as an Energy-Efficient Solution for Concrete Curing

引用

IEEE Access 2025年 13卷 79237-79246页

作者： Kim, Dong-Jin Ahn, Chi-Hyung Korea University of Technology and Education School of Electrical Electronics and Communication Engineering Dongnam-gu Cheonan-si31253 Korea Republic of

In this paper, we propose an innovative, environmentally friendly, and efficient technology for concrete curing through induction heating. This technology is complemented by a comprehensive design methodology aimed at achieving uniform heating across the formwork surface. Induction heating concrete curing is a new approach based on the principle of electromagnetic induction for the purpose of concrete curing, and it offers improved heating efficiency and safety in formwork heating. However, the heating range of the formwork surface is limited depending on the induction heating coil. Thus, we introduce a coil structure derived from evaluations of formwork’s surface temperature distribution through finite element analysis simulations. Heating experiments were conducted on formwork without concrete and formwork with concrete, using a coil designed through simulation. In the experiment without concrete, a uniform temperature distribution was confirmed with a small temperature deviation of 4.28°C. For the experiment with concrete, induction heating experiments and steam curing experiments were conducted at low and room temperatures to compare surface temperature distribution, power consumption, and concrete strength. The experimental results show that room temperature induction heating has 11.39% higher strength and 41.86% less power consumption than low temperature induction heating curing. It also shows 3.51% higher strength and 30.56% less power consumption than steam curing. Room temperature induction heating curing is superior to steam curing in terms of strength and power consumption. Also, although low temperature induction heating curing has lower performance than steam curing, the difference is not significant, so it can be a feasible alternative. © 2013 IEEE.

关键词： Induction heating

来源：评论

学校读者我要写书评

暂无评论

AN 8-WAY RF FEED NETWORK FOR 5.3 GHZ C-BAND SAR: DESIGN, SIMULATION, AND ANALYSIS USING STANDARD METHODS

引用

Telecommunications and Radio engineering (English translation of Elektrosvyaz and Radiotekhnika) 2025年第5期84卷 21-43页

作者： Kothapudi, Venkata Kishore Thokala, Niharika Vallabhaneni, Ramyasri Sanaka, Poojitha Kumar, Vijay Pappula, Lakshman Gottapu, Srinivasa Kiran Bhattacharyya, Arka Guntur District Andhra Pradesh Vadlamudi522213 India Department of Communication Engineering School of Electronics Engineering Vellore Institute of Technology Tamil Nadu Vellore632014 India Department of Electrical Electronics and Communication Engineering GITAM School of Technology GITAM University Andhra Pradesh Visakhapatnam530045 India School of Science and Engineering Engineering and Computer Science Department Benedict College 1600 Harden St ColumbiaSC29204 United States

This study proposes the design and analysis of an eight-way power divider for unequal division at 5.3 GHz for C-band frequency. Many transmission line pieces make up the current power divider. These transmission lines were printed on a substrate material (RT/duroid 5880, 1.6 mm thick, permittivity 2.2, and loss tangent 0.0002). The eight-way power divider’s unequal amplitudes are constructed using Taylor’s 1-parameter method to achieve the side projection levels –17 dB down as from the main beam. The proposed power divider consists of using two transmission lines with a 3/4 frequency connected to branched microstriplines, providing excellent isolation and matching between the power divider’s two output ports. This power divider is composed of three stages. A 1:2 power divider is used in the first stage, a 1:4 power divider is used in the second stage, and a 1:8 power divider is used in the third stage. According to the simulation results, 1:2 delivers half of the input power, 1:4 produces one-fourth of the power, and 1:8 delivers one-eighth of the input power. All of these assessments were done with the advanced design system (ADS) simulation program. According to both simulated and measured data, the return loss exceeds –35 dB, and the isolation across the output terminals exceeds –40 dB. The proposed circuit operates effectively at C-band frequencies in terms of return loss and isolation (4 GHz and 8 GHz). © 2025 by Begell House, Inc.

关键词： Integrated circuit design

来源：评论

学校读者我要写书评

暂无评论

RNN with High Precision and Noise Immunity:A Robust and Learning-Free Method for Beamforming

引用

IEEE Internet of Things Journal 2025年第11期12卷 15779-15791页

作者： Lin, Cong Jiang, Zhihui Cong, Jingyu Zou, Lilan Guangdong Ocean University School of Electronics and Information Engineering Zhanjiang524088 China Sun Yat-sen University School of Electronics and Communication Engineering Shenzhen518107 China

Recurrent Neural Networks (RNNs), recognized for their high accuracy and strong robustness. However, the adoption of RNN-based solutions for array signal beamforming is still in its infancy, as RNNs are very sensitive to noise and cannot easily overcome the impact of environmental noise on the solution. To address these limitations, this study proposes the Dynamic Integrated Enhanced Neural Network (DIENN) for array signal beamforming, which incorporates an error integral feedback mechanism. This mechanism enhances the robustness and noise immunity of the model, enabling it to maintain stable performance under dynamic noise environments. Compared with state-of-the-art methods, the proposed model has higher stability in beamforming tasks while providing excellent results under three interference conditions where the other algorithms of the comparison failed. The residual accuracy achieved in the case of time-varying disturbance was 10-15. The feasibility of the model was verified by applying it to experimental data. To our knowledge, this is the first work to develop a zero-reset RNN for array signal processing. © 2014 IEEE.

关键词： Beamforming

来源：评论

学校读者我要写书评

暂无评论

Securing Telehealth Electronic Health Records with Adaptive Chaotic Encryption and Dynamic Key Generation

引用

SN Computer Science 2025年第5期6卷 1-19页

作者： Leelavathi, Venkatala Venkatappa Rashmi Priyadarshini, Bajanemane Krishnamurthy Puttegowda, Kiran School of Electronics and Communication Engineering REVA University Karnataka Bengaluru India Department of Electronics and Communication Engineering Vidyavardhaka College of Engineering Karnataka Mysuru India Department of Electronics and Communication Engineering Government Engineering College Chamarajanagara Karnataka India

The increasing use of electronic health records (EHRs) has led to a growing concern about the security of sensitive medical information. Medical images hold critical diagnostic and therapeutic information about patients, making their confidentiality essential, especially in telehealth contexts. Data encryption is essential for protecting sensitive information, especially in the realm of image encryption, which is vital for safeguarding visual data. This paper proposes a new encryption algorithm of EHR medical images using chaotic Henon map systems with self-adaptive approach. This approach creates as many correlated security keys as there are elements contained in the medical images delivered by EHR by means of Plaintext-Associated Key Generation (PAKG). This self-adaptive framework guarantees that the generated key revolves entirely around the characteristics of the plaintext image. For the pixel confusion phase, a block-based generalization technique along with multi-level map is used and for the diffusion phase, the threshold-based generalization technique involving the chaotic Henon map systems is used. Through this method, medical image encryption is made possible with low computational power required thus striking more of the balance between the two factors, efficiency and security. Our presented scheme has been implemented and simulated over standard EHR images;the security aspects have been weighed and measured in terms of entropy, key sensitivity, key space, correlation, and histogram analysis. Finally, comparing the results of our scheme with other existing methods it proven that our scheme performs substantially better in case of similar conditions. As for the strengths of our approach, it is necessary to mention that security keys are generated dynamically for each of the input medical images and the information of the image itself is considered for-key creation. The entropy values closely approximate optimal threshold value 8, affirming the efficiency o

关键词： Diffusion EHRs Encryption Henon map Medical images Permutation

来源：评论

学校读者我要写书评

暂无评论

Attention deficit hyperactivity disorder subtypes classification: a machine learning approach with phenotypic information and brain tissue volume

引用

Multimedia Tools and Applications 2025年第14期84卷 12901-12924页

作者： Usha Rupni, K. Aruna Priya, P. Department of Electronics and Communication Engineering College of Engineering and Technology SRM Institute of Science and Technology Chengalpattu District Tamil Nadu Kattankulathur603203 India

Attention deficit hyperactivity disorder (ADHD) is a critical neurodevelopmental disorder that needs to be diagnosed and treated early to lower the risk of related health issues. This research study uses a machine-learning (ML) classification model to provide a computer-aided diagnosis method for ADHD subtypes namely ADHD-inattentive and ADHD-combined. The brain tissue volume and phenotypic information of the children are used to train the ML classification model. The grey matter and white matter brain tissues are segmented from T1-weighted brain MRI of children using a modified fuzzy c-means clustering algorithm. A novel thresholding and pixel-based volume calculation method generates volume for segmented tissues. The highest accuracy of 92.98% is achieved for classifying ADHD subtypes and typically developing (TD) with Extreme Gradient Boosting (XGBoost) classifier among the other ML classifiers. An interpretative approach provides the insight of the classification model and it predicts that medication status, intelligence quotient, gender, and grey matter volume that are used in this research are the key factors in distinguishing between ADHD subtypes and TD individuals. In conclusion, T1-weighted MRI brain tissue volume of children can help healthcare providers diagnose ADHD and its subtypes along with the symptom-based diagnosis. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.

关键词： Magnetic resonance imaging

来源：评论

学校读者我要写书评

暂无评论

Reinforcement Learning Techniques for Real Time Battery Cell Balancing Using Modified Cascaded H-Bridge Converters

引用

Tehnicki Vjesnik 2025年第3期32卷 1010-1017页

作者： Hemalatha, M. Sampoornam, K.P. Department of Electrical and Electronics Engineering Bannari Amman Institute of Technology India Department of Electronics and Communication Engineering Bannari Amman Institute of Technology India

This study presents an innovative approach to individual battery cell balancing, focusing on the implementation of a modified cascaded H-Bridge (MCHB) multi-level converter utilizing a reinforcement learning (RL) protocol. In battery systems, particularly those employing extended-lifecycle cells that exhibit intrinsic mismatches, achieving effective cell balancing is crucial for enhancing overall performance and lifespan. The proposed design connects each converter module directly to individual battery cells, facilitating precise balancing without the need for additional complex circuitry. This approach is especially beneficial for applications involving elongated facility-lifecycle grid storage, where optimal performance and reliability are paramount. The research employs an advanced control technique driven by reinforcement learning algorithms to dynamically regulate the interaction between the Battery State of Charge (SoC) and the outputs of the modified cascaded H-Bridge. By continuously adapting to varying operational conditions, the RL protocol ensures that each cell is charged and discharged optimally, minimizing the effects of imbalance and maximizing the overall efficiency of the battery system. The implementation of this technique allows for real-time monitoring and adjustment, significantly improving the response time and accuracy of the balancing process. In the context of grid storage applications, this novel methodology not only enhances the longevity of battery cells but also contributes to more sustainable energy management practices. The experimental results demonstrate that the RL-based MCHB converter significantly outperforms traditional balancing methods, offering a promising solution for future battery management systems. This work paves the way for the development of more efficient energy storage solutions, ultimately supporting the transition to cleaner energy sources and advancing the field of energy technology. Further research will focus on

关键词： State of charge

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：