检索结果-内蒙古大学图书馆

Explainable Artificial Intelligence Driven Segmentation for Cervical Cancer Screening

IEEE Access 2025年 13卷 71306-71322页

作者： Sritharan, Niruthikka Gnanavel, Nishaanthini Inparaj, Prathushan Meedeniya, Dulani Yogarajah, Pratheepan University of Moratuwa Department of Computer Science and Engineering Moratuwa10400 Sri Lanka Ulster University School of Computing Engineering and Intelligent Systems LondonderryBT48 7JL United Kingdom

Cervical cancer remains an important global health challenge among women. Early and accurate identification of abnormal cervical cells is crucial for effective treatment and improved survival rates. This paper addresses the development of a novel weakly supervised segmentation framework that combines binary classification, Explainable Artificial Intelligence (XAI) techniques, and GraphCut to automate cervical cancer screening. Unlike traditional segmentation methods that rely on pixel-level annotations of medical images, which are costly, laborious, and require expertise in medical imaging, our approach leverages classification-driven insights to segment the nucleus, cytoplasm, and background regions. A key innovation of our framework is the use of XAI techniques such as Grad-CAM++ and LRP combined with GraphCut, to enable annotation-free segmentation using only classification-level labels. This represents a pioneering application of explainability techniques in the context of cervical cancer screening. Among the classification models explored, including fine-tuned variants of VGGNet and XceptionNet, VGG16-Adapted128 achieved the highest performance, marked by an accuracy of 0.94, precision of 0.94, recall of 0.94, and an F1 score of 0.94. This novel segmentation framework employed LRP and GradCAM++ as XAI techniques to gain insight into the decision-making process of classification models, with GradCAM++ demonstrating greater effectiveness. The performance of these XAI methods was assessed through both visual inspection and quantitative metrics, including entropy and pixel flipping. This innovative approach to segmentation is formally introduced through two algorithms detailed in this paper. The weakly supervised segmentation framework achieved a Dice Similarity Coefficient (DSC) of 62.05% and an Intersection over Union (IoU) of 61.89%. In addition, it has received high satisfaction ratings from expert evaluations and has been seamlessly integrated into a user-frie

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Deep learning-based method for detection of copy-move forgery in videos

引用

Neural Computing and Applications 2025年第14期37卷 8451-8464页

作者： Elbarougy, Reda Abdelfatah, Osama Behery, G.M. El-Badry, Noha M. Department of Artificial Intelligence and Data Science College of Computer Science and Engineering University of Ha’il Ha’il81481 Saudi Arabia Department of Information Technology Faculty of Computer and Artificial Intelligence Damietta University New Damietta Kafr Saad Egypt Department of Computer Science Faculty of Computer and Artificial Intelligence Damietta University New Damietta Kafr Saad Egypt Department of Mathematics Faculty of Science Damietta University New Damietta Kafr Saad Egypt

Video forgery is one of the most serious problems affecting the credibility and reliability of video content. Therefore, detecting video forgery presents a major challenge for researchers due to the diversity of forgery types, the modernity of the programs used in forgery operations, and the abundance of information and content present in videos. The seriousness of this issue arises from the widespread use of videos in vital fields that require very high accuracy with no room for doubt or error, such as courtrooms, journalism, and others. Copy-move forgery is one of the most common and dangerous types of video forgery because of the difficulty in identifying it with the naked eye and the great diversity of forgery techniques involved in this particular type. One of the biggest challenges researchers have faced in the past is the complexity of the steps required to detect forgery in videos, leading to significant computational complexity and time consumption. The proposed method also aims to achieve better results than previous methods while reducing computational operations. Ultimately, forgery is detected with great efficiency. Compared to previous methods used for detecting copy-move forgery in the Rewind dataset, the proposed method achieves the highest F1, reaching 0.86, with a significant difference of 0.13 compared to the best result of the previous methods. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Steady-State Solution ofMHD Flow with Induced Magnetic Field

引用

Fluid Dynamics & Materials Processing 2025年第2期21卷 233-252页

作者： Saykat Poddar JuiSaha Badhan Neogi Mohammad Sanjeed Hasan Muhammad Minarul Islam Giulio Lorenzini Md.Mahmud Alam Department of Mathematics Faculty of ScienceBangabandhu Sheikh Mujibur Rahman Science and Technology UniversityGopalganj8100Bangladesh Department of Industrial Systems and Technologies Engineering University of ParmaParma43124Italy Mathematics Discipline ScienceEngineering and Technology SchoolKhulna UniversityKhulna9208Bangladesh

This study presents a numerical analysis of the steady-state solution for transient magnetohydrodynamic(MHD)dissipative and radiative fluid flow,incorporating an inducedmagnetic field(IMF)and considering a relatively high concentration of foreign mass(accounting for Soret and Dufour effects)over a vertically oriented semi-infinite *** governing equations were normalized using boundary layer(BL)*** resulting nonlinear system of partial differential equations(PDEs)was discretized and solved using an efficient explicit finite difference method(FDM).Numerical simulations were conducted using MATLAB R2015a,and the developed numerical code was verified through comparison with another code written in FORTRAN *** ensure the reliability of the results,both mesh refinement and steady-state time validation tests were ***,a comparison with existing published studies was made to confirm the accuracy of the *** dimensionless equations revealed the impacts of several key *** IMF initially intensifies near the plate before gradually diminishing as the magnetic parameter *** the range 0≤y≤1.8(where y is the horizontal direction),the IMF decreases with a rise in the magnetic Prandtl number;however,for 1.8≤y≤7(approximately),the magnetic field begins to *** this,the profile of the magnetic field becomes somewhat irregular through the remaining part of the BL.

关键词： MHD stability analysis finite difference method(FDM)

来源：评论

学校读者我要写书评

暂无评论

A survey of artificial intelligence models for wireless capsule endoscopy videos for superior automatic diagnosis: problems and solutions

引用

Multimedia Tools and Applications 2025年 1-35页

作者： El-Gammal, Eman M. El-Shafai, Walid Taha, Taha E. El-Fishawy, Adel S. Abd El-Samie, Fathi E. Department of Electronics and Electrical Communications Engineering Faculty of Electronic Engineering Menoufia University Menouf32952 Egypt Center of Advanced Software and Biomedical Clinical Engineering Consultations Faculty of Engineering Cairo University Giza Egypt Computer Science Department Prince Sultan University Riyadh11586 Saudi Arabia Department of Information Technology College of Computer and Information Sciences Princess Nourah Bint Abdulrahman University 84428 Riyadh11671 Saudi Arabia

Wireless Capsule Endoscopy (WCE) emerged as an innovative and patient-centric approach for non-invasive and painless examination of the gastrointestinal (GI) tract. It serves as a pivotal tool in helping medical practitioners in the early detection of anomalies within the intricate domain of the human GI tract. The automated identification of aberrations assumes paramount significance, contributing not only to temporal efficiency but also to timely diagnosis. Within the realm of academic literature, a plethora of Artificial Intelligence (AI) methodologies have been proposed to effectuate the automatic classification, segmentation, and synthesis of anomalies inherent in WCE images. This scholarly endeavor undertakes a comprehensive and meticulous survey, elucidating the spectrum of anomaly classification, summarization, and detection techniques harnessed for computer-Aided Diagnosis (CAD) in the context of WCE images. The survey begins by delineating the methodologies underpinning WCE image classification and video processing. Subsequently, an exhaustive evaluation of techniques for analyzing WCE images is furnished, accompanied by an in-depth exploration of pertinent surveys, in addition to a judicious assessment of their merits and demerits. Furthermore, this study undertakes a rigorous evaluation of prevailing datasets utilized to benchmark the efficacy of various WCE techniques. In summary, this survey serves as a vanguard, delineating promising avenues for future investigations in the realm of WCE image analysis models. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2025.

关键词： Endoscopy

来源：评论

学校读者我要写书评

暂无评论

ADNeuroNet: a neuroevolution-based neural network algorithm for the diagnosis of neurodegenerative diseases

引用

Neural Computing and Applications 2025年第14期37卷 8593-8627页

作者： Khan, Afreen Zubair, Swaleha Ali, Irfan Department of Computer Application Faculty of Engineering & IT Integral University Lucknow226026 India Department of Statistics and Operations Research Aligarh Muslim University Aligarh India Department of Computer Science Aligarh Muslim University Aligarh India

Neurodegenerative disorders such as dementia and Alzheimer’s disease (AD) have adversely devastated the health and well-being of the older community. Given that early detection might help prevent or delay cognitive disorders associated with AD, developing an effective diagnostic technique is deemed appropriate to control the disease. Despite advances in clinical diagnostic standards and treatment techniques, the global prevalence of cognitive disturbance and behavioral problems remains steep. Besides the above challenge, the scarcity of AD-related open-source raw data prompted us to tailor ADNeuroNet, a neuroevolution-based neural network (NN) designed for predicting cognitively normal, mild cognitive impairment, and AD instances. To construct a robust NN algorithm that can predict the disease with high accuracy and improved efficiency, we worked on the combined cognitive and demographic clinical data to create a comprehensive model. Primarily, the Alzheimer’s Disease Neuroimaging Initiative (ADNI) repository was used to develop this program. We employed three distinct datasets and developed a predictive model that achieved the maximum performance accuracy of 93.42%. Utilizing only baseline details, the as-developed model successfully diagnosed and predicted neurodegenerative disorders and is likely to emerge as an effective clinical tool in future endeavors. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.

关键词： Disease control

来源：评论

学校读者我要写书评

暂无评论

High quality dehazed image and video based on enhanced multi-scale guided filtering dehazing technique

引用

Cluster Computing 2025年第5期28卷 1-24页

作者： Ayoub, Abeer El-Shafai, Walid El-Samie, Fathi E. Abd Hamad, Ehab K. I. El-Rabaie, El-Sayed M. Department of Electronics and Electrical Communications Engineering Faculty of Electronic Engineering Menoufia University Menouf32952 Egypt Computer Science Department Prince Sultan University Riyadh11586 Saudi Arabia Electrical Engineering Department Faculty of Engineering Aswan University Aswan81542 Egypt Department of Information Technology College of Computer and Information Sciences Princess Nourah bint Abdulrahman University P.O. Box 84428 Riyadh11671 Saudi Arabia

Haze is one of the most common challenges faced by outdoor imaging systems. Hazy images are characterized by low contrast and a white overlay, which make it difficult to produce high-quality results. To address this issue, researchers have developed various haze removal methods aimed at achieving superior image enhancement, efficient processing, and cost-effectiveness. This paper introduces an enhanced Multi-scale Guided Filtering (MGF) technique for dehazing. The proposed technique incorporates several key components: a fast dehazing method, homomorphic processing, and Contrast Limited Adaptive Histogram Equalization (CLAHE) as preprocessing steps, followed by a fuzzy logic approach to enhance contrast. The core of the technique is the MGF dehazing method, which is supported by fast dehazing and fuzzy logic proessing as enhancement strategies to reduce noise and improve the dynamic range of images. The MGF dehazing process gets the transmission map and atmospheric light at the coarsest level, and then guided filtering is employed iteratively to produce a smoothed transmission map, resulting in dehazed images free from artifacts. The haze model is applied to obtain the final dehazed image. This technique effectively combines two traditional approaches—fast dehazing and MGF dehazing—into a unified framework. The performance of the proposed technique is evaluated using visible and Near Infrared (NIR) video frames, as well as real hazy images. To demonstrate its efficacy, its results are compared with those of the standalone MGF dehazing method. Additionally, a comparative analysis is conducted with other dehazing techniques using both visible and NIR frames. Evaluation metrics include both referenced and non-referenced quantitative measures such as Peak Signal-to-Noise Ratio (PSNR), correlation (to measure similarity between hazy and dehazed images), entropy (quantifying the amount of information), Feature Similarity Index (FSIM), Chronic Feature Similarity index (FSI

关键词： Image enhancement

来源：评论

学校读者我要写书评

暂无评论

A Study of Predicting Social Networking Sites(SNS) Media Disorder Survey Data Using a Machine Learning Approach

A Study of Predicting Social Networking Sites(SNS) Media Dis...

引用

2025 International Conference on Multi-Agent systems for Collaborative Intelligence, ICMSCI 2025

作者： Munia, Jerin Akther Tahmiduzzaman, K.B.M. Saad, Izazul Haque Islam, Md Mynul Akash, Abdul Hady Zumma, Md. Thoufiq Daffodil International University Department of Computer Science and Engineering Dhaka Bangladesh University of Technology Sydney Faculty of Engineering and Information Technology Sydney Australia University of Asia Pacific Department of Computer Science and Engineering Dhaka Bangladesh

ISBN: (纸本)9798331509828

This study uses survey data and machine learning algorithms to forecast social media disorder in people. A total of 600 individuals answered questions on their social media usage patterns, internet habits, demographics, and signs of social media disorder. This dataset was used to train a number of machine learning models, such as Gaussian Naive Bayes (GaussianNB), Random Forest, K-Nearest Neighbors (KNN), Decision Tree, and Support Vector Machine (SVM). According to the findings, the GaussianNB model had the greatest accuracy of 97%, followed by SVM (94%), Random Forest (91%), Decision Tree (92%), and KNN (90%). These results demonstrate how well machine learning can identify those who may be at risk for social media disorder. In order to improve prediction performance, future research will concentrate on increasing feature size, diversifying datasets, and refining models. Furthermore, incorporating these models into user-friendly apps and intervention programs may offer useful resources for the early diagnosis and treatment of social media illness. © 2025 IEEE.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

Enhanced Multiple Sound Event Detection and Classification Using Physical Signal Properties in Recurrent Spiking Neural Networks

引用

IEEE Access 2025年 13卷 81312-81325页

作者： Roozbehi, Zahra Narayanan, Ajit Mohaghegh, Mahsa Saeedinia, Samaneh-Alsadat Auckland University of Technology School of Electrical and Computer Engineering Auckland New Zealand Auckland University of Technology Faculty of Design and Creative Technologies Auckland New Zealand Iran University of Science and Technology Department of Electrical and Computer Engineering Tehran Iran

Sound event detection and classification present significant challenges, particularly in noisy environments with multiple overlapping sources. This paper introduces an innovative architecture for multiple sound event detection and classification utilizing recurrent spiking neural networks (SNNs). Our method uniquely leverages temporal data to detect and classify multiple sound sources simultaneously, integrating the physical concept of signal power matching with neuronal output power and employing a binaural strategy to enhance detection accuracy in real-world scenarios. The architecture processes spatiotemporal data to dynamically update synaptic weights, enabling precise identification of sound event categories and their occurrences. Our simulations reveal substantial performance improvements, achieving the highest precision of 73% in classification tasks, including multilayer perceptrons (MLP), convolutional recurrent neural networks (CRNN), and recurrent neural networks (RNN). Statistical analysis indicates that these improvements are significant (p-value © 2013 IEEE.

关键词： Recurrent neural networks

来源：评论

学校读者我要写书评

暂无评论

A Comparison of Deep Learning and Machine Learning Methods for Water Quality Prediction 3

A Comparison of Deep Learning and Machine Learning Methods f...

引用

3rd International Conference on Intelligent Data Communication Technologies and Internet of Things, IDCIoT 2025

作者： Srinivasa Gupta, N. Valarmathi, B. Hiwanj, Ishaan Department of Manufacturing Engineering School of Mechanical Engineering Vellore Institute of Technology Tamil Nadu Vellore India Department of Software and Systems Engineering School of Computer Science Engineering and Information Systems Vellore Institute of Technology Tamil Nadu Vellore India Department of Information Technology School of Computer Science Engineering and Information Systems Vellore Institute of Technology Tamil Nadu Vellore India

ISBN: (纸本)9798331527549

Predicting water quality is essential to preserving human health and environmental sustainability. Traditional water quality assessment methods often face scalability and real-time monitoring limitations. With accuracies of 62%, 72%, 83%, 69%, 63%, 66%, 71%, 63%, and 64%, respectively, the current techniques utilized were Logistic Regression, Decision Trees, Random Forest Regressor, Extreme Gradient Boosting, Naive Bayes, K-nearest neighbors, Support Vector Machine, AdaBoost, and Bagging [9]. This study addresses these challenges by leveraging Adaptive Synthetic Sampling (ADASYN) to balance the dataset and evaluating model performance on datasets of 5,000 and 10,000 entries per class. A robust dataset obtained from Kaggle was used, with five models – Long Short-Term Memory (LSTM), Feed Forward Neural Network (FFNN), Categorical Boosting (CatBoost), Extreme Gradient Boosting (XGBoost), and Random Forest – evaluated and compared. The proposed methods demonstrate significant improvements in accuracy, with XGBoost achieving the highest accuracy of 95.53%, followed by Random Forest at 93.98%. This work underscores the importance of advanced machine learning techniques in addressing the limitations of traditional methods, enhancing accuracy, scalability, and adaptability in water quality prediction. These findings contribute to advancing environmental monitoring and management practices with reliable, data-driven insights. © 2025 IEEE.

关键词： Support vector machines

来源：评论

学校读者我要写书评

暂无评论

Enhancing single-lead electrocardiogram arrhythmia detection with empirical mode decomposition

引用

Neural Computing and Applications 2025年 1-23页

作者： Issa, Mohamed F. Yousry, Ahmed Tuboly, Gergely Wang, Zeyu Zoltan, Juhasz Selim, Mazen M. AbuEl-Atta, Ahmed H. Department of Scientific Computing Faculty of Computers and Artificial Intelligence Benha University Benha13511 Egypt Department of Computer Science Faculty of Computers and Artificial Intelligence Benha University Benha13511 Egypt Department of Electrical Engineering and Information Systems University of Pannonia Veszprem8200 Hungary Department of Mechatronics Delta University for Science and Technology Gamasa35712 Egypt

Cardiac arrhythmias pose a significant challenge to health care, requiring accurate and reliable detection methods to enable early diagnosis and treatment. However, traditional ECG beat classification methods often lack robustness and fail to generalize effectively to diverse clinical scenarios, particularly when using single-lead-I recordings, which are less studied compared to lead II. To address this gap, this paper proposes a novel method that combines deep learning and empirical mode decomposition (EMD) for arrhythmia detection using single-lead-I ECG recordings obtained with electrodes on each hand. The hybrid model comprises 149 layers of one-dimensional convolution, max pooling, and residual blocks. By leveraging EMD, it classifies ECG beats into normal beats, premature ventricular contractions, and atrial premature beats with high sensitivity (99.86%, 99.72%, and 99.84%, respectively) using the first four intrinsic mode functions (IMFs). The proposed method is applicable in clinical and out-of-hospital monitoring scenarios, and an accompanying web application provides real-time analysis and diagnostic results. This approach addresses a critical gap in lead-I ECG analysis, enhancing diagnostic precision and expanding the scope of ECG applications in health care. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.

关键词： Empirical mode decomposition

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：