检索结果-内蒙古大学图书馆

Design and Analysis of deep learning Method for Fragmenting Brain Tissue in MRI images

INTERNATIONAL JOURNAL OF ADVANCED COMPUTER SCIENCE AND APPLICATIONS 2024年第1期15卷 95-111页

作者： Yang, Ting Sun, Jiabao Shaoxing Univ Sch Informat & Elect Engn Yuanpei Coll Shaoxing 312000 Zhejiang Peoples R China

An essential component of medical image processing is brain tumour segmentation. The process of giving each pixel a label is called image segmentation in order forpixels bearing the same label to share characteristics and help distinguish the target. A higher fatality rate and additional dangers can be avoided with early identification. It can be challenging and time-consuming to manually (man-made) segment brain tumours from the numerous MRI pictures generated during medical procedures in order to diagnose malignancy. This is the fundamental reason why brain tumour imaging has to be automated. The deep learning technique for the segmentation of brain tissue in magnetic resonance imaging (MRI) pictures was examined and enhanced in this work. Researchers are using deep learning techniques-convolutional neural networks in particular-to tackle the complex problem of biological image fragmentation object recognition. In contrast to traditional classification techniques that take in manually constructed qualities, convolutional neural networks automatically extract the required complicated features from the data itself. This solves a number of problems

关键词： Brain tumor deep learning neural networks magnetic resonance imaging

来源：评论

学校读者我要写书评

暂无评论

Predicting real-time Spectra from High-Speed Imaging: An Ultrafast Machine Vision Framework for Online Optical Control in Microfluidics

引用

ADVANCED MATERIALS TECHNOLOGIES 2024年第15期9卷

作者： Zhang, Xucheng Guo, Jia Wang, Changchun Fudan Univ Dept Macromol Sci State Key Lab Mol Engn Polymers 2005 Songhu Rd Shanghai 200438 Peoples R China

Automated laboratory pipelines empowered by the advent of artificial intelligence (AI) have been reshaping the paradigm of chemical and material discovery. While intelligent decision-making strategies have been developed that enable fast convergence to optimal reaction conditions, the overall efficiency is still bottlenecked by the slow acquisition of online characterization units, especially in microfluidics-based systems. In response, an ultrafast machine vision framework is hereby developed to directly predict optical spectra from real-time high-speed imaging in this article. Utilizing uniquely engineered deep-learning algorithms, this framework automatically detects droplets and recapitulates high-quality spectra from embedded image features in a sub-millisecond timeframe. In a proof-of-concept example, droplets are detected with an average precision of 99.75% within 0.51 ms, and their spectra spanning the entire visible wavelengths are predicted with an average peak error of 5.32 nm, which is equivalent to 1.6% of the visible range. The work demonstrates the first example of an all-imaging-based machine learning framework deployed in a real-time environment for the full replacement of an analytical instrument. It highlights the broader possibility of AI in material researches to break the physical limitations of online diagnostics and to further boost efficiencies in self-driving laboratories.

关键词： deep learning machine vision microfluidics self-driving laboratories spectroscopy

来源：评论

学校读者我要写书评

暂无评论

Fluorescence excitation-scanning hyperspectral imaging with scalable 2D-3D deep learning framework for colorectal cancer detection

引用

SCIENTIFIC REPORTS 2024年第1期14卷 1-16页

作者： Oswald, Willaim Browning, Craig Yasmin, Ruthba Deal, Joshua Rich, Thomas C. Leavesley, Silas J. Gong, Na Univ S Alabama Dept Elect & Comp Engn Mobile AL 36688 USA Univ S Alabama Dept Syst Engn Mobile AL 36688 USA Univ S Alabama Dept Chem & Biomol Engn Mobile AL 36688 USA Nikon Instruments Melville NY 11747 USA Univ S Alabama Dept Pharmacol Mobile AL 36688 USA Univ S Alabama Ctr Lung Biol Mobile AL 36688 USA

Colorectal cancer is one of the top contributors to cancer-related deaths in the United States, with over 100,000 estimated cases in 2020 and over 50,000 deaths. The most common screening technique is minimally invasive colonoscopy using either reflected white light endoscopy or narrow-band imaging. However, current imaging modalities have only moderate sensitivity and specificity for lesion detection. We have developed a novel fluorescence excitation-scanning hyperspectral imaging (HSI) approach to sample image and spectroscopic data simultaneously on microscope and endoscope platforms for enhanced diagnostic potential. Unfortunately, fluorescence excitation-scanning HSI datasets pose major challenges for data processing, interpretability, and classification due to their high dimensionality. Here, we present an end-to-end scalable Artificial Intelligence (AI) framework built for classification of excitation-scanning HSI microscopy data that provides accurate image classification and interpretability of the AI decision-making process. The developed AI framework is able to perform real-time HSI classification with different speed/classification performance trade-offs by tailoring the dimensionality of the dataset, supporting different dimensions of deep learning models, and varying the architecture of deep learning models. We have also incorporated tools to visualize the exact location of the lesion detected by the AI decision-making process and to provide heatmap-based pixel-by-pixel interpretability. In addition, our deep learning framework provides wavelength-dependent impact as a heatmap, which allows visualization of the contributions of HSI wavelength bands during the AI decision-making process. This framework is well-suited for HSI microscope and endoscope platforms, where real-time analysis and visualization of classification results are required by clinicians.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Survey of Incremental deep learning for Defect Detection in Manufacturing

引用

BIG DATA AND COGNITIVE COMPUTING 2024年第1期8卷 7页

作者： Mohandas, Reenu Southern, Mark O'Connell, Eoin Hayes, Martin Univ Limerick Dept Elect & Comp Engn Limerick V94 T9PX Ireland Univ Limerick Sch Engn Limerick V94 T9PX Ireland Univ Limerick Enterprise Res Ctr Limerick V94 T9PX Ireland

deep learning based visual cognition has greatly improved the accuracy of defect detection, reducing processing times and increasing product throughput across a variety of manufacturing use cases. There is however a continuing need for rigorous procedures to dynamically update model-based detection methods that use sequential streaming during the training phase. This paper reviews how new process, training or validation information is rigorously incorporated in real time when detection exceptions arise during inspection. In particular, consideration is given to how new tasks, classes or decision pathways are added to existing models or datasets in a controlled fashion. An analysis of studies from the incremental learning literature is presented, where the emphasis is on the mitigation of process complexity challenges such as, catastrophic forgetting. Further, practical implementation issues that are known to affect the complexity of deep learning model architecture, including memory allocation for incoming sequential data or incremental learning accuracy, is considered. The paper highlights case study results and methods that have been used to successfully mitigate such real-time manufacturing challenges.

关键词： deep learning incremental learning continuous learning catastrophic forgetting self-healing processes defect detection concept drift

来源：评论

学校读者我要写书评

暂无评论

Surface inspection of annular ceramic metal coating based on small sample data augmentation and deep learning

引用

SIGNAL image AND VIDEO processing 2025年第1期19卷 1-10页

作者： Chu, Jia Le Liang, Dan Xi, Yong Long Ye, Min Jie Ningbo Univ Part Rolling Key Lab Zhe Jiang Prov 818 Fenghua Rd Ningbo 315211 Peoples R China Ningbo Univ Ningbo Key Lab Micronano Mot & Intelligent Control 818 Fenghua Rd Ningbo 315211 Peoples R China

The defects in the metal coating surface of annular ceramic workpiece have significant effects on the conductivity and reliability. Due to the irregularity, small area, and few sample number of defects, it is difficult to achieve efficient and accurate inspection. This paper presents a defect inspection framework based on deep learning for the metal coating surface of annular ceramic workpiece. Firstly, an image acquisition system for the coating surface is designed, and the defects characteristics are analyzed. Then, a surface image data set is constructed through five data augmentation strategies in order to solve the problem of insufficient samples. Finally, a defect detection framework based on improved YOLOv7 for Ceramic Metal Coating (YOLOV7-CMC) is established. By optimizing the clustering algorithm of target box, introducing an attention mechanism, and improving the MPConv structure, the efficient and precise identification of different defects is realized. Experimental results show that the recognition rate of defects including scratch, deficiency, scuffing, and dot is higher than 94%, and the average detection time is about 30 ms. The proposed detection framework based on deep learning shows great application potential in the fields of precise coating and manufacturing of ceramic materials.

关键词： Annular ceramic Metal coating Defect inspection Data augmentation

来源：评论

学校读者我要写书评

暂无评论

real-time deep learning Based image Compression Techniques: Review 6th

Real-Time Deep Learning Based Image Compression Techniques: ...

引用

6th International conference on communication and computational technologies, ICCCT 2024

作者： Abdulredah, Ali A. Kherallah, Monji Charfi, Faiza National School of Electronics and Telecoms of Sfax University of Sfax Sfax Tunisia Faculty of Sciences of Sfax University of Sfax Sfax Tunisia College of Computer Science and Information Technology University of Sumer Thi Qar Iraq

ISBN: (纸本)9789819774227

The emergence of deep learning techniques has solved many image processing problems using traditional methods. It has provided pioneering solutions, especially in image compression, for the urgent need for storage and transmission. This paper aims to review modern techniques that use image compression using several neural networks and deep learning methods. These networks have shown promising results in complex cognitive tasks by providing high compression ratios while maintaining visual image quality. However, this field lacks further exploration and testing to evaluate the effectiveness of deep learning across different types of images, especially in medical images, which has its own challenges and requirements. Therefore, image compression has become extremely important. In this article, we begin with an overview of the basics of image compression. A brief introduction to the types of networks based on deep learning, then a comprehensive summary of previous literature, and finally, we discuss prospects for image compression methods based on deep learning. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Contrastive learning

来源：评论

学校读者我要写书评

暂无评论

Prediction of Antenna Performance based on Scalable Data-informed Machine learning Methods

引用

APPLIED COMPUTATIONAL ELECTROMAGNETICS SOCIETY JOURNAL 2024年第4期39卷 275-290页

作者： Chen, Yiming Demir, Veysel Bhupatiraju, Srirama Elsherbeni, Atef Z. Gavilan, Joselito Stoynov, Kiril Colorado Sch Mines Dept Elect Engn Golden CO 80401 USA Dept Elect Engn Northern Illinois Univ Dept Elect Engn De Kalb IL 60115 USA Antenna Grp Nvidia SC 95050 USA

- This paper proposes a scalable architecture for predicting antenna performance using various data- informed machine learning (DIML) methods. By utilizing the computation power of graphics processing units (GPUs), the architecture takes advantage of hardware (HW) acceleration from the beginning of electromagnetic (EM) full-wave simulation to the final machine learning (ML) validation. A total of 49152 full-wave simulations of a classical microwave patch antenna forms the ML dataset. The dataset contains the performance of patch antenna on six commonly used materials and two standard thicknesses in a wide frequency range from 0.1 to 20 GHz. A total of 13 base ML models are stacked and ensembled in a tabular workflow with performance as 0.970 and 0.933 F 1 scores for two classification models, as well as 0.912 and 0.819 R 2 scores for two regression models. Moreover, an image-based workflow is proposed. The image-based workflow yields the 0.823 R 2 score, indicating a near real-time prediction for all S 11 values from 0.1 to 20 GHz. The proposed architecture requires neither the fine-tuned hyperparameters in the ML-assisted optimization (MLAO) model for specified antenna design nor the pre-knowledge required in the physics-informed models. The fully automated process with data collection and the customized ML pipeline provides the architecture with robust scalability in future work where more antenna types, materials, and performance requirements can be involved. Also, it could be wrapped as a pre-trained ML model as a reference for other antenna designs.

关键词： Index Terms - Data informed ensemble full-wave sim- ulation machine learning scalability stacking

来源：评论

学校读者我要写书评

暂无评论

Sensorless End-to-End Freehand 3-D Ultrasound Reconstruction With Physics-Guided deep learning

引用

IEEE TRANSACTIONS ON ULTRASONICS FERROELECTRICS AND FREQUENCY CONTROL 2024年第11期71卷 1514-1525页

作者： Dou, Yimeng Mu, Fangzhou Li, Yin Varghese, Tomy Univ Wisconsin UW Sch Med & Publ Hlth Dept Med Phys Madison WI 53705 USA UW Madison Dept Elect & Comp Engn Madison WI 53706 USA UW Madison Dept Comp Sci Madison WI 53706 USA Nvidia Corp Santa Clara CA 95051 USA UW Sch Med & Publ Hlth Dept Biostat & Med Informat Madison WI 53726 USA

Three-dimensional ultrasound (3-D US) imaging with freehand scanning is utilized in cardiac, obstetric, abdominal, and vascular examinations. While 3-D US using either a "wobbler" or "matrix" transducer suffers from a small field of view and low acquisition rates, freehand scanning offers significant advantages due to its ease of use. However, current 3-D US volumetric reconstruction methods with freehand sweeps are limited by imaging plane shifts along the scanning path, i.e., out-of-plane (OOP) motion. Prior studies have incorporated motion sensors attached to the transducer, which is cumbersome and inconvenient in a clinical setting. Recent work has introduced deep neural networks (DNNs) with 3-D convolutions to estimate the position of imaging planes from a series of input frames. These approaches, however, fall short for estimating OOP motion. The goal of this article is to bridge the gap by designing a novel, physics-inspired DNN for freehand 3-D US reconstruction without motion sensors, aiming to improve the reconstruction quality and, at the same time, to reduce computational resources needed for training and inference. To this end, we present our physics-guided learning-based prediction of pose information (PLPPI) model for 3-D freehand US reconstruction without 3-D convolution. PLPPI yields significantly more accurate reconstructions and offers a major reduction in computation time. It attains a performance increase in the double digits in terms of mean percentage error, with up to 106% speedup and 131% reduction in graphic processing unit (GPU) memory usage, when compared to the latest deep learning methods.

关键词： Three-dimensional displays image reconstruction Speckle Imaging Correlation Tracking Decorrelation deep learning medical image processing ultrasonic imaging

来源：评论

学校读者我要写书评

暂无评论

Enhanced Approach to Study the Neural Stem Cell Migration through image processing using deep learning

Enhanced Approach to Study the Neural Stem Cell Migration th...

引用

International Conference on Power, Energy, Control and Transmission Systems (ICPECTS)

作者： M. Vaithiyanathan K. Bala Department of Computer Science and Engineering Bharath Institute of Higher Education and Research Chennai Tamil Nadu India Department or Computer Science and Engineering School of Computing Bharath Institute of Higher Education and Research Chennai Tamil Nadu India

ISBN: (数字)9798331508845

ISBN: (纸本)9798331508852

Neural stem cells are a specific type of stem cells that are unique in their capacity to self-renew and differentiate into neurons and glial cells, including neurons, oligodendrocytes and astrocytes the latter contributing to the developmental process of the nervous system. In a bid to implement this approach in identifying imaging features related to neural stem cells, this study aims at using deep convolutional neural networks (CNNs) at a middle level. In the study, the primary focus is laid on the consideration of the distribution of the number of CD133+ neural stem cells as well as their possible migration in rat brain areas. Absolute quantification studies delineated marked qualitative differences in the localization of these cells across distinct regions of the brain. Most importantly, we found out that CD133+ neural stem cells could move from subventricular zone (SVZ) to almost near different parts of the brain including ganglia, corpus callosum and cerebral cortex. Before undergoing surgery, the activity of the rats was trained through water maze learning, respectively recording their escape latency (EL) at the time point of one week, two weeks, and one month. The findings clearly showed that EL was longer in the cerebral ischemia-reperfusion as compared with the sham-operated control group. Nevertheless, a significant decrease in EL was seen in the group treated with electroacupuncture after ischemia-reperfusion. From this it could be extrapolated that electroacupuncture may improve the viability and differentiation of newly generated neural stem cells into both glial and neuronal lineages. In extension, electroacupuncture promoted new neuron formation and a subset was found to be functional when they secreted acetylcholine. Taken together, the present study showed that electroacupuncture facilitated neurological recovery in cerebral ischemia-reperfusion rats.

关键词： deep learning Cerebral cortex Neurons Imaging Surgery Stem cells Rats Convolutional neural networks Ganglia Corpus callosum

来源：评论

学校读者我要写书评

暂无评论

WaveGNet: Wavelet-Guided deep learning for Efficient Aerial Power Line Detection

引用

IEEE SENSORS JOURNAL 2024年第14期24卷 22037-22044页

作者： An, Deyu Chao, Jianshu Li, Ting Fang, Li Fujian Agr & Forestry Univ Dept Elect & Mech Engn Fuzhou 350000 Peoples R China Univ Chinese Acad Sci Fujian Coll Fuzhou 350000 Peoples R China Quanzhou Inst Equipment Mfg Fujian Inst Res Struct Matter Chinese Acad Sci Quanzhou 362000 Peoples R China Fujian Normal Univ Dept Comp & Cyberspace Secur Fuzhou 350000 Peoples R China

Numerous approaches exist for unmanned aerial vehicles (UAVs) to perceive the surrounding environment. CMOS and charge coupled device (CCD) image sensors are commonly used in machine perception due to their cost-effectiveness and ease of acquisition. The use of image sensors for UAV power line (PL) inspection presents challenges due to the disorganized nature of the PL aerial images and the abundance of redundant background information. To address these challenges, we propose a novel approach that combines wavelet transform theory and deep learning methods. The image acquired by the image sensor is decomposed into low-frequency coefficients and high-frequency coefficients using wavelet transform, and the high-frequency coefficients are used to guide the low-frequency coefficients to generate a feature map of weak semantic information. This approach helps suppress complex and redundant background feature information, enhancing the detection of slender PLs. Furthermore, we incorporate an asymmetric dilated convolution global attention mechanism, guided by wavelet decomposition, to further enhance the features of slender PLs. This attention mechanism focuses on key regions and details, improving the PL detection process. Finally, we present a lightweight PL detection algorithm, WaveGNet, which combines deep learning and wavelet transform. Through extensive experiments on pinhole and fish-eye aerial PL datasets, WaveGNet achieves a remarkable trade-off of speed and accuracy, surpassing current lightweight segmentation models. This algorithm offers a novel approach for PL detection using image sensors and provides valuable insights for future deployment on UAVs.

关键词： Wavelet transforms Transforms Convolution deep learning Semantics image edge detection Discrete wavelet transforms Asymmetric dilated convolution power line (PL) detection real-time segmentation wavelet transform

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：