检索结果-内蒙古大学图书馆

Aortic KD former: aortic multiclass segmentation using SegFormer via knowledge distillation

Machine Learning: Science and technology 2025年第2期6卷

作者： Mohamed Soliman, Nancy Awadalla, Medhat Elhelw, Mohamed Elattar, Mustafa Medical Imaging and Image Processing Research Group Center for Informatics Science Nile University Giza Egypt Department of Computers and Systems Helwan University Helwan Cairo Egypt School of Information Technology and Computer Science Nile University Giza Egypt

In this paper, we investigate the effectiveness of knowledge distillation for semantic segmentation of aortic structures using deep learning models. We employ SegFormer B5 as the teacher model and SegFormer B0 as the student model. Knowledge distillation is applied through three methods: output-based distillation, inner feature maps-based distillation, and integrated distillation, which combines both logits and feature maps. Our results show that while the student model (SegFormer B0) has significantly fewer parameters (3,715,686) compared to the teacher model (SegFormer B5) with 84,597,958 parameters, it achieves performance closely approaching that of the teacher. Among the evaluated techniques, inner feature map distillation yields the best performance, with a mean IoU of 0.9994, a mean Dice score of 0.9549, and a loss of 0.0020, demonstrating its superior capability in transferring knowledge effectively. The study highlights the potential of inner feature map distillation as the most effective method for achieving high performance with reduced model complexity. © 2025 The Author(s). Published by IOP Publishing Ltd.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Elucidating Deep Transfer Learning Approach for Early Plant Disease Detection Through Spot and Lesion Analysis

Elucidating Deep Transfer Learning Approach for Early Plant ...

引用

Disruptive Technologies (ICDT), International Conference on

作者： Abha Kaushik Shree Harsh Attri Sansar Singh Chauhan Department of Computer Science and Engineering Sharda University Greater Noida Uttar Pradesh INDIA Department of Computer Science and Engineering Centre for AI in Medicine Imaging and Forensic (CAIMIF) Sharda University Greater Noida Uttar Pradesh INDIA Computer Science and Engineering G.L. Bajaj Institute of Technology and Management Greater Noida INDIA

ISBN: (数字)9798331519582

ISBN: (纸本)9798331519599

The research field of detecting plant diseases through spot and lesion identification has become a crucial field because it helps minimize agricultural losses and enhances food security. The survey analyzes the latest progress in deep transfer learning methods which assist in detecting plant diseases during their early stages. A review of numerous research papers demonstrates how transfer learning techniques are used to achieve early detection of disease symptoms. The discussion includes significant development such as the application of CNN models ResNet, MobileNet, and EfficientNet alongside techniques like data augmentation and fine-tuning to overcome issues like limited labeled data and computational restrictions. The analysis focuses on commonly utilized data sets in plant disease detection research, which include the widely utilized PlantVillage dataset along with datasets from real-time field datasets. The paper highlights the dataset capabilities in managing data diversity, environmental variations, and scalability, but also represents dataset limitations when applied to real-world situations. A detailed overview discussed existing methodology in early plant disease detection through transfer learning and identifying major research gaps. It identifies critical obstacles, like unavailability of real-time based datasets, dataset imbalances, recommends future research based on development of lightweight models for devices with limited resources utilizing federated learning for the diagnostic process. This research presents essential insights to researchers and field experts to boost precision agriculture using novel AI methods.

关键词： Surveys Precision agriculture Plant diseases Reviews Scalability Transfer learning Food security Real-time systems Lesions Object recognition

来源：评论

学校读者我要写书评

暂无评论

AugGPT: Leveraging ChatGPT for Text Data Augmentation

引用

IEEE Transactions on Big Data 2025年第03期11卷 907-918页

作者： Dai, Haixing Liu, Zhengliang Liao, Wenxiong Huang, Xiaoke Cao, Yihan Wu, Zihao Zhao, Lin Xu, Shaochen Zeng, Fang Liu, Wei Liu, Ninghao Li, Sheng Zhu, Dajiang Cai, Hongmin Sun, Lichao Li, Quanzheng Shen, Dinggang Liu, Tianming Li, Xiang University of Georgia School of Computing AthensGA United States South China University of Technology School of Computer Science and Engineering China Lehigh University Department of Computer Science and Engineering BethlehemPA United States Carnegie Mellon University Heinz College of Information Systems and Public Policy PittsburghPA United States Massachusetts General Hospital Harvard Medical School Department of Radiology BostonMA United States Mayo Clinic Department of Radiation Oncology PhoenixAZ United States University of Virginia School of Data Science CharlottesvilleVA United States The University of Texas at Arlington Department of Computer Science and Engineering ArlingtonTX United States ShanghaiTech University School of Biomedical Engineering Shanghai201210 China Shanghai United Imaging Intelligence Co. Ltd. Shanghai200230 China Shanghai Clinical Research and Trial Center Shanghai201210 China

Text data augmentation is an effective strategy for overcoming the challenge of limited sample sizes in many natural language processing (NLP) tasks. This challenge is especially prominent in the few-shot learning (FSL) scenario, where the data in the target domain is generally much scarcer and of lowered quality. A natural and widely used strategy to mitigate such challenges is to perform data augmentation to better capture data invariance and increase the sample size. However, current text data augmentation methods either can't ensure the correct labeling of the generated data (lacking faithfulness), or can't ensure sufficient diversity in the generated data (lacking compactness), or both. Inspired by the recent success of large language models (LLM), especially the development of ChatGPT, we propose a text data augmentation approach based on ChatGPT (named"AugGPT"). AugGPT rephrases each sentence in the training samples into multiple conceptually similar but semantically different samples. The augmented samples can then be used in downstream model training. Experiment results on multiple few-shot learning text classification tasks show the superior performance of the proposed AugGPT approach over state-of-the-art text data augmentation methods in terms of testing accuracy and distribution of the augmented samples. Codes of AugGPT are available at https://***/yhydhx/AugGPT. © 2015 IEEE.

关键词： Zero-shot learning

来源：评论

学校读者我要写书评

暂无评论

Deep learning-based multi-task prediction of response to neoadjuvant chemotherapy using multiscale whole slide images in breast cancer:A multicenter study

引用

Chinese Journal of Cancer Research 2025年第1期37卷 28-47页

作者： Qin Wang Feng Zhao Haicheng Zhang Tongpeng Chu Qi Wang Xipeng Pan Yuqian Chen Heng Zhou Tiantian Zheng Ziyin Li Fan Lin Haizhu Xie Heng Ma Lan Liu Lina Zhang Qin Li Weiwei Wang Yi Dai Ruijun Tang Jigang Wang Ping Yang Ning Mao School of Information and Electronic Engineering Shandong Technology and Business UniversityYantai 264005China Big Data and Artificial Intelligence Laboratory Yantai Yuhuangding Hospital of Qingdao UniversityYantai 264000China Shandong Provincial Key Medical and Health Laboratory of Intelligent Diagnosis and Treatment for Women’s Diseases Yantai Yuhuangding HospitalYantai 264000China Department of Radiology Yantai Yuhuangding Hospital of Qingdao UniversityYantai 264000China School of Computer Science and Technology Shandong Technology and Business UniversityYantai 264005China School of Computer Science and Information Security Guilin University of Electronic TechnologyGuilin 541004China School of Medical Imaging Binzhou Medical UniversityYantai 264003China Department of Radiology Jiangxi Cancer Hospitalthe Second Affiliated Hospital of Nanchang Medical CollegeNanchang 330006China Department of Radiology the First Affiliated Hospital of China Medical UniversityShenyang 400042China Department of Radiology Weifang Hospital of Traditional Chinese MedicineWeifang 262600China Department of Medical Imaging Affiliated Hospital of Jining Medical UniversityJining 272029China Department of Radiology the Peking University Shenzhen HospitalShenzhen 518036China Department of Pathology Guilin Traditional Chinese Medicine HospitalGuilin 541002China Department of Pathology the Affiliated Hospital of Qingdao UniversityQingdao 266555China Department of Pathology Yantai Yuhuangding Hospital of Qingdao UniversityYantai 264000China

Objective:Early predicting response before neoadjuvant chemotherapy(NAC)is crucial for personalized treatment plans for locally advanced breast cancer *** aim to develop a multi-task model using multiscale whole slide images(WSIs)features to predict the response to breast cancer NAC more ***:This work collected 1,670 whole slide images for training and validation sets,internal testing sets,external testing sets,and prospective testing sets of the weakly-supervised deep learning-based multi-task model(DLMM)in predicting treatment response and pCR to *** approach models two-by-two feature interactions across scales by employing concatenate fusion of single-scale feature representations,and controls the expressiveness of each representation via a gating-based attention ***:In the retrospective analysis,DLMM exhibited excellent predictive performance for the prediction of treatment response,with area under the receiver operating characteristic curves(AUCs)of 0.869[95%confidence interval(95%CI):0.806−0.933]in the internal testing set and 0.841(95%CI:0.814−0.867)in the external testing *** the pCR prediction task,DLMM reached AUCs of 0.865(95%CI:0.763−0.964)in the internal testing and 0.821(95%CI:0.763−0.878)in the pooled external testing *** the prospective testing study,DLMM also demonstrated favorable predictive performance,with AUCs of 0.829(95%CI:0.754−0.903)and 0.821(95%CI:0.692−0.949)in treatment response and pCR prediction,*** significantly outperformed the baseline models in all testing sets(P<0.05).Heatmaps were employed to interpret the decision-making basis of the ***,it was discovered that high DLMM scores were associated with immune-related pathways and cells in the microenvironment during biological basis ***:The DLMM represents a valuable tool that aids clinicians in selecting personalized treatment strategies for breast cancer patients.

关键词： Artificial intelligence breast cancer digital pathology whole slide images

来源：评论

学校读者我要写书评

暂无评论

Unlocking Alzheimer’s Disease Heterogeneity: The Role of Normative Modeling, Dynamic Connectivity, and Beyond

引用

Biological Psychiatry 2025年第11期97卷 1016-1017页

作者： Qian Wang Shile Qi Rongtao Jiang Jing Sui State Key Laboratory of Cognitive Neuroscience and Learning Beijing Normal University Beijing China College of Computer Science and Technology Nanjing University of Aeronautics and Astronautics Nanjing China Department of Radiology and Biomedical Imaging Yale School of Medicine New Haven Connecticut

来源：评论

学校读者我要写书评

暂无评论

Multi-Center Study on Deep Learning-Assisted Detection and Classification of Fetal Central Nervous System Anomalies Using Ultrasound imaging

arXiv

引用

arXiv 2025年

作者： Qi, Yang Cai, Jiaxin Lu, Jing Xiong, Runqing Chen, Rongshang Zheng, Liping Ma, Duo School of Computer and Information Engineering Xiamen University of Technology China School of Mathematics and Statistics Xiamen University of Technology China Department of Obstetrics Xiamen University Affiliated First Hospital China Department of Ultrasound Imaging Xiamen Medical College Affiliated Second Hospital China

Prenatal ultrasound evaluates fetal growth and detects congenital abnormalities during pregnancy, but the examination of ultrasound images by radiologists requires expertise and sophisticated equipment, which would otherwise fail to improve the rate of identifying specific types of fetal central nervous system (CNS) abnormalities and result in unnecessary patient examinations. We construct a deep learning model to improve the overall accuracy of the diagnosis of fetal cranial anomalies to aid prenatal diagnosis. In our collected multi-center dataset of fetal craniocerebral anomalies covering four typical anomalies of the fetal central nervous system (CNS): anencephaly, encephalocele (including meningocele), holoprosencephaly, and rachischisis, patient-level prediction accuracy reaches 94.5%, with an AUROC value of 99.3%. In the subgroup analyzes, our model is applicable to the entire gestational period, with good identification of fetal anomaly types for any gestational period. Heatmaps superimposed on the ultrasound images not only provide a visual interpretation for the algorithm but also provide an intuitive visual aid to the physician by highlighting key areas that need to be reviewed, helping the physician to quickly identify and validate key areas. Finally, the retrospective reader study demonstrates that by combining the automatic prediction of the DL system with the professional judgment of the radiologist, the diagnostic accuracy and efficiency can be effectively improved and the misdiagnosis rate can be reduced, which has an important clinical application prospect. All code is available at https://***/xiaqi7/Fetal-ultrasound. Copyright © 2025, The Authors. All rights reserved.

关键词： Ultrasonic imaging

来源：评论

学校读者我要写书评

暂无评论

Learning Accurate Rigid Registration for Longitudinal Brain MRI from Synthetic Data

Learning Accurate Rigid Registration for Longitudinal Brain ...

引用

IEEE International Symposium on Biomedical imaging

作者： Jingru Fu Adrian V. Dalca Bruce Fischl Rodrigo Moreno Malte Hoffmann Division of Biomedical Imaging KTH Royal Institute of Technology Huddinge Sweden Athinoula A. Martinos Center for Biomedical Imaging Charlestown USA Department of Radiology Massachusetts General Hospital Boston USA Department of Radiology Harvard Medical School Boston USA Computer Science & Artificial Intelligence Laboratory MIT Cambridge USA

ISBN: (数字)9798331520526

ISBN: (纸本)9798331520533

Rigid registration aims to determine the translations and rotations necessary to align features in a pair of images. While recent machine learning methods have become state-of-the-art for linear and deformable registration across subjects, they have demonstrated limitations when applied to longitudinal (within-subject) registration, where achieving precise alignment is critical. Building on an existing framework for anatomy-aware, acquisition-agnostic affine registration, we propose a model optimized for longitudinal, rigid brain registration. By training the model with synthetic within-subject pairs augmented with rigid and subtle nonlinear transforms, the model estimates more accurate rigid transforms than previous cross-subject networks and performs robustly on longitudinal registration pairs within and across magnetic resonance imaging (MRI) contrasts.

关键词： Training Neuroimaging Deep learning Image registration Accuracy Magnetic resonance imaging Buildings Transforms Brain modeling Synthetic data

来源：评论

学校读者我要写书评

暂无评论

Scale Feature-Aware Generative Adversarial Network Improve MRI Device Data Imbalance for Healthy Consumption

引用

IEEE Transactions on Consumer Electronics 2025年

作者： Zhao, Yawu Wang, Shudong Qiao, Sibo Zhang, Yulin Wang, Jiehuan Yin, Wenjing Liu, Tiyao Pang, Shanchen China University of Petroleum Qingdao Institute of Software School of Computer Science and Technology Shandong Qingdao China Tiangong University The School of Software Tianjin China Shandong University of Science and Technology College of Mathematics and System Science Shandong Qingdao China Affliated Hospital of Jining Medical University Department of Medical Imaging Shandong Jining China

In consumer healthcare, the imbalance between the imaging data of MRI devices creates a significant barrier to the generalization ability of medical image segmentation. We employ generative adversarial networks to address the challenges inherent in the heterogeneity of data distribution among different MRI devices. First, we propose the Scale Feature Awareness Module (SFAM), which can skillfully capture image details at different scales and broader contextual information. Then, we propose the Dynamic Scale Attention Module (DSAM), which aims to combine feature mappings at different scales for different paths dynamically. We can assign different dynamic weights to each scale to enhance the feature representation. Finally, we propose a multi-scale discriminator to guide the generation of aneurysm images with different diameters. Based on the above modules, we designed Scale Feature Aware Generative Adversarial Network (SFAGAN) to generate medical images with the same distribution. It is experimentally demonstrated that SFAGAN improves PSNR, SSIM, and FID values by 0.64, 0.62, and 7.99, respectively, over the SOTA method. In addition, we use the generated data for downstream segmentation tasks to demonstrate the quality of the generated images. Notably, our SFAGAN has significant performance, making its application in medical systems very feasible. © 1975-2011 IEEE.

关键词： Dynamic contrast enhanced MRI

来源：评论

学校读者我要写书评

暂无评论

Spatiotemporal Feature Extraction and Fusion for Longitudinal Alzheimer’s Disease Diagnosis 17th

Spatiotemporal Feature Extraction and Fusion for Longitudina...

引用

17th International Conference on Brain Informatics, BI 2024

作者： Guan, Zhenghua Yang, Peng Lei, Haijun Yang, Bao Song, Xuegang Dong, Lei Yan, Xueqin Wei, Cuimei Liang, Chunhua Xiao, Xiaohua Wang, Tianfu Lei, Baiying School of Biomedical Engineering Guangdong Key Laboratory for Biomedical Measurements and Ultrasound Imaging National-Regional Key Technology Engineering Laboratory for Medical Ultrasound Shenzhen University Medical schoo1 Shenzhen518060 China Affiliated Hospital of Shenzhen University Health Science Center Shenzhen University Shenzhen Second People’s Hospital Shenzhen China School of Computer Science and Software Engineering Shenzhen University Shenzhen518060 China

ISBN: (纸本)9789819632961

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder characterized by a gradual decline in cognitive function. The brain structure of AD patients progressively atrophies over time. Longitudinal data can capture the changes in imaging over time, providing important information for early detection and diagnosis of the disease, which is not achievable with images from a single time point. We propose a novel longitudinal AD diagnosis model that leverages deformation field data obtained by registering images from adjacent time points. Each pixel in the deformation field represents the displacement vector of the corresponding pixel between two scans, reflecting information about brain tissue atrophy. By effectively extracting both temporal and spatial features from the longitudinal data, our model enhances early AD diagnosis. The input data for the model consists of processed sMRI images and deformation fields calculated between each pair of time points in the longitudinal sMRI data. The model is composed of three modules: Convolutional Long Short-Term Memory (CLSTM), Gate Fusion Long Short-Term Memory (GFLSTM), and an attention module. CLSTM extracts temporal features while preserving spatial features by maintaining the 3D shape of the features, GFLSTM achieves spatiotemporal feature fusion within a cascaded network, and the attention module further enhances feature representation. Our proposed method was evaluated on the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database, and experiments demonstrate that it outperforms several state-of-the-art methods. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Label-free hyperspectral multiphoton microscopy

引用

Optics Letters 2025年第5期50卷 1484-1487页

作者： De La Cadena, Alejandro Renteria, Carlos A. Aksamitiene, Edita Boppart, Stephen A. Beckman Institute for Advanced Science and Technology University of Illinois Urbana-Champaign UrbanaIL61801 United States Department of Bioengineering University of Illinois Urbana-Champaign UrbanaIL61801 United States Department of Electrical and Computer Engineering University of Illinois Urbana-Champaign UrbanaIL61801 United States Interdisciplinary Health Science Institute University of Illinois Urbana-Champaign UrbanaIL61801 United States University of Illinois Urbana-Champaign NIH/NIBIB P41 Center for Label-Free Imaging and Multiscale Biophotonics United States

Label-free multiphoton microscopy is a powerful tool for investigating pristine biological specimens. This imaging modality leverages optical signals originating from the nonlinear response of native biomolecules to intense optical radiation, nonlinear signals that allow localizing and quantifying the constituents of specimens, driving applications in biology and medicine. However, since its inception over three decades ago, this approach has operated with a narrowband detection scheme, relying on narrow bandwidths from the entire spectra to derive imaging contrast. This detection scheme hinders the analytical power of the nonlinear microscope, preventing the rigorous unmixing of co-localized constituents with spectral overlap. In this Letter, we shift from the narrowband paradigm to the broadband, demonstrating label-free hyperspectral multiphoton imaging of biological specimens. We validate this configuration by disentangling the constituents of fresh murine tissues by virtue of nonlinear spectra. © 2025 Optica Publishing Group. All rights, including for text and data mining (TDM), Artificial Intelligence (AI) training, and similar technologies, are reserved.

关键词： Nonlinear optics

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：