检索结果-内蒙古大学图书馆

XSleepFusion: A dual-stage information bottleneck fusion framework for interpretable multimodal sleep analysis

学校读者我要写书评

暂无评论

Information Fusion 2025年 123卷

作者： Hu, Shuaicong Wang, Yanan Liu, Jian Yang, Cuiwei Department of Biomedical Engineering Fudan University Shanghai200433 China Key Laboratory of Medical Imaging Computing and Computer Assisted Intervention of Shanghai Shanghai200093 China

Sleep disorders affect hundreds of millions globally, with accurate assessment of sleep apnea (SA) and sleep staging (SS) essential for clinical diagnosis and early intervention. Manual analysis by sleep experts is time-consuming and subject to inter-rater variability. Deep learning (DL) approaches offer automation potential but face fundamental challenges in multi-modal physiological signal integration and interpretability. This paper presents XSleepFusion, a cross-modal fusion framework based on information bottleneck (IB) theory for automated sleep analysis. The framework introduces a dual-stage IB mechanism that systematically processes physiological signals: first eliminating intra-modal redundancy, then optimizing cross-modal feature fusion. An evolutionary attention Transformer network (EAT-Net) backbone extracts temporal features at multiple scales, providing interpretable attention patterns. Experimental validation on eight clinical datasets comprising over 15,000 sleep recordings demonstrates the framework's effectiveness in polysomnogram (PSG)-based SA detection, electrocariogram (ECG)-based SA detection, and SS. The architecture achieves superior generalization across varying signal qualities and modal combinations, while the dual-stage design enables flexible integration of diverse physiological signals. Through interpretable feature representations and robust cross-modal fusion capabilities, XSleepFusion establishes a reliable and adaptable foundation for clinical sleep monitoring. © 2025 Elsevier B.V.

关键词： Electrocardiography

REMOTE: Real-time Ego-motion Tracking for Various Endoscopes via Multimodal Visual Feature Learning

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Shao, Liangjing Chen, Benshuang Zhao, Shuting Chen, Xinrong Academy for Engineering & Technology Fudan University China Shanghai Key Laboratory of Medical Image Computing and Computer-Assisted Intervention Fudan University China

Real-time ego-motion tracking for endoscope is a significant task for efficient navigation and robotic automation of endoscopy. In this paper, a novel framework is proposed to perform real-time ego-motion tracking for endoscope. Firstly, a multi-modal visual feature learning network is proposed to perform relative pose prediction, in which the motion feature from the optical flow, the scene features and the joint feature from two adjacent observations are all extracted for prediction. Due to more correlation information in the channel dimension of the concatenated image, a novel feature extractor is designed based on an attention mechanism to integrate multi-dimensional information from the concatenation of two continuous frames. To extract more complete feature representation from the fused features, a novel pose decoder is proposed to predict the pose transformation from the concatenated feature map at the end of the framework. At last, the absolute pose of endoscope is calculated based on relative poses. The experiment is conducted on three datasets of various endoscopic scenes and the results demonstrate that the proposed method outperforms state-of-the-art methods. Besides, the inference speed of the proposed method is over 30 frames per second, which meets the real-time requirement. The project page is here: *** © 2025, CC BY.

关键词： Endoscopy

Weakly Semi-supervised Whole Slide Image Classification by Two-level Cross Consistency Supervision

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Qu, Linhao Li, Shiman Luo, Xiaoyuan Liu, Shaolei Guo, Qinhao Wang, Manning Song, Zhijian Digital Medical Research Center School of Basic Medical Science Fudan University Shanghai China Shanghai Key Lab of Medical Image Computing and Computer Assisted Intervention Shanghai China Department of Gynecologic Oncology Shanghai Cancer Center Fudan University Shanghai China Department of Oncology Shanghai Medical College Fudan University Shanghai China

computer-aided Whole Slide Image (WSI) classification has the potential to enhance the accuracy and efficiency of clinical pathological diagnosis. It is commonly formulated as a Multiple Instance Learning (MIL) problem, where each WSI is treated as a bag and the small patches extracted from the WSI are considered instances within that bag. However, obtaining labels for a large number of bags is a costly and time-consuming process, particularly when utilizing existing WSIs for new classification tasks. This limitation renders most existing WSI classification methods ineffective. To address this issue, we propose a novel WSI classification problem setting, more aligned with clinical practice, termed Weakly Semi-supervised Whole slide image Classification (WSWC). In WSWC, a small number of bags are labeled, while a significant number of bags remain unlabeled. The MIL nature of the WSWC problem, coupled with the absence of patch labels, distinguishes it from typical semi-supervised image classification problems, making existing algorithms for natural images unsuitable for directly solving the WSWC problem. In this paper, we present a concise and efficient framework, named CroCo, to tackle the WSWC problem through two-level Cross Consistency supervision. CroCo comprises two heterogeneous classifier branches capable of performing both instance classification and bag classification. The fundamental idea is to establish cross-consistency supervision at both the bag-level and instance-level between the two branches during training. Extensive experiments conducted on four datasets demonstrate that CroCo achieves superior bag classification and instance classification performance compared to other comparative methods when limited WSIs with bag labels are available. To the best of our knowledge, this paper presents for the first time the WSWC problem and gives a successful resolution. Copyright © 2025, The Authors. All rights reserved.

关键词： Semi-supervised learning

DM-Mamba: Dual-domain Multi-scale Mamba for MRI Reconstruction

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Meng, Yucong Yang, Zhiwei Song, Zhijian Shi, Yonghong Fu, Kexue Digital Medical Research Center School of Basic Medical Science Fudan University Shanghai200032 China Shanghai Key Laboratory of Medical Image Computing and Computer Assisted Intervention Shanghai200032 China Academy of Engineering and Technology Fudan University Shanghai200433 China Jinan China

The accelerated MRI reconstruction poses a challenging ill-posed inverse problem due to the significant undersampling in k-space. Deep neural networks, such as CNNs and ViT, have shown substantial performance improvements for this task while encountering the dilemma between global receptive fields and efficient computation. To this end, this paper pioneers exploring Mamba, a new paradigm for long-range dependency modeling with linear complexity, for efficient and effective MRI reconstruction. However, directly applying Mamba to MRI reconstruction faces three significant issues: (1) Mamba’s row-wise and column-wise scanning disrupts k-space’s unique spectrum, leaving its potential in k-space learning unexplored. (2) Existing Mamba methods unfold feature maps with multiple lengthy scanning paths, leading to long-range forgetting and high computational burden. (3) Mamba struggles with spatially-varying contents, resulting in limited diversity of local representations. To address these, we propose a dual-domain multi-scale Mamba for MRI reconstruction from the following perspectives: (1) We pioneer vision Mamba in k-space learning. A circular scanning is customized for spectrum unfolding, benefiting the global modeling of k-space. (2) We propose a multi-scale Mamba with an efficient scanning strategy in both image and k-space domains. It mitigates long-range forgetting and achieves a better trade-off between efficiency and performance. (3) We develop a local diversity enhancement module to improve the spatially-varying representation of Mamba. Extensive experiments are conducted on three public datasets for MRI reconstruction under various undersampling patterns. Comprehensive results demonstrate that our method significantly outperforms state-of-the-art methods with lower computational cost. Implementation code will be available in https://***/XiaoMengLiLiLi/DM-Mamba. Copyright © 2025, The Authors. All rights reserved.

关键词： Deep neural networks

A Contrast-Enhanced Null Subtraction imaging Method using Dynamic DC Bias in Ultrafast Ultrasound imaging

学校读者我要写书评

暂无评论

IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control 2025年第6期PP卷 PP页

作者： Yan, Xin Yang, Xiaodong Jing, Lingling Guo, Wei Wang, Yingqi Su, Xinwei Wang, Yuanyuan Shenyang University of Technology School of Information Science and Engineering Shenyang110870 China Ultrasound Department of Anshan Central Hospital Anshan114000 China CO. Ltd Suzhou215123 China Fudan University Department of Electronic Engineering and the Key Laboratory of Medical Imaging Computing and Computer Assisted Intervention of Shanghai Shanghai200433 China

Enhancing the resolution and contrast of ultrafast ultrasound imaging is imperative for the accuracy of clinical diagnostics. Null subtraction imaging (NSI) is a nonlinear beamforming technique capable of significantly enhancing lateral resolution. However, it suffers from issues of low-quality speckle pattern and poor contrast performance. To address this issue, we propose a novel contrast-enhanced NSI method that utilizes dynamic DC bias. Innovatively, we construct the dynamic DC bias using a generalized coherence factor (GCF) and a sigmoid transformation function that adapts the DC value based on the signal characteristics of different imaging regions. Furthermore, a normalization scheme is proposed to optimize the beamforming output, ensuring uniform pixel intensity throughout the final image. Simulation, phantom, and in vivo data are utilized for ultrasound beamforming to evaluate the performance of the proposed method. Quantitative results show that the proposed method significantly enhances the contrast ratio (CR) by 197%, the contrast-to-noise ratio (CNR) by 341%, the speckle signal-to-noise ratio (sSNR) by 302%, and the generalized contrast-to-noise ratio (gCNR) by 106% compared to the original NSI (in phantom). Point target imaging results indicate that the proposed method achieves a main lobe width slightly wider than the original NSI method, but much narrower than those of DAS and GCF. These findings confirm that the proposed method significantly enhances imaging contrast while preserving high resolution, which is of great significance for the further clinical application of ultrafast ultrasound imaging. © 1986-2012 IEEE.

关键词： Phantoms

Exploring CLIP’s Dense Knowledge for Weakly Supervised Semantic Segmentation

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Yang, Zhiwei Meng, Yucong Fu, Kexue Tang, Feilong Wang, Shuo Song, Zhijian Academy for Engineering and Technology Fudan University Shanghai200433 China Shanghai Key Laboratory of Medical Image Computing and Computer Assisted Intervention China Digital Medical Research Center School of Basic Medical Sciences Fudan University China Shandong Computer Science Center National Supercomputer Center in Jinan China

Weakly Supervised Semantic Segmentation (WSSS) with image-level labels aims to achieve pixel-level predictions using Class Activation Maps (CAMs). Recently, Contrastive Language-Image Pre-training (CLIP) has been introduced in WSSS. However, recent methods primarily focus on image-text alignment for CAM generation, while CLIP’s potential in patch-text alignment remains unexplored. In this work, we propose ExCEL to explore CLIP’s dense knowledge via a novel patch-text alignment paradigm for WSSS. Specifically, we propose Text Semantic Enrichment (TSE) and Visual Calibration (VC) modules to improve the dense alignment across both text and vision modalities. To make text embeddings semantically informative, our TSE module applies Large Language Models (LLMs) to build a dataset-wide knowledge base and enriches the text representations with an implicit attribute-hunting process. To mine fine-grained knowledge from visual features, our VC module first proposes Static Visual Calibration (SVC) to propagate fine-grained knowledge in a non-parametric manner. Then Learnable Visual Calibration (LVC) is further proposed to dynamically shift the frozen features towards distributions with diverse semantics. With these enhancements, ExCEL not only retains CLIP’s training-free advantages but also significantly outperforms other state-of-the-art methods with much less training cost on PASCAL VOC and MS COCO. Code is available at https://***/zwyang6/ExCEL. Copyright © 2025, The Authors. All rights reserved.

关键词： Semantic Segmentation

Anatomy-inspired model for critical landmark localization in 3D spinal ultrasound volume data

学校读者我要写书评

暂无评论

medical Image Analysis 2025年 103卷

作者： Huang, Yi Jiao, Jing Yu, Jinhua Zheng, Yongping Wang, Yuanyuan Biomedical Engineering Center Fudan University Shanghai 200433 China Key Laboratory of Medical Imaging Computing and Computer Assisted Intervention of Shanghai Fudan University 200433 China Department of Biomedical Engineering The Hong Kong Polytechnic University SAR Hong Kong China Research Institute for Smart Ageing The Hong Kong Polytechnic University SAR Hong Kong China

Three-dimensional (3D) spinal ultrasound imaging has demonstrated its promising potential in measuring spinal deformity through recent studies, and it is more suitable for massive early screening and longitudinal follow-up of adolescent idiopathic scoliosis (AIS) compared with X-ray imaging due to its radiation-free superiority. Moreover, some deformities with low observability, such as vertebral rotation, in X-ray images can also be reflected by critical landmarks in 3D ultrasound data. In this paper, we propose a localization network (LLNet) to extract lamina in 3D ultrasound data, which has been indicated as a meaningful anatomy for measuring vertebral rotation by clinical studies. First, the LLNet skillfully establishes a parallel anatomical prior embedding branch that implicitly explores the anatomical correlation between the lamina and another anatomy with more stable observability (spinous process) during the training phase and then introduces the correlation to highlight the potential region of the lamina in the inferring one. Second, since the lamina is a tiny target, the information loss caused by continuous convolutional and pooling operations has a profound negative effect on detecting the lamina. We employ an optimization mechanism to mitigate this problem, which refines feature maps according to information from the original image and further reuses them to polish output. Furthermore, a modified global-local attention module is deployed on skip connections to mine global dependencies and contextual information to construct an effective image pattern. Extensive comparisons and ablation studies are performed on actual clinical data. Results indicate that the capability of our model is better than other outstanding detection models, and functional modules effectively contribute to this, with a 100.0 % detection success rate and an 8.9 % improvement of mean intersection over the union. Thus, our model is promising to become a helpful part of a computer-ass

关键词： Ultrasonic imaging

PLTN: Noisy label learning in long-tailed medical images with adaptive prototypes

学校读者我要写书评

暂无评论

Neurocomputing 2025年 645卷

作者： Zhiqing He Jie Xu Peng Huang Qin Wang Yuanyuan Wang Yi Guo Department of Biomedical Engineering School of Information Science and Technology Fudan University Shanghai China Key Laboratory of Medical Imaging Computing and Computer Assisted Intervention of Shanghai Shanghai China

label noise poses a significant challenge in medical image classification, particularly for minority diseases in long-tailed distribution. Most Noisy label Learning methods rely on clean sample selection to learn representative patterns. However, the imbalanced data distribution and intra-class variability cause model bias towards head classes at the inter-class level and dominant patterns at the intra-class level. Existing methods show limited performance in handling these biases simultaneously, causing inefficient clean sample selection and hindering minority class training. Thus, we propose PLTN, named P rototype-guided L ong- T ail N oisy Learning, a novel approach that selects comprehensive clean samples from each class and enhances minority class training to improve diagnostic accuracy of rare disease in long tailed medical data. First, Adaptive Multi-Prototype Learning (AMPL) module is designed to enable precise selection of clean samples, especially for minority classes. It achieves this by capturing diverse feature patterns through a dynamic prototype learning approach. The clean samples are then treated as labeled data while the noisy samples are used as unlabeled data in a semi-supervised learning paradigm. Second, we introduce Sharpness-aware Distributional Mixup (SDM) to enhance minority class training. SDM employs distributional mixup to enrich tail classes representations, simultaneously minimizing loss landscape sharpness to enhance model generalization for all classes, benefiting tail classes. Experiments on three typical medical datasets demonstrate that our method outperforms state-of-the-art methods. On HAM10000 dataset with 20 % label noise, we achieve an 80.34 % Macro-F1 score, with notable gains in tail class performance while maintaining head class performance, showing potential for real-world clinical applications.

关键词： Semi-supervised learning

DDFP: Data-dependent frequency prompt for source free domain adaptation of medical image segmentation

学校读者我要写书评

暂无评论

Knowledge-Based Systems 2025年 324卷

作者： Siqi Yin Shaolei Liu Manning Wang Digital Medical Research Center School of Basic Medical Science Fudan University Shanghai 200032 China Shanghai Key Laboratory of Medical Image Computing and Computer Assisted Intervention Shanghai 200032 China

Domain adaptation addresses the challenge of model performance degradation caused by domain gaps. In the typical setup for unsupervised domain adaptation, labeled data from a source domain and unlabeled data from a target domain are used to train a target model. However, access to labeled source domain data, particularly in medical datasets, can be restricted due to privacy policies. As a result, research has increasingly shifted to source-free domain adaptation (SFDA), which requires only a pretrained model from the source domain and unlabeled data from the target domain data for adaptation. Existing SFDA methods often rely on domain-specific image style translation and self-supervision techniques to bridge the domain gap and train the target domain model. However, the quality of domain-specific style-translated images and pseudo-labels produced by these methods still leaves room for improvement. Moreover, training the entire model during adaptation can be inefficient under limited supervision. In this paper, we propose a novel SFDA framework to address these challenges. Specifically, to effectively mitigate the impact of domain gap in the initial training phase, we introduce preadaptation to generate a preadapted model, which serves as an initialization of target model and allows for the generation of high-quality enhanced pseudo-labels without introducing extra parameters. Additionally, we propose a data-dependent frequency prompt to more effectively translate target domain images into a source-like style. To further enhance adaptation, we employ a style-related layer fine-tuning strategy, specifically designed for SFDA, to train the target model using the prompted target domain images and pseudo-labels. Extensive experiments on cross-modality abdominal and cardiac SFDA segmentation tasks demonstrate that our proposed method outperforms existing state-of-the-art methods. Our code is available online.

关键词：