Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks332 from public data
- EfficientViT: Memory Efficient Vision Transformer with Cascaded Group Attention912
This paper proposes a family of high-speed vision transformers named Efficient ViT, and designs a new building block with a sandwich layout, i.e., using a single memory-bound MHSA between efficient FFN layers, which improves memory efficiency while enhancing channel communication.
- Selective Feature Aggregation Network with Area-Boundary Constraints for Polyp Segmentation445
This work proposes a novel selective feature aggregation network with the area and boundary constraints and introduces three up-concatenations between encoder and decoders and embedding Selective Kernel Modules into convolutional layers which can adaptively extract features from different size of kernels.
- CLIP-Driven Universal Model for Organ Segmentation and Tumor Detection391
The CLIP-Driven Universal Model is proposed, which incorporates text embedding learned from Contrastive Language-Image Pre-training (CLIP) to segmentation models and is computationally more efficient compared with dataset-specific models, generalized better to CT scans from varying sites, and shows stronger transfer learning performance on novel tasks.
- 354
- 296
- 234
- Measurement of Branching Fractions for B→ηcK(*) Decays229
Measurements of branching fractions for charged and neutral B-->eta( c)K decays where the eta(c) meson is reconstructed in the K(0)(S)K+/-pi(-/+), K+K-pi(0), K(*0)K- pi(+), and pp; decay channels are reported.
- Integrated lithium niobate microwave photonic processing engine219
The image edge detector is leveraged to enable a photonic-assisted image segmentation model that could effectively outline the boundaries of melanoma lesion in medical diagnostic images, achieving orders of magnitude faster processing speed and lower power consumption than conventional electronic processors.
- 216
- Deep learning for polyp recognition in wireless capsule endoscopy images194
The comprehensive results have demonstrated that the proposed SSAEIM can provide descriptive characterization for WCE images and recognize polyps in a WCE video accurately and could be further utilized in the clinical trials to help physicians from the tedious image reading work.
- 188
- DICCCOL: Dense Individualized and Common Connectivity-Based Cortical Landmarks172
This work reports a dense and consistent map of 358 cortical landmarks, named Dense Individualized and Common Connectivity-based Cortical Landmarks (DICCCOLs), defined by group-wise consistent white-matter fiber connection patterns derived from diffusion tensor imaging data.
- 170
- Ground-Aware Monocular 3D Object Detection for Autonomous Driving161
This work identifies how the ground plane provides additional clues in depth reasoning in 3D detection in driving scenes and improves the processing of 3D anchors and introduces a novel neural network module to fully utilize such application-specific priors in the framework of deep learning.
- Source free domain adaptation for medical image segmentation with fourier style mining158
A novel source free domain adaptation framework with fourier style mining, where only a well-trained source segmentation model is available for the adaptation to the target domain, and delivers impressive performance compared with state-of-the-art domain adaptation methods.
- 140
- 138
- Causal Disentanglement Domain Generalization for time-series signal fault diagnosis136
A DGFD method named Causal Disentanglement Domain Generalization (CDDG), which can reestablish the data-generating process by disentangling time-series data into the causal factors (fault-related representation) and no-casual factors (domain-related representation) with a structural causal model.
- 136
- Prostate cancer classification with multiparametric MRI transfer learning model133
The experiment results showed that the proposed MPTL model can learn discriminative features in prostate images and classify the cancer accurately, and could be further applied in the clinical practice to provide valuable information for cancer treatment and precision medicine.
- Development of deep neural network for individualized hepatobiliary toxicity prediction after liver SBRT125
The novel paradigm for toxicity prediction by leveraging the enormous potential of deep learning and go beyond the existing dose/volume histogram-based metric is proposed, which offers clinically accurate tools for hepatobiliary toxicity prediction and automatic identification of anatomical regions that are critical to spare during SBRT.
- 122
- 113
- A Heterogeneous Group CNN for Image Super-Resolution107
A heterogeneous group SR CNN (HGSRCNN) via leveraging structure information of different types to obtain a high-quality image and a parallel upsampling mechanism is developed to train a blind SR model.
- 106
- Superpixel-based and boundary-sensitive convolutional neural network for automated liver segmentation106
The proposed SBBS-CNN provides an accurate and effective tool for automated liver segmentation and showed superior performance in comparison with state-of-art methods, including U-Net, pixel-based CNN, active contour, level-sets and graph-cut algorithms.
- 98
- 97
- SCAN: Cross Domain Object Detection with Semantic Conditioned Adaptation97
This work proposes a novel Semantic Conditioned AdaptatioN (SCAN) framework such that well-modeled unbiased semantics can support semantic conditioned adaptation for precise domain adaptive object detection.
- 96
- 95
- 94
- Study of e+e−→γωJ/ψ and Observation of X(3872)→ωJ/ψ92
The X(3872) resonance is observed for the first time in the ωJ/ψ system with a significance of more than 5σ and the resonant parameters of the X(3915) agree with those of the Y(3940) in B→KωJ/ ψ and of theX(39 15) in γγ→ωJ /ψ observed by the Belle and BABAR experiments within errors.
- 90
- DRMF: Degradation-Robust Multi-Modal Image Fusion via Composable Diffusion Prior88
Dgradation-Robust Multi-modality image Fusion is presented, leveraging the powerful generative properties of diffusion models to counteract various degradations during image fusion, enabling effective multi-modal image fusion.
- GLENet: Boosting 3D Object Detectors with Generative Label Uncertainty Estimation87
This paper proposes GLENet, a generative framework adapted from conditional variational autoencoders, to model the one-to-many relationship between a typical 3D object and its potential ground-truth bounding boxes with latent variables and proposes an uncertainty-aware quality estimator architecture in probabilistic detectors to guide the training of the IoU-branch with predicted localization uncertainty.
- MetaCorrection: Domain-aware Meta Loss Correction for Unsupervised Domain Adaptation in Semantic Segmentation87
A MetaCorrection framework is advanced, where a Domain-aware Meta-learning strategy is devised to benefit Loss Correction (DMLC) for UDA semantic segmentation and the proposed DMLC strategy is employed to provide matched and compatible supervision signals for different level features, thereby ensuring deep adaptation.
- 81
- Semi-supervised WCE image classification with adaptive aggregated attention78
A synergic network to learn discriminative image features, consisting of two branches: an abnormal regions estimator and an abnormal information distiller, that achieves 93.17% overall accuracy in a fourfold cross-validation for WCE image classification.
- Hypermethylation of BRCA1 gene: implication for prognostic biomarker and therapeutic target in sporadic primary triple-negative breast cancer77
It is demonstrated that epigenetic alteration of key tumor suppressor gene can be a promising biomarker for the prognosis of triple-negative breast cancer/basal-like breast cancer.
- 71
- 69
- Artificial General Intelligence for Medical Imaging Analysis67
A thorough overview of the key features and enabling techniques of LLMs and AGI are provided, and the roadmaps guiding the evolution and implementation of AGI models in the medical sector are examined, summarizing their present applications, potentialities, and associated challenges.
- 67
- 66
- Universal and extensible language-vision models for organ segmentation and tumor detection from abdominal computed tomography66
A novel language-driven parameter generator that leverages language embeddings from large language models, enriching semantic encoding compared with one-hot encoding, and facilitates the extensibility to new classes while alleviating the catastrophic forgetting of previously learned classes is introduced.
- Multi‐subtype classification model for non‐small cell lung cancer based on radiomics: SLS model66
The SLS model can accurately classify and diagnose the four subtypes of NSCLC based on CT images, and thus it has the potential to be used in the clinical practice to provide valuable information for lung cancer treatment and further promote the personalized medicine.
- 61
- A Comprehensive Study of the Robustness for LiDAR-Based 3D Object Detectors Against Adversarial Attacks60
This paper presents the first comprehensive evaluation and analysis of the robustness of LiDAR-based 3D detectors under adversarial attacks, benchmarking the robustness of state-of-the-art LiDAR-based 3D object detectors against attacks on the KITTI and Waymo datasets.
- 58
- 57
- Contour-aware multi-label chest X-ray organ segmentation56
The proposed contour-aware segmentation improved the segmentation performance, and when compared against existing algorithms on the same publicly available database of 247 chest X-ray images, the UNet architecture with the ResNeXt50 encoder combined with the contours-aware approach resulted in the best overall segmentsation performance.
- 55
- 54
- 54
- Inferring Group-Wise Consistent Multimodal Brain Networks via Multi-View Spectral Clustering54
A novel approach to inferring group-wise consistent brain subnetworks from multimodal DTI/resting-state fMRI datasets via multi-view spectral clustering of cortical networks, which were constructed upon the recently developed and validated large-scale cortical landmarks - DICCCOL.
- Consolidated domain adaptive detection and localization framework for cross-device colonoscopic images53
This paper proposes a consolidated domain adaptive detection and localization framework to bridge the domain gap between different colonosopic datasets effectively, consisting of two parts: the pixel-level adaptation and the hierarchical feature- level adaptation.
- 53
- 52
- Fine-Grained Image-Text Alignment in Medical Imaging Enables Explainable Cyclic Image-Report Generation51
An AdaMatch-based bidirectional large language model for Cyclic CXR-report generation (AdaMatch-Cyclic) employs the AdaMatch to obtain the keywords for CXR images and `keypatches' for medical reports as hints to guide CXR-report generation.
- 50
- 50
- Personalized Retrogress-Resilient Framework for Real-World Medical Federated Learning50
This work devise a Progressive Fourier Aggregation at the server to achieve more stable and effective global knowledge gathering by integrating client models from low-frequency to high-frequency gradually and proposes a personalized retrogress-resilient framework to produce a superior personalized model for each client.
- 49
- Categorical Relation-Preserving Contrastive Knowledge Distillation for Medical Image Classification49
A novel Categorical Relation-preserving Contrastive Knowledge Distillation (CRCKD) algorithm, which takes the commonly used mean-teacher model as the supervisor, is proposed, which can distill the relational knowledge more comprehensively.
- Triple ANet: Adaptive Abnormal-aware Attention Network for WCE Image Classification48
A novel Angular Contrastive loss (AC Loss) is proposed to reduce the intra-class variances and enlarge the inter-class differences effectively effectively and showed better performance compared with state-of-the-art WCE image classification methods.
- Flaws can be Applause: Unleashing Potential of Segmenting Ambiguous Objects in SAM47
This paper introduces an optimization framework based on a conditional variational autoencoder, which jointly models the prompt and the granularity of the object with a latent probability distribution, and enables the model to adaptively perceive and represent the real ambiguous label distribution.
- 47
- Towards Robust Adaptive Object Detection under Noisy Annotations47
This paper represents the first effort to formulate noisy DAOD and proposes a Noise Latent Transferability Exploration (NLTE) framework to address this issue and a thorough evaluation on benchmark DAOD datasets with noisy source annotations validates the effectiveness of NLTE.
- 47
- 47
- 46
- P 2 SAM: Probabilistically Prompted SAMs Are Efficient Segmentator for Ambiguous Medical Images45
P²SAM is a novel framework that leverages the Segment Anything Model (SAM)'s prior knowledge for ambiguous object segmentation by transforming SAM's sensitivity to prompts into an advantage, and introduces a prior probabilistic space for prompts.
- 45
- UN-SAM: Universal Prompt-Free Segmentation for Generalized Nuclei Images43
This work proposes the Universal prompt-free SAM framework for Nuclei segmentation (UN-SAM), by providing a fully automated solution with remarkable generalization capabilities, and devise a multi-scale Self-Prompt Generation (SPGen) module to revolutionize clinical workflow.
- 41
- 41
- 41
- Diagnose Like A Pathologist: Weakly-Supervised Pathologist-Tree Network for Slide-Level Immunohistochemical Scoring39
A Pathologist-Tree Network (PTree-Net) is proposed to sparsely model the whole-slide image (WSI) of immunohistochemistry test of biopsy tissue efficiently in multi-scale manner and outperforms state-of-the-art by a large margin.
- 39
- 39
- Visual analytics of brain networks39
This paper presents a novel approach of visual analytics and its open-source software for ROI definition and brain network construction and demonstrates the effectiveness of this approach to an operation span working memory fMRI/DTI dataset, a schizophrenia DTI/resting state fMRI (R-fMRI) dataset, and a mild cognitive impairmentDTI/R- fMRI dataset.
- SimT: Handling Open-set Noise for Domain Adaptive Semantic Segmentation38
A simplex noise transition matrix (SimT) is proposed to model the mixed noise distributions in DA semantic segmentation and formulate the problem as estimation of SimT to correct noise issues in pseudo labels and promote the generalization ability of segmentation model on target domain data.
- 38
- 37
- 37
- Holistic Evaluation of GPT-4V for Biomedical Imaging36
A large-scale evaluation probing GPT-4V's capabilities and limitations for biomedical image analysis advances understanding of multimodal large language models (LLMs) and guides future work toward impactful healthcare applications.
- Biomedical image analysis competitions: The state of current participation practice36
An international survey was designed that was issued to all participants of challenges conducted in conjunction with the IEEE ISBI 2021 and MICCAI 2021 conferences, covering participants' expertise and working environments, their chosen strategies, as well as algorithm characteristics.
- 35
- 34
- 34
- 34
- MonoSplat: Generalizable 3D Gaussian Splatting from Monocular Depth Foundation Models33
MonoSplat is introduced, a novel framework that leverages rich visual priors from pre-trained monocular depth foundation models for robust Gaussian reconstruction and achieves superior reconstruction quality and generalization capability compared to existing methods while maintaining computational efficiency with minimal trainable parameters.
- 33
- Deep learning for identification of critical regions associated with toxicities after liver stereotactic body radiation therapy33
This work proposed and validated a universal deep learning-based solution for the identification of radiosensitive anatomical regions and recognized the importance of hepatobiliary tract sparing during liver SBRT without any prior anatomical knowledge.
- Non-equivalent images and pixels: Confidence-aware resampling with meta-learning mixup for polyp segmentation32
A novel Meta-Learning Mixup (MLMix) data augmentation method and a Confidence-Aware Resampling (CAR) strategy for polyp segmentation are proposed to facilitate the representation ability of model and ensure the stability of the training procedure.
- 31
- Wearable technologies for assisted mobility in the real world31
How wearable technologies are transforming assisted mobility towards more adaptive, intuitive, and user-centered solutions is discussed, highlighting human-in-the-loop and cognitive–sensorimotor integration.
- 31
- Meta-Analysis of Functional Roles of DICCCOLs31
This work aims to take the advantage of existing literature fMRI studies reported and aggregated in the BrainMap database to examine the possible functional roles of 358 DICCCOLs via meta-analysis and offers a starting point for future elucidation of fine-grainedfunctional roles of cortical landmarks.
- 30
- 30
- 30
- RIIS-DenseNet: Rotation-Invariant and Image Similarity Constrained Densely Connected Convolutional Network for Polyp Detection30
A rotation invariant and image similarity constrained Densely Connected Convolutional Network (RIIS-DenseNet) model is proposed to provide end-to-end polyp detection workflow and has superior performance compared with state-of-the-art methods.
- UN-SAM: Domain-adaptive self-prompt segmentation for universal nuclei images29
The domain-adaptive self-prompt SAM framework for Universal Nuclei segmentation (UN-SAM) is proposed, by providing a fully automated solution with superior performance across different domains by eliminating the labor-intensive requirement of per-nuclei annotations for prompt.
- 29
- Dual domain distribution disruption with semantics preservation: Unsupervised domain adaptation for medical image segmentation29
This work proposes a novel UDA framework known as Dual Domain Distribution Disruption with Semantics Preservation (DDSP), which makes the model domain-agnostic and focus on anatomical structural information by leveraging semantic information as constraints to guide the model to adapt to images with disrupted distributions in both source and target domains.
- Medical federated learning with joint graph purification for noisy label learning29
This work proposes a federated framework with joint Graph Purification (FedGP) to address the label noise in FL through server and clients collaboration and proposes a global centroid aggregation on the server side to produce a robust classifier with global knowledge, which can be optimized collaboratively in the FL framework.
- 29
- Instance importance-Aware graph convolutional network for 3D medical diagnosis29
This work proposes the Instance Importance-aware Graph Convolutional Network (I2GCN) under the multi-instance learning (MIL), which outperforms state-of-the-art methods by a large margin and devise the importance-based Sub-Graph Augmentation (SGA) to effectively regularize the framework training.
- IDEA-Net: Dynamic 3D Point Cloud Interpolation via Deep Embedding Alignment29
This paper proposes IDEA-Net, an end-to-end deep learning framework, which disentangles the problem of temporally interpolating dynamic 3D point clouds with large non-rigid deformation under the assistance of the explicitly learned temporal consistency.
- Advancing Embodied Intelligence in Robotic-Assisted Endovascular Procedures: A Systematic Review of AI Solutions28
This review systematically analyzes the integration of EI into endovascular robotics, identifying challenges such as the heterogeneity in validation standards and the gap between human mimicry and machine-native capabilities.
- 27
- Incremental Cross-view Mutual Distillation for Self-supervised Medical CT Synthesis27
A novel medical slice synthesis to increase the inter-slice resolution by introducing the incremental cross-view mutual distillation strategy to accomplish this task in the self-supervised learning manner and demonstrating the effectiveness of the proposed approach.
- Artificial Intelligence in Industrial Systems27
Although artificial intelligence has shown great success in solving real-world problems, there is still room for improvement due to the complexity of industrial tasks.
- 27
- Towards a general-purpose foundation model for fMRI analysis25
NeuroSTORM (Neuroimaging Foundation Model with Spatial-Temporal Optimized Representation Modeling) is introduced that learns generalizable representations directly from 4D fMRI volumes and enables efficient transfer to diverse downstream applications and is suggested to become a standardized foundation model for reproducible and transferable fMRI analysis.
- Bi-VLGM: Bi-Level Class-Severity-Aware Vision-Language Graph Matching for Text Guided Medical Image Segmentation24
A novel Bi-level class-severity-aware Vision-Language Graph Matching (Bi-VLGM) for text guided medical image segmentation to exploit the class-severity-aware relation among visual-textual features and its superiority to existing methods is introduced.
- 24
- 23
- 23
- 23
- 23
- A novel AI model for detecting periapical lesion on CBCT: CBCT-SAM22
CBCT-SAM is capable of providing expert-level assistance in the identification of periapical lesions on CBCT and slightly surpassed PAL-Net in the diagnostic and segmentation tasks.
- CLIFF: Continual Latent Diffusion for Open-Vocabulary Object Detection22
This work explores the advanced generative paradigm with distribution perception and proposes a novel framework based on the diffusion model, coined Continual Latent Diffusion (CLIFF), which formulates a continual distribution transfer among the object, image, and text latent space probabilistically.
- 22
- 22
- Dynamic-weighting hierarchical segmentation network for medical images22
Comprehensive experiments on three benchmark datasets show the newly proposed HieraSeg and DW-HieraSeg Networks achieve state-of-the-art performance, which clearly demonstrates the effectiveness of the proposed approaches for medical image segmentation.
- Domain Knowledge Based Brain Tumor Segmentation and Overall Survival Prediction22
A 3D dilated multi-fiber network (DMFNet) with weighted dice loss for brain tumor segmentation and a DenseNet based 3D neural network with position encoding convolutional layer (PECL) for OS prediction are proposed.
- CLOUD DETECTION METHOD BASED ON FEATURE EXTRACTION IN REMOTE SENSING IMAGES22
Experimental results show that the clouds and ground objects can be separated efficiently, and the method can implement rapid clouds detection and cloudiness calculation.
- 21
- 21
- Joint Spatial-Wavelet Dual-Stream Network for Super-Resolution21
The tailor-made Wavelet Features Adaptation (WFA) to adjust the wavelet coefficients for better compatibility with networks and Wavelet-Aware Convolutional blocks (WAC) to exploit features in the wavelets domain efficiently is designed.
- 20
- GaussianStego: A Generalizable Stenography Pipeline for Generative 3D Gaussians Splatting20
GaussianStego represents an initial exploration into the novel challenge of embedding customizable, imperceptible, and recoverable information within the renders produced by current 3D generative models, while ensuring minimal impact on the rendered content's quality.
- Instruction-ViT: Multi-Modal Prompts for Instruction Learning in ViT20
This work provided an innovative strategy to fuse multi-modal prompts with better performance and faster adaptability for visual classification models.
- Quorum sensing signal synthases enhance Vibrio parahaemolyticus swarming motility20
The data presented here suggest an important strategy of swarming regulation by the integration of QS and c‐di‐GMP signaling pathways in V. parahaemolyticus.
- 20
- Self-Supervised Learning of LiDAR 3D Point Clouds via 2D-3D Neural Calibration19
A novel self-supervised learning framework for enhancing 3D perception in autonomous driving scenes that focuses on 2D-3D neural calibration, and confirms that joint learning from different modalities significantly enhances the network’s understanding abilities and effectiveness of learned representation.
- 19
- 19
- Automatic Microscopy Analysis with Transfer Learning for Classification of Human Sperm19
A transfer learning method based on AlexNet to automatically classify the sperms into four different categories in terms of the World Health Organization (WHO) standards by analyzing their morphology, which shows that automatic sperm classification has great potential to replace manual sperm classification in the future.
- Complementary Network with Adaptive Receptive Fields for Melanoma Segmentation19
A novel complementary network with adaptive receptive filed learning is proposed, which introduces a foreground network to detect melanoma lesions and a background network to mask non-melanoma regions and enables the semi-supervised learning and improve the boundary-sensitivity.
- EndoBench: A Comprehensive Evaluation of Multi-Modal Large Language Models for Endoscopy Analysis18
EndoBench establishes a new standard for evaluating and advancing MLLMs in endoscopy, highlighting both progress and persistent gaps between current models and expert clinical reasoning.
- 18
- Models for multiple attribute decision making with picture fuzzy information18
This paper investigates the picture fuzzy multiple attribute decision making problems where the information about attribute weights is partly known or completely unknown, and establishes the grey relational analysis models to measure the grey relational degree between each alternative and the picture fuzzy ideal point.
- 18
- Dynamic feature splicing for few-shot rare disease diagnosis17
A dynamic feature splicing framework for few-shot rare disease diagnosis that explores cross-image channel relations and selects base-class channels with semantic consistency for explicit knowledge transfer and achieves superior performance against state-of-the-art approaches.
- 17
- Gradient modulated contrastive distillation of low-rank multi-modal knowledge for disease diagnosis17
This work proposes a novel Momentum-enriched Multi-Modal Low-Rank (M3LR) constraint to explore the high-order correlations and complementary information among different modalities, thus yielding more accurate multi-modal diagnosis.
- Comprehensive learning and adaptive teaching: Distilling multi-modal knowledge for pathological glioma grading17
A comprehensive learning and adaptive teaching framework is proposed to improve the performance of pathological grading models by transferring the privileged knowledge from the multi-modal teacher to the pathology student and a novel Saliency-Aware Masking strategy to explore richer disease-related features from both modalities.
- Comparing three types of real-time data collection techniques: Counting cameras, Wi-Fi sensors and GPS trackers17
This paper focuses on local and global real-time measurements, collected during the SAIL Amsterdam event using counting cameras, Wi-Fi sensors and GPS trackers, and estimates of counts, travel times, density and flow are estimated.
- Real‐Time Guidewire Tip Tracking Using a Siamese Network for Image‐Guided Endovascular Procedures16
A novel tracking framework based on a Siamese network with dual attention mechanisms combines self‐ and cross‐attention strategies for robust guidewire tip tracking that handles visual ambiguities, tissue deformations, and imaging artifacts through enhanced spatial‐temporal feature learning.
- 16
- 16
- 16
- Towards a general-purpose foundation model for functional MRI analysis15
The Neuroimaging Foundation Model with Spatial-Temporal Optimized and Representation Modelling (NeuroSTORM), which learns generalizable representations directly from four-dimensional fMRI volumes and enables efficient transfer to diverse downstream applications, is introduced.
- Polyp-Gen: Realistic and Diverse Polyp Image Generation for Endoscopic Dataset Expansion15
The Polyp-Gen is presented, the first full-automatic diffusion-based endoscopic image generation framework that can generate realistic and diverse endoscopic images for building reliable ADS and has shown remarkable zeroshot generalizability on other datasets.
- ConcealGS: Concealing Invisible Copyright Information in 3D Gaussian Splatting15
ConcealGS, a novel 3D steganography method that embeds implicit information into the explicit 3D representation of Gaussian Splatting, paves the way for advanced copyright protection and secure data transmission in the evolving landscape of 3D content creation and distribution.
- DEeR: Deviation Eliminating and Noise Regulating for Privacy-Preserving Federated Low-Rank Adaptation15
A novel privacy-preserving federated finetuning framework called Deviation Eliminating and Noise Regulating (DEeR) is proposed, which theoretically proves that the necessary condition to eliminate aggregation deviation is guaranteeing the equivalence between LoRA parameters of clients and achieves robust privacy protection and excellent finetuning performance.
- 15
- 15
- 15
- 15
- 14
- 14
- 14
- The autoPET3 Challenge: Automated Lesion Segmentation in Whole-Body PET/CT $\unicode{x2013}$ Multitracer Multicenter Generalization13
Three main conclusions can be drawn: (1) in-domain multitracer PET/CT segmentation is sufficient and probably approaching reader agreement; (2) compositional generalization to unseen tracer-center combinations remains an open problem mainly driven by systematic volume overestimation; (3) heterogeneity and case difficulty drive performance variation substantially more than the choice of algorithm among top-ranked teams.
- 13
- Relation-Guided Versatile Regularization for Federated Semi-Supervised Learning13
A novel FSSL framework, Relation-guided Versatile Regularization (FedRVR), consisting of versatile regularization at clients and relation-guided directional aggregation strategy at the server, which proves the superiority of FedRVR over state-of-the-art counterparts across various federated learning settings.
- STAR-RL: Spatial-Temporal Hierarchical Reinforcement Learning for Interpretable Pathology Image Super-Resolution13
This paper reformulate the SR problem as a Markov decision process of interpretable operations and adopt the hierarchical recovery mechanism in patch level, to avoid sub-optimal recovery in pathology image super-resolution problem.
- 13
- Bioactive Polyurethane–Poly(ethylene Glycol) Diacrylate Hydrogels for Applications in Tissue Engineering13
These hydrogels represent a unique and highly tailorable system for synthesizing PU-based synthetic extracellular matrices for tissue engineering applications and reveal that PU–PEG blended hydrogels support cell adhesion and viability when cell adhesion peptides are crosslinked within the hydrogel matrix.
- 13
- Unknown-Oriented Learning for Open Set Domain Adaptation13
A novel Unknown-Oriented Learning (UOL) framework for OSDA, and it is composed of three stages: true unknown excavation, false unknown suppression and known alignment, to excavate the diverse semantic information in unknown category.
- 13
- 13
- Comparison of ILK and ERP29 expressions in benign and malignant pancreatic lesions and their clinicopathological significances in pancreatic ductal adenocarcinomas13
Positive ILK and negative ERP29 expressions are associated with the progression of PDAC and poor prognosis in patients with PDAC.
- 12
- 12
- 11
- Track Any Anomalous Object: A Granular Video Anomaly Detection Pipeline11
An innovative VAD framework called Track Any Anomalous Object (TAO) is proposed, which introduces a Granular Video Anomaly Detection Framework that, for the first time, integrates the detection of multiple fine-grained anomalous objects into a unified framework.
- 11
- 11
- 11
- 11
- 11
- 11
- 11
- 10
- 10
- MAP: evaluation and multi-agent enhancement of large language models for inpatient pathways9
The IPDS benchmark is developed, comprising 51,274 cases across 9 triage departments, 17 major disease categories, and 16 treatment options, and the Multi-Agent Inpatient Pathways (MAP) framework is proposed, containing three specialized clinical agents: a triage agent for patient admission, a diagnosis agent for diagnostic decision-making, and a treatment agent for care planning.
- $\mathbf{X}^{\mathbf{2}}$-Gaussian: 4D Radiative Gaussian Splatting for Continuous-Time Tomographic Reconstruction9
This work proposes X2 Gaussian, a novel framework that enables continuoustime 4D-CT reconstruction by integrating dynamic radiative Gaussian splatting with self-supervised respiratory motion learning and introduces a physiology-driven periodic consistency loss that learns patient-specific breathing cycles directly from projections via differentiable optimization.
- DeMo: Deep Motion Field Consensus with Learnable Kernels for Two-view Correspondence Learning9
DeMo is proposed, a novel and cutting-edge network for outlier rejection, which possesses the capacity to fully capture global motion consensus clues by way of consensus interpolation over the entire high-dimensional motion field generated by putative correspondences.
- FedBM: Stealing knowledge from pre-trained language models for heterogeneous federated learning9
A novel framework called Federated Bias eliMinating (FedBM) to get rid of local learning bias in heterogeneous federated learning (FL), which mainly consists of two modules, i.e., Linguistic Knowledge-based Classifier Construction (LKCC) and Concept-guided Global Distribution Estimation (CGDE).
- 9
- 9
- 8
- 8
- 8
- 8
- 8
- 8
- 8
- 8
- Generalized Gradient Flow Based Saliency for Pruning Deep Convolutional Neural Networks8
The proposed gradient flow based saliency method, GFBS, is capable to identify subnetworks with comparable performance with the baseline model at an early training stage and could feasibly cooperate with the baseline networks and compress them with only negligible performance drop.
- A Morphometric Analysis of Commonly Used Craniometric Approaches for Freehand Ventriculoperitoneal Shunting8
Keen's point was the most accurately sited and was a predictor for optimal catheter position, and catheter tip location, not the entry site, predicted shunt survival.
- 8
- Exploring Gradient Flow Based Saliency for DNN Model Compression8
A new model pruning method from a new perspective of gradient flow that theoretically analyze the channel's influence based on Taylor expansion by integrating the effects of BN layer and ReLU activation function and the incorporation of the first-order Talyor polynomial of the scaling parameter and the shifting parameter in the BN layers is suggested to effectively indicate the significance of a channel in a DNN.
- Automated hepatobiliary toxicity prediction after liver stereotactic body radiation therapy with deep learning-based portal vein segmentation8
A novel framework for automated prediction of hepatobiliary (HB) toxicity after liver stereotactic body radiation therapy (SBRT) is proposed by combining deep learning-based auto-segmentation, PV anatomy analysis and the previously reported HB toxicity model.
- Ventricle Shape Analysis for Centenarians, Elderly Subjects, MCI and AD Patients8
Comparisons of volume and shape analysis suggest that ventricle shape characteristics could potentially be a more sensitive biomarker of AD progression than the traditionally assumedventricle volumes.
- Voxel-Level Brain States Prediction Using Swin Transformer7
This work shows promising evidence that the spatiotemporal organization of the human brain can be learned by a Swin Transformer model, at high resolution, which provides a potential for reducing the fMRI scan time and the development of brain-computer interfaces in the future.
- 7
- Harnessing Lightweight Transformer With Contextual Synergic Enhancement for Efficient 3D Medical Image Segmentation7
Light-UNETR is proposed, a lightweight transformer designed to achieve model efficiency and a Contextual Synergic Enhancement (CSE) learning strategy, which aims to boost the data efficiency of Transformers.
- 7
- 7
- 7
- 7
- 7
- X-GRM: Large Gaussian Reconstruction Model for Sparse-view X-rays to Computed Tomography6
X-GRM (X-ray Gaussian Reconstruction Model), a large feedforward model for reconstructing 3D CT volumes from sparse-view 2D X-ray projections, employs a scalable transformer-based architecture to encode sparse-view X-ray inputs, where tokens from different views are integrated efficiently.
- 6
- Mask-aware transformer with structure invariant loss for CT translation6
A Mask-Aware Transformer with structure invariant loss for CT translation, which presents the first effort to exploit this domain knowledge for CT translation and its superiority to the state-of-the-art CT translation methods.
- 6
- 6
- 5
- 5
- 5
- Global and multi-partition local network analysis of scalp EEG in West syndrome before and after treatment5
The results show that the bidirectional network has better significant differentiation and localized brain regions variability in features and differences in the ability to interact with information between brain regions have potential as biomarkers for medication assessment in WEST syndrome.
- 5
- 5
- 5
- Gaze-based attention to improve the classification of lung diseases5
The proposed study has demonstrated the potential benefits of using gaze information in computerized diagnostic solutions and improved the diagnostic accuracy to 0.714 of the area under receiving operator curve (AUC) from 0.681 AUC obtained for the same ResNet34 trained to generate binary pathology labels alone.
- Deep Learning-Driven Models for Endoscopic Image Analysis5
This chapter takes two tasks in endoscopic image analysis as examples, to survey the state of the art, recent advances, and future directions of CAD applications, especially with regard to deep learning models.
- 5
- 4
- Infproto-Powered Adaptive Classifier and Agnostic Feature Learning for Single Domain Generalization in Medical Images4
This work designs a hierarchical domain-adaptive classifier to elasticize the model for varying domains and introduces probabilistic modeling and provide a theoretic upper bound to implicitly perform the infinite prototype sampling in the optimization of InfProto.
- 4
- 4
- 4
- Deep 3D Dose Analysis for Prediction of Outcomes After Liver Stereotactic Body Radiation Therapy4
This paper pioneers the concept of deep 3D dose analysis for outcome prediction after liver stereotactic body RT (SBRT), develops tools for unification of dose plans into the same anatomy space, and assembles the first volumetric liver atlas of the critical-to-spare liver regions.
- Clip-Driven Universal Model for Multi-Material Decomposition in Dual-Energy CT3
A CLIP-Driven Universal Model for adaptive Multi-Material Decomposition (MMD) that utilizes the semantic capabilities of text embeddings from Contrastive Language-Image Pre-training (CLIP), allowing a single network to manage structured feature embedding for multiple materials.
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- Liver Lesion Detection Based on Two-Stage Saliency Model with Modified Sparse Autoencoder3
A novel saliency model for lesion detection in CT images by segmenting the image into multi-scale patch sizes and adopting Locality-constrained Linear Coding method to encode the obtained discriminative features of each patch is proposed.
- 3
- Hierarchical key frame extraction for wireless capsule endoscopy video based on the saliency map3
A novel hierarchical key frame extraction algorithm based on the saliency map to automatically select a small number of key informative frames in the WCE video to assist the diagnostic process.
- 3
- 3
- FoME: A foundation model for EEG using adaptive temporal-lateral attention scaling2
FoME (Foundation Model for EEG), a novel approach using adaptive temporal-lateral attention scaling to address above-mentioned challenges, establishes a new paradigm for EEG analysis, offering a versatile foundation that advances brain-computer interfaces, clinical diagnostics, and cognitive research across neuroscience and related fields.
- 2
- 2
- 2
- 2
- 2
- 2
- MAN-GAN: a mask-adaptive normalization based generative adversarial networks for liver multi-phase CT image generation2
MAN-GAN demonstrates the feasibility of liver MPECT image translation based on non-contrast images and achieves state-of-the-art performance via the subtraction strategy and has great potential for solving the dilemma of liver CT contrast canning and aiding further liver interaction clinical scenarios.
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- Contrastive machine learning reveals species -shared and -specific brain functional architecture2
A novel contrastive machine learning method, called shared-unique variation autoencoder (SU-VAE), to allow disentanglement of the species-shared and species-specific functional connectome variation between macaque and human brains on large-scale resting-state fMRI datasets.
- Rethinking Data Perturbation and Model Stabilization for Semi-supervised Medical Image Segmentation2
DPMS is proposed, which adopts a plain teacher-student framework with a standard supervised loss and unsupervised consistency loss and can obtain new state-of-the-art performance on the public 2D ACDC and 3D LA datasets across various semi-supervised settings.
- 2
- Neural Representation of Exogenous and Endogenous Temporal Expectations Based on fMRI1
Temporal expectations are considered as implicit timing, which is different from explicit timing and exogenous and endogenous temporal expectations could be divided into exogenous and endogenous temporal expectations.
- 1
- 1
- 1
- 1
- GAGM: Geometry-aware graph matching framework for weakly supervised gyral hinge correspondence1
A Geometry-Aware Graph Matching framework is proposed for weakly supervised gyral hinge correspondence solely based on brain prior information, which leverages prior knowledge of the multi-scale relationship between GHs and brain regions and incorporates inter-scale semantic consistency to ensure both intra-region consistency and inter-region variability of GH features, ultimately achieving accurate GH matching.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- Current Research on the Application of Xuefu Zhuyu Decoction and Its Modifications in the Treatment of Lung Cancer1
The mechanisms of action and clinical efficacy of Xuefu Zhuyu Decoction and its modifications in treating lung cancer and associated conditions are summarized and analyzed.
- 1
- 1
- 1
- 1
- 1
- 1
- MRM++: Enhanced Masked Relation Modeling for Multi-Modal Medical Pre-training–
This work attempts to harness the complementary information of multi-modal medical data to perform self-supervised pre-training and proposes MRM++, an enhanced masked relation modeling paradigm that can adaptively transform knowledge from the pre-trained model to be compatible with downstream tasks while maintaining transferable information.
- –
- Dual Adaptive Grouped Multivariate Variational Mode Decomposition for Physiological Signals–
DA-GMVMD makes multichannel EEG and ECG decomposition less sensitive to a manually preset group number and common mode number and adapts the channel groups and their mode numbers to the signal content while retaining the evaluated EEG and ECG information in the reconstructed signals.
- –
- Toward Pathology-guided Illumination Optimization for Neuromodulation with PhomiNeuro–
PhomiNeuro is presented, a foundation-model-encoded differentiable surrogate for time-resolved LF modeling and pathology-guided inverse design and position PhomiNeuro as a highly extensible translational framework toward personalized treatment planning and digital twin development in precision neuromodulation.
- Brain network construction and analysis for epilepsy: A methodology review–
This work reorganizes the literature around five prevalent research tasks: seizure prediction, seizure detection, seizure type classification, clinical correlation analysis, and epileptic foci localization, and consolidates the conceptual foundations and construction methods for functional, structural, and effective networks, together with their dynamic and multimodal extensions.
- –
- –
- –
- –
- –
- –
- –
- Search for e + e − → φ χ c 0 and φ η c 2 ( 1 D ) at center-of-mass energies from 4.47 to 4.95 GeV–
Higher limits are set on the dressed cross sections for electron-positron collisions recorded by the BESIII detector at the BEPCII storage ring and across center-of-mass energies from 4.95 GeV.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-14. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.