System logs have been critical for analyzing the operational status and abnormal behavior of highly distributed and heterogeneous edge computing nodes. In edge environments, logs exhibit cross-event and cross-field structural interactions, making it difficult to uncover potential anomaly patterns from isolated events. Moreover, sparse annotations and varying log formats limit the effectiveness of existing methods. To address these challenges, we propose a graph neural network (GNN) anomaly detection framework with prompt learning. It leverages few-shot prompt learning to automatically extract key fields and constructs a weighted directed graph that jointly models semantic embeddings and temporal dependencies, fully representing the structural interactions and semantic associations across events and fields. Furthermore, the framework performs graph-level anomaly detection by jointly optimizing graph representation learning and classification objective within an enhanced one-class directed graph convolutional network, enabling effective identification of global structural anomaly patterns in log graphs. Experimental results demonstrate that the proposed method achieves an average F1-score of 93.3%, surpassing the current state-of-the-art (SOTA) methods by 6.93%.

Designing a quad mesh that meets aesthetic, anatomical, and numerical requirements often requires meticulous manual effort in conventional methods, making quadrilateral remeshing an “art of design”. Neural networks hold significant promise for automating this process. However, current approaches that directly predict cross fields cannot properly handle the discontinuous behavior of smooth cross fields: minor shape variations can lead to substantial changes in the cross field, even when singularities remain largely unchanged. Therefore, such methods often result in non-smooth outputs when combining multiple singularity instances. To avoid such discontinuity, we propose to learn the sparse singularities, including their locations and indices, then let the non-neural conventional method to smoothly connect them. The imbalanced ratio of singular and regular vertices poses a significant challenge for learning. We convert them into a geodesic distance field and an over-sampled index field to address it. This carefully designed two-stage strategy satisfies several key requirements, such as coordinate invariance and tessellation insensitivity, while enabling the generation of smooth cross fields with varying topologies. By shifting the focus from directly learning the cross field to learning singularities, we also simplify the dataset preparation process by requiring only sparse annotations.

Scribble-based weakly supervised segmentation methods have shown promising results in medical image segmentation, significantly reducing annotation costs. However, existing approaches often rely on auxiliary tasks to enforce semantic consistency and use hard pseudo labels for supervision, overlooking the unique challenges faced by models trained with sparse annotations. These models must predict pixel-wise segmentation maps from limited data, making it crucial to handle varying levels of annotation richness effectively. In this paper, we propose MaCo, a weakly supervised model designed for medical image segmentation, based on the principle of "from few to more." MaCo leverages Masked Context Modeling (MCM) and Continuous Pseudo Labels (CPL). MCM employs an attention-based masking strategy to perturb the input image, ensuring that the model's predictions align with those of the original image. CPL converts scribble annotations into continuous pixel-wise labels by applying an exponential decay function to distance maps, producing confidence maps that represent the likelihood of each pixel belonging to a specific category, rather than relying on hard pseudo labels. We evaluate MaCo on three public datasets, comparing it with other weakly supervised methods. Our results show that MaCo outperforms competing methods across all datasets, establishing a new record in weakly supervised medical image segmentation.

Pixel-level annotation of lung cavities (LCs) in computed tomography (CT) images is challenging due to their morphological diversity and complexity. Weakly supervised semantic segmentation (WSSS) methods, which utilize sparse annotations (e.g., image-level labels), offer a promising solution. However, existing WSSS approaches often generate coarse pseudo-labels and lack sufficient spatial supervision, resulting in under- or over-segmentation of irregular lesions. To address these limitations, we introduce several key innovations. First, we propose a novel Graph-based Affinity Network (GA-Net) that, unlike conventional methods relying on low-level pixel features, models long-range contextual relationships and structural dependencies using a superpixel graph and learned edge inference kernel, enabling structure-aware pseudo-label refinement for complex lesion morphology. Second, we introduce region-wise affinity propagation, which refines segmentation by propagating activations within semantically coherent 3D regions, offering more precise control over under-/over-segmentation compared to global affinity methods. Additionally, we incorporate Exponential Moving Average (EMA) ensembling for training stability and a scribble-based segmentation module that utilizes pseudo-label contours to provide direct boundary supervision. Extensive experiments on three benchmark datasets demonstrate that our method outperforms existing state-of-the-art medical WSSS techniques, achieving precise and reliable segmentation of complex LCs in CT scans.

Urban airborne laser scanning (ALS) point clouds cover extensive geographical areas, rendering dense point-level annotation economically prohibitive and limiting the feasibility of fully supervised learning. In weakly supervised settings for urban ALS data, the natural long-tailed class distribution—where ground and building points dominate and smaller objects are rare—combined with the use of fixed pseudo-label thresholds under sparse annotations exacerbates confirmation bias and increases prediction uncertainty. This ultimately restricts the effective utilization of unlabeled data during training. To overcome these challenges, we propose a pseudo-label confidence-calibrated curriculum learning framework designed for weakly supervised ALS point cloud classification. The framework introduces a confidence-aware self-adaptive soft gating (CSS) mechanism that dynamically adjusts category-specific thresholds online using exponential moving average statistics and scene-aware normalization, eliminating the need for manual scheduling while improving pseudo-label quality. In addition, a reliability-driven soft selection (RSS) constraint is incorporated, in which each point is assigned a comprehensive reliability score that integrates prediction confidence, entropy clarity, and cross-augmentation consistency, enabling adaptive soft weighting to replace hard pseudo-label selection and achieve more balanced sample utilization. These components are further integrated into a unified pseudo-label confidence-calibrated curriculum learning framework (P3CL) that progressively shifts the model’s focus from high-certainty samples to more ambiguous ones, effectively mitigating confirmation bias. Extensive experiments on three public ALS benchmarks demonstrate that the proposed method consistently outperforms existing weakly supervised approaches and achieves competitive performance compared with several fully supervised models.

IntroductionGiven the high prevalence of vertebral fractures following radiotherapy in patients with metastatic spine disease, torso muscle segmentation is necessary for biomechanical modeling of vertebral loading, permitting individualized evaluation of fracture risk.MethodsIn this study, we developed and validated a deep-learning model for full volumetric segmentation of the thoracic and abdominal spinal musculature in cancer patients with metastatic spine disease from sparsely annotated clinical CT image data. We obtained CT data for 148 metastatic spine disease patients undergoing radiotherapy treatment, and an external set of randomly selected 30 subjects from the National Lung Screening Trial. We extracted 1924 axial CT images at the midpoint of each vertebral level (T4 to L4) and manually labeled the key extensor and flexor muscles (up to 8 muscles per side) at each level. We trained a 2D nnU-Net deep-learning (DL) model to segment each muscle and, using these sparse annotations, trained the model to segment each muscle’s 3D volume per spine. Two experienced radiologists independently and blindly evaluated the anatomical fidelity of the segmentations using a Likert scale, for 1) manual- and 2) DL-segmentation, 3) random test samples from the muscle’s 3D volume and 4) an external NLST CT data.ResultsThe DL method achieved comparable performance to manual segmentations with a mean Dice score above 0.769. Mann-Whitney test analysis showed that the radiologist ratings of DL-generated muscle segmentations were noninferior to the manual segmentation for each muscle.DiscussionDemonstrating excellent performance for rapid, high-anatomical fidelity 3D segmentation of the main flexor, extensor, and stabilizing thoracolumbar muscles, the DL model from clinical CT scans, this development holds significant potential for reducing the manual effort required to generate individualized musculoskeletal models in cancer patients.

Atrial fibrillation, characterized by high prevalence and poor prognosis, presents a significant global health burden. Accurate segmentation and measurement of left ventricular and left atrial appendage morphology and function are essential for reliable risk assessment. However, these tasks are hindered by ambiguous bound-aries, complex cardiac motion, and sparse annotations. To address these challenges, we propose a Keypoint-Guided Medical Video Segmentation Model with Spatiotemporal Feature Fusion (KG-STS). First, we propose a shape-constrained point encoder that explicitly encodes boundary points to improve the representation of ambiguous boundaries. Next, we introduce a motion-aware alignment module that models cardiac motion by forming coherent motion information across frames. Building on these two modules, we develop a keypoint-guided spatiotemporal feature fusion module that integrates spatial boundary representations with temporal motion cues to enhance decoding features under sparse annotations, enabling temporally consistent segmentation and supporting morphological measurement. We evaluate the segmentation and measurement performance of our method on a self-constructed multi-view transesophageal echocardiography dataset and two publicly available transthoracic echocar-diography datasets. The results demonstrate that KG-STS achieves superior temporal consistency in segmentation and higher accuracy in morphological measurements compared to competing methods.

Pheochromocytoma is a rare urological adrenal tumor disease. Automated segmentation of pheochromocytomas from computed tomography (CT) is essential for diagnosis and treatment. However, this task is a challenging one due to issues such as blurred boundaries, irregular shapes, variations in location and size, and the lack of annotated images for training. To address these issues, we propose a semi-supervised framework for pheochromocytoma segmentation that primarily consists of a dynamic uncertainty rectification mechanism and a supervised strategy based on SAM-Med3D prior knowledge. First, we design a semi-supervised segmentation model comprising a shared encoder and multiple independent decoders that dynamically select pseudo labels from the different decoder outputs. To mitigate the risk of unreliable predictions caused by sparse annotations during training, we introduce uncertainty estimation to prioritize reliable outputs. Additionally, an Attentional Convolution Block (ACB) is designed in the encoding stage to fully utilize both global and local features, improving tumor recognition in segmentation. Furthermore, SAM-Med3D prior knowledge is incorporated into the framework as supplementary supervisory information, aiding the model in learning from limited labeled data. To eliminate the labor-intensive requirement for manual prompts in SAM-Med3D, we leverage pseudo labels to generate high-quality mask prompts, thus transforming the clinical workflow. Experiments on two pheochromocytoma datasets from different centers demonstrate that our proposed method achieves competitive performance.

Recently, deep neural networks have improved the performance of polarimetric synthetic aperture radar (PolSAR) image classification. Nonetheless, traditional approaches based on deep learning are highly dependent on substantial labeled datasets, which is difficult to obtain for PolSAR images. To tackle this issue, self-supervised learning is introduced to the studied task, exploiting hidden information within massive unlabeled data. Current self-supervised methods predominantly rely on a single paradigm, either generative or contrastive learning, yet each captures only partial features critical for PolSAR image classification. Therefore, to achieve comprehensive PolSAR image interpretation, we propose GHSS-Net, a graph-enhanced hybrid self-supervised framework for PolSAR image classification. Integrating graph-based generative learning and pixel-level contrastive learning through a dual-branch architecture, GHSS-Net improves classification results in scenarios with sparse annotations. Specifically, the generative learning branch utilizes superpixel-based graph neural networks with a low-rank constraint for representation learning by masking and reconstructing graphs. To compensate for the lack of fine-grained information, the contrastive learning branch captures pixel-level features through an implicit comparison process. We validate the effectiveness of GHSS-Net on four benchmark datasets. Experimental results substantiate that GHSS-Net attains state-of-the-art performance compared with other competitors with limited labels.

Multi-object tracking (MOT) is crucial for applications such as autonomous driving and robotics, yet traditional image-based methods struggle in high-speed scenarios due to motion blur and temporal gaps caused by low frame rates. Spike cameras, with their ability to continuously record spatiotemporal signals, overcome these limitations. However, existing spike-based methods often rely on intermediate image reconstruction or discrete clustering, limiting real-time performance and temporal continuity. To address this, we propose SNNTracker, the first fully spiking neural network (SNN)-based MOT algorithm tailored for spike cameras. SNNTracker integrates a dynamic neural field (DNF)-based attention mechanism for target detection and a winner-take-all (WTA)-based tracking module with online spike-timing-dependent plasticity (STDP) for adaptive learning of object trajectories. By directly processing spike streams without reconstruction, SNNTracker reduces latency, computational overhead, and dependency on image quality, making it ideal for ultra-high-speed environments. It maintains robust, continuous tracking even under occlusions, severe lighting variations, or temporary object disappearance, by leveraging SNN-estimated motion predictions and long-term online clustering. We construct three types of spike-camera MOT datasets covering dense and sparse annotations across diverse real-world scenarios, including camera ego-motion, deformable and ultra-fast motion (up to 2600 RPM), occlusion, indoor/outdoor lighting changes, and low-visibility tracking. Extensive experiments demonstrate that SNNTracker consistently outperforms state-of-the-art MOT methods-both ANN- and SNN-based-achieving MOTA scores above 96% and up to 100% in many sequences. Our results highlight the advantages of spike-driven SNNs for low-latency, high-speed, and label-free multi-object tracking, advancing neuromorphic vision for real-time perception.

Pixel-level annotation of lung cavities (LCs) in computed tomography (CT) images is challenging due to their morphological diversity and complexity. Weakly supervised semantic segmentation (WSSS) methods, which utilize sparse annotations (e.g., image-level labels), offer a promising solution. However, existing WSSS approaches often generate coarse pseudo-labels and lack sufficient spatial supervision, resulting in under- or over-segmentation of irregular lesions. To address these limitations, we introduce several key innovations. First, we propose a novel Graph-based Affinity Network (GA-Net) that, unlike conventional methods relying on low-level pixel features, models long-range contextual relationships and structural dependencies using a superpixel graph and learned edge inference kernel, enabling structure-aware pseudo-label refinement for complex lesion morphology. Second, we introduce region-wise affinity propagation, which refines segmentation by propagating activations within semantically coherent 3D regions, offering more precise control over under-/over-segmentation compared to global affinity methods. Additionally, we incorporate Exponential Moving Average (EMA) ensembling for training stability and a scribble-based segmentation module that utilizes pseudo-label contours to provide direct boundary supervision. Extensive experiments on three benchmark datasets demonstrate that our method outperforms existing state-of-the-art medical WSSS techniques, achieving precise and reliable segmentation of complex LCs in CT scans.

Whole-slide histology images (WSIs) can exceed 100 k × 100 k pixels, making direct pixel-level segmentation infeasible and requiring patch-level classification as a practical alternative for downstream WSI segmentation. However, most approaches either treat patches independently, ignoring spatial and biological context, or rely on deep graph models prone to oversmoothing and loss of local tissue detail. We present WSI-GT (Pseudo-Label Guided Graph Transformer), a simple yet effective architecture that addresses these challenges and enables accurate WSI-level tissue segmentation. WSI-GT combines a lightweight local graph convolution block for neighborhood feature aggregation with a pseudo-label guided attention mechanism that preserves intra-class variability and mitigates oversmoothing. To cope with sparse annotations, we introduce an area-weighted sampling strategy that balances class representation while maintaining tissue topology. WSI-GT achieves a Macro F1 of 0.95 on PATH-DT-MSU WSS2v2, improving by up to 3 percentage points over patch-based CNNs and by about 2 points over strong graph baselines. It further generalizes well to the Placenta benchmark and standard graph node classification datasets, highlighting both clinical relevance and broader applicability. These results position WSI-GT as a practical and scalable solution for graph-based learning on extremely large images and for generating clinically meaningful WSI segmentations.

Leveraging sparse annotations for leukemia diagnosis on the large leukemia dataset.

Long-range diffusion for weakly camouflaged object segmentation.

Intrinsic motivation-based exploration for enhancing tuberculosis lesion discovery in sparse annotation chest X-ray datasets.

The success of deep learning in 3D medical image segmentation hinges on training with a large dataset of fully annotated 3D volumes, which are difficult and time-consuming to acquire. Although recent foundation models (e.g., segment anything model, SAM) can utilize sparse annotations to reduce annotation costs, segmentation tasks involving organs and tissues with blurred boundaries remain challenging. To address this issue, we propose a region uncertainty estimation framework for Computed Tomography (CT) image segmentation using noisy labels. Specifically, we propose a sample-stratified training strategy that stratifies samples according to their varying quality labels, prioritizing confident and fine-grained information at each training stage. This sample-to-voxel level processing enables more reliable supervision information to propagate to noisy label data, thus effectively mitigating the impact of noisy annotations. Moreover, we further design a boundary-guided regional uncertainty estimation module that adapts sample hierarchical training to assist in evaluating sample confidence. Experiments conducted across multiple CT datasets demonstrate the superiority of our proposed method over several competitive approaches under various noise conditions. Our proposed reliable label propagation strategy not only significantly reduces the cost of medical image annotation and robust model training but also improves the segmentation performance in scenarios with imperfect annotations, thus paving the way towards the application of medical segmentation foundation models under low-resource and remote scenarios. Code will be available at https://github.com/KHan-UJS/NoisyLabel.

BackgroundThe development and application of deep learning-based models have seen significant success in medical image segmentation, transforming diagnostic and treatment processes. However, these advancements often rely on large, fully annotated datasets, which are challenging to obtain due to the labour-intensive and costly nature of expert annotation. Therefore, we sought to explore the feasibility and efficacy of training 2D models under severe annotation constraints, aiming to optimise segmentation performance while minimising annotation costs.MethodsWe propose an incremental 2D self-labelling framework for segmenting 3D medical volumes from a single annotated slice per volume. A 2D U-Net is first trained on these central slices. The model then iteratively generates and filters pseudo-labels for adjacent slices, progressively fine-tuning itself on an expanding dataset. This process is repeated until the entire training set is pseudo-labelled to produce the final model.ResultsOn brain MRI and liver CECT datasets, our self-labelling approach improved segmentation performance compared to using only the sparse ground-truth data, increasing the Dice Similarity Coefficient and Intersection over Union by up to 15.95% and 26.75%, respectively. It also improved 3D continuity, reducing the 95th percentile Hausdorff Distance from 69.88 mm to 36.46 mm. Parameter analysis revealed that a gradual propagation of high-confidence pseudo-labels was most effective.ConclusionOur framework demonstrates that a computationally efficient 2D model can be leveraged through self-labelling to achieve robust 3D segmentation performance and coherence from extremely sparse annotations, offering a viable solution to reduce the annotation burden in medical imaging.

Localizing macromolecules in crowded cellular cryo-electron tomography (cryo-ET) images or tomograms is crucial for determining their in situ structures. Traditional template matching-based approaches for this task suffer from template-specific biases and have low throughput. Given these problems, learning-based solutions are necessary. However, the paucity of annotated data for training poses substantial challenges for such learning-based methods. Moreover, preparing extensively annotated cellular tomograms for training macromolecule localization methods is extremely time-consuming and burdensome due to the large volume and low signal-to-noise ratio of the tomograms. In this work, we developed TomoPicker, an annotation-efficient macromolecule localization method for tomograms. To achieve such annotation-efficiency, TomoPicker regards macromolecule localization as a voxel classification problem and solves it with two different positive-unlabeled learning approaches. We evaluated TomoPicker on two experimental cryo-ET datasets of crowded eukaryotic cells and one experimental dataset of relatively less crowded prokaryotic cell. We observed that, with only 10 annotated macromolecule locations, TomoPicker with positive unlabeled learning achieved a performance comparable to that of state-of-the-art supervised methods trained with several hundred annotations. In other words, TomoPicker achieved plausible segmentation with up to 98% less data compared with supervised learning-based methods. Furthermore, it demonstrated substantial improvements over existing learning-based macromolecule localization methods under sparse annotation scenarios.

The accurate identification of pharmaceutical pills is crucial for ensuring medication safety, enhancing healthcare delivery, and supporting regulatory compliance. While manual verification remains common, it is often time-consuming and prone to human error. In response, this chapter proposes an advanced deep learning-based system for real-time pill detection and classification using the YOLOv5s object detection framework. A novel component of this system is the integration of a Deep Text Spotter (DTS) module, further enhanced by a character-level Recurrent Neural Network (RNN) with coordinate encoding to improve imprint recognition. This combination effectively addresses spatial inconsistencies and mitigates optical character recognition (OCR) errors that typically arise in incomplete or unclear pill imprints. To address limitations in small-object detection and sparse annotations, a training strategy is introduced that enables generalization from single-object scenarios to multi-object contexts. The system is trained on a hybrid dataset that includes the National Library of Medicine (NLM) Pill Dataset and real-world pill images with diverse morphologies and imprint styles. The proposed framework achieves a detection accuracy of 97.8%, outperforming baseline models across precision, recall, and F1-score metrics. These results suggest strong potential for adoption in pharmaceutical quality assurance, clinical verification systems, and forensic pill analysis.

Coronary artery calcification segmentation with sparse annotations in intravascular OCT: Leveraging self-supervised learning and consistency regularization.