Medical artificial intelligence (AI) plays a central role in advancing the rapid development of AI-driven healthcare technologies. Among these applications, medical visual question answering (VQA) has emerged as a critical component, garnering significant research attention. However, current approaches to medical VQA predominantly prioritize achieving high accuracy on constrained datasets, overlooking the multifaceted challenges inherent to real-world clinical deployment. In practical settings, two key factors significantly influence the input reliability of medical VQA systems. First, medical imaging processes inherently introduce noise due to technical and environmental variability. Second, linguistic diversity in patient-provider interactions leads to heterogeneous phrasing of clinical questions. Despite their inevitability in real-world scenarios, these challenges remain understudied in existing research, largely due to the absence of dedicated benchmarks. To address this gap, we propose a benchmark for robust medical VQA (RoM-VQA), a novel benchmark designed to evaluate model robustness from both medical imaging and linguistic perspectives. This benchmark extends existing datasets through systematic enhancements, simulating real-world noise and language variability. During its development, a multiLLM collaborative framework is implemented to minimize human intervention. With the active involvement of medical experts, our paper constructs a diverse, high-quality, and large-scale medical visual question answering dataset designed to evaluate model robustness. Evaluation of existing medical VQA models on RoM-VQA demonstrates their instability when confronted with diverse noise and variability. These results highlight a significant gap in medical AI research, where current models lack the robustness necessary for practical deployment.

Federated semi-supervised learning (FSSL) enables collaborative training on distributed medical data while preserving privacy, but faces challenges from data heterogeneity and class imbalance. These issues degrade pseudo-labeling quality and introduce confirmation bias. To overcome these limitations, this paper proposes FedPPL, a novel framework for federated medical image classification. FedPPL comprises two key components: Prototype-Aware Thresholding (PAT), which adaptively adjusts pseudo-labeling thresholds using global to mitigate confirmation bias, and Prototype Contrastive Learning (PCL), which enhances feature discriminability to boost accuracy. Experiments on FedISIC2019 and MedMNIST demonstrate that FedPPL achieves more robust and balanced performance than state-of-the-art methods, proving its potential for building reliable and privacy-preserving diagnostic models.

Automatic medical report generation (MRG) enhances clinical efficiency but still remains challenging due to cross-modal misalignment between lesions and text, exacerbated by data bias from scarce abnormalities and excessive normal cases diluting focus on critical lesions. To rectify these deficiencies, this paper proposes a new method called Multi-level Contrastive Learning with Feature Enhancement Network (MCL-FENet). Visual Feature Enhancement Module (VFEM) enhances channel-wise discriminability to improve lesion localization. Cross-Modal Feature Enhancement Module (CMFEM) integrates historical report semantics with current images via multimodal fusion, emulating radiologists' diagnostic reasoning. Furthermore, Global Contrastive Learning Module (GCLM) and Local Contrastive Learning Module (LCLM) jointly improve cross-modal alignment at both semantic and fine-grained levels. Experiments on the IU X-ray dataset demonstrate that MCL-FENet outperforms existing state-of-the-art methods.

Automatically generating natural, diverse and rhythmic human dance movements driven by music is vital for virtual reality and film industries. However, generating dance that naturally follows music remains a challenge, as existing methods lack proper beat alignment and exhibit unnatural motion dynamics. In this paper, we propose Danceba, a novel framework that leverages gating mechanism to enhance rhythm-aware feature representation for music-driven dance generation, which achieves highly aligned dance poses with enhanced rhythmic sensitivity. Specifically, we introduce Phase-Based Rhythm Extraction (PRE) to precisely extract rhythmic information from musical phase data, capitalizing on the intrinsic periodicity and temporal structures of music. Additionally, we propose Temporal-Gated Causal Attention (TGCA) to focus on global rhythmic features, ensuring that dance movements closely follow the musical rhythm. We also introduce Parallel Mamba Motion Modeling (PMMM) architecture to separately model upper and lower body motions along with musical features, thereby improving the naturalness and diversity of generated dance movements. Extensive experiments confirm that Danceba outperforms state-of-the-art methods, achieving significantly better rhythmic alignment and motion diversity. Project page: https://danceba.github.io/ .

In medical image segmentation tasks, diffusion models have exhibited significant potential. However, mainstream diffusion models show drawbacks including multiple sampling times and slow prediction results. Recently, as a standalone generative network, consistency models have resolved the existing issue. Compared to diffusion models, consistency models can lower the sampling times to once, not only achieving similar generative effects but also significantly accelerating training and prediction. However, they are not suitable for image segmentation tasks. Meanwhile, their application in the medical imaging field has not yet been investigated. Therefore, this study employs the consistency model to perform medical image segmentation tasks, designing multi-scale feature signal supervision modes and loss function guidance to realize model convergence. Experiments have demonstrated that the CTS model is capable of obtaining better medical image segmentation results with a single sampling during the test phase.

Medical Visual Question Answering (Med-VQA) aims to accurately answer clinical questions related to medical images. Due to the challenges in collecting medical images and the limited scale of datasets, extracting image features has become highly challenging. Traditional Med-VQA tasks are often treated as multi-class classification problems, overlooking the relationships between candidate answers. In this paper, a novel answer selection-based loss function is proposed to quantify the differences between original answers, addressing the aforementioned issues. For the answer with a higher probability among the selected answers, the similarity between the predicted answer and the true answer is calculated, and the loss is derived. Additionally, a representation enhancement module is introduced that focuses more on regions receiving common attention from various feature extractors. Experimental results demonstrate that the proposed method exhibits strong performance on multiple Med-VQA datasets and effectively improves the accuracy of existing Med-VQA models. Furthermore, the method can be seamlessly integrated into existing Med-VQA models to enhance their accuracy.

Federated learning enables collaborative learning across distributed medical institutions without centralizing data. However, existing studies often overlook class imbalance in medical images, which can degrade model performance, especially for minority classes. In this paper, we propose FedCRGP, a method that uses Gaussian-distributed prototypes to address class imbalance in federated medical image classification. These prototypes incorporate class variance to enhance representation learning. Class-aware Gaussian Prototype Learning (CGPL) is introduced to improve intra-class similarity and reduce interclass similarity by aligning class features around the Gaussian-distributed prototypes. Local and global features sampled from these prototypes are fused to retrain the classifier, mitigating bias caused by class imbalance. Experiments on two medical image classification datasets demonstrate its superior performance, particularly in multi-class scenarios.

Drag-based image editing using generative models provides precise control over image contents, enabling users to manipulate anything in an image with a few clicks. However, prevailing methods typically adopt $n$-step iterations for latent semantic optimization to achieve drag-based image editing, which is time-consuming and limits practical applications. In this paper, we introduce a novel one-step drag-based image editing method, i.e., FastDrag, to accelerate the editing process. Central to our approach is a latent warpage function (LWF), which simulates the behavior of a stretched material to adjust the location of individual pixels within the latent space. This innovation achieves one-step latent semantic optimization and hence significantly promotes editing speeds. Meanwhile, null regions emerging after applying LWF are addressed by our proposed bilateral nearest neighbor interpolation (BNNI) strategy. This strategy interpolates these regions using similar features from neighboring areas, thus enhancing semantic integrity. Additionally, a consistency-preserving strategy is introduced to maintain the consistency between the edited and original images by adopting semantic information from the original image, saved as key and value pairs in self-attention module during diffusion inversion, to guide the diffusion sampling. Our FastDrag is validated on the DragBench dataset, demonstrating substantial improvements in processing time over existing methods, while achieving enhanced editing performance.

Medical images are complex, and the annotation of medical images requires high expertise. It would be time-consuming and costly to annotate directly for experts. As a result, one of the primary challenges currently faced by medical visual question answering (VQA) is the lack of large-scale annotated data. To address this issue, a Weak Supervision-based Robust Pretraining (WSRP) method for medical VQA is proposed. Specifically, our method builds upon a contrastive language-image pretraining framework by introducing adversarial training. However, the contrastive language-image pretraining framework, by treating each image-text pair as a separate category, may lead to class collision problems, thereby affecting the quality of image representation. Therefore, weakly supervised contrastive learning is introduced to generate weak labels, enabling the model to learn fine-grained feature representations. The proposed weak supervision-based robust pretraining method for medical VQA is empirically evaluated and experimental results on public datasets demonstrate its superior performance.

Sequential recommendations are designed to capture user preferences based on their past actions and predict the items they may interact with in the next moment. Benefiting from the self-attention mechanism, methods that utilize side information (such as item categories or brand) to improve the prediction performance of sequential recommendation have yielded promising results. Previous approaches typically directly fuses side information embeddings into item embeddings as inputs to the model. However, this fusion approach overlooks the distinctions in various types of information in sequential pattern inference, and also failing to fully model the relationship between items and side information. In this work, we propose a Dual-Path Side Information Fusion method (DPIF) to better utilize side information for improved recommendation performance. Our model employs two parallel paths for side information fusion modeling. One path obtains the relationship representation within the items and the side information, and the other path obtains the relationship representation between the items and the side information. Subsequently, an attention-based adaptive fusion module is utilized to combine inter-attribute relationship and intra-attribute relationship representation, generating the final user preferences. Extensive experiments were conducted on four real-world datasets, demonstrating the effectiveness of the introduced model. Our source code is available at https://github.com/ZhangYu-x/DPIF.

Abstract Skin cancer is one of the deadliest cancers, and it has been widely developed worldwide since the last decade. Malignant melanoma is currently the most deadly skin cancer. If malignant melanoma is diagnosed at an early stage, the probability of patients being cured will be greatly improved. At present, most existing skin lesion image classification methods only use deep learning. However, the multi‐modal features of skin lesions in the medical domain are not well utilized and integrated. To reduce the classification error of the skin lesion images caused by the complexity and subjectivity of visual interpretation, a malignant melanoma dermoscopy image classification method based on multi‐modal medical features is proposed in this paper which is inspired by the fuzzy decision‐making process of doctors. It can reduce the subjective difference in the image classification process and assist dermatologists to analyze the skin lesion area. Firstly, the feature detection method based on the extension theory can effectively quantify the difference between different colour features. Then, an interpretable segmentation edge of the skin lesion is established by using the neutrosophic theory which can convert the image into the neutrosophic space. The edge of the skin lesion is captured by applying the Hierarchical Gaussian Mixture Model (HGMM) method. Next, the edge sequence is established by segmenting the edge, and the contour regularity, symmetry, and uniformity of the edge of the skin lesion are analyzed. Finally, the extracted multi‐feature sets are used for dermoscopy image classification. Experiments are carried out on real datasets, and the classification accuracy of four kernel functions is verified. The experimental results show that the authors’ method can effectively improve the classification accuracy of benign dermoscopy images and malignant dermoscopy images.

Interpreting ocean remote sensing images is still a challenge that is worth studying because they can carry valuable information for various important applications. Due to the absence of labeled datasets, unsupervised object-based image analysis (OBIA) methods provide an effective solution to understand remote sensing images with the advantage of grouping local similar pixels into a homogeneous area. However, ocean remote sensing images usually have the characteristics of large size, large background, and coexisting of large and small objects, which results in previous OBIA methods easily falling into the difficulty of accurately segmenting the large and small objects at the same time and the dilemma of time-consuming computation. To solve this problem, a novel multiscale and multisubgraph (MSMSG)-based image segmentation method is presented in this article. First, a coarse-to-fine superpixel generation method is designed to generate optimal superpixels, which can not only solve the problem of coexisting large objects and small objects but also the problem of manually setting the initial segmentation number. Second, the proposed background removal strategy helps to eliminate the trouble of large background areas in ocean remote sensing images. Third, a multisubgraph is constructed with the help of background removal. Finally, the MSMSG merging strategy is addressed to group all similar superpixels into the same cluster, which not only reduces the useless computation of nonadjacent superpixels but also avoids segmentation errors with the same scale. Experiments conducted on three different datasets show that the proposed segmentation method is high-performance and high-efficiency.

Aspect-based sentiment analysis aims to identify the sentiment polarity of a certain aspect in the context sentence. It only use the syntax dependency tree to construct graph convolution in previous sentiment analysis methods. But one aspect of sentiment sometimes only be determined by a few words, and relying entirely on the syntax tree may distract the model's attention. Also, due to the limitations of the corpus, the model is only able to learn limited knowledge. To address the above limitations, we propose a sentiment classification model based on aspect-dependent heterogeneous graph convolutional network (named ADHGCN). The model prunes the dependency tree, and then it can reduce the influence of featherweight information on the results. In addition, the model fuses multiple feature relationships between words by constructing a heterogeneous graph, and applies graph convolutional network (GCN) to seek meaningful representations for each node. It makes the model further integrate a variety of information on the original basis, and then it is no longer purely dependent on one relation. At the same time, this paper will introduce a common-sense knowledge base to participate in the construction of heterogeneous graphs, so that the model can learn knowledge beyond the corpus, thereby improving the accuracy of sentiment classification. Through experiments, we have demonstrated that our network performs better than others on five public datasets.

With the growing problem of forest fires, thermal infrared imaging technology is gradually applied to monitor and control forest fires. The segmentation of thermal infrared images is of great significance as an important part of this technology. This paper proposes an image segmentation model based on K-means clustering and variational (K-V model), which is used to alleviate the problem that the forest fire thermal infrared image is difficult to be segmented due to the presence of smoke masking, boundary blur of the fire area and regional dispersion of the fire area. Experiments are on a data set obtained by transforming the forest fire thermal infrared images collected on the Internet. This paper tests the running time and qualitative segmentation results of the proposed K-V model, and obtains convincing performance.

With the continuous improvement of people’s quality of life, air quality issues have become one of the topics of daily concern. How to achieve accurate predictions of air quality in a variety of complex situations is the key to the rapid response of local governments. This paper studies two problems: (1) how to predict the air quality of any monitoring station based on the existing weather and environmental data while considering the spatiotemporal correlation among monitoring stations and (2) how to maintain the accuracy and stability of the forecast even when the available data is severely insufficient. A prediction model combining Long Short‐Term Memory networks (LSTM) and Graph Attention (GAT) mechanism is proposed to solve the first problems. A metalearning algorithm for the prediction model is proposed to solve the second problem. LSTM is used to characterize the temporal correlation of historical data and GAT is used to characterize the spatial correlation among all the monitoring stations in the target city. In the case of insufficient training data, the proposed metalearning algorithm can be used to transfer knowledge from other cities with abundant training data. Through testing on public data sets, the proposed model has obvious advantages in accuracy compared with baseline models. Combining with the metalearning algorithm, it gives a much better performance in the case of insufficient training data.

The main task of medical image mining is to effectively analyze medical image data. Medical image classification algorithms have a high error rate near the threshold. To address the problem, the paper adopts a hybrid approach which combines computers algorithm and crowdsourcing system for image classification. A hybrid framework is proposed, which can achieve a higher accuracy significantly than only use classification algorithms. At the same time, it only processes the images that classification algorithms perform not well, so it has a lower monetary cost. In this framework, a range threshold is generated by using an efficient algorithm that assigns an image to a crowdsourcing or classification algorithm. To ensure the quality of crowdsourcing answers, this paper presents two worker models, Worker Quality Evaluation Model (WQEM)and Worker Performance Prediction Model(WPPM) respectively. Due to the lack of the crowdsourcing platform for processing medical information, medical image classification results are difficult to collect, so this paper proposed a crowdsourcing platform for medical image classification.

Low-dose computed tomography image denoising is a challenging task that has been studied by many researchers. Current denoising methods based on deep learning tend to produce a blur effect on the final results, especially at high noise levels, which are prone to over-smoothed edges and loss of details. In this paper, we propose a deep learning approach based on deep convolutional and edge constraints to mitigate these problems. Firstly, to avoid the loss of shallow layers details while obtaining semantically-richer features information, we use dilated convolution instead of standard convolution, and fusion feature maps of different levels to aggregate information from different receptive field. Secondly, in order to improve the network's ability to distinguish between noise and image content, we have designed an attention block to adaptively recalibrate the information relationship of the fusion feature maps. Finally, we incorporate edge prior knowledge into LDCT image denoising task, guiding the network to pay more attention on texture and structure information by edge constraints loss. Extensive experiments demonstrate that the proposed method achieves significant improvements over the state-of-the-art methods.

A new document retrieval method based on graph was proposed in this paper. Queries and documents are represented by graphs. The paper also proposes the concept of the document semantic unit in consideration of the overhead of graph computing. The size of semantic unit is used as the granularity for graph construction. This new method puts queries and documents in an unequal level instead of regarding them as equivalent entities which conventional IR system does. The paper further proposes the similarity calculating method of graphs based on general maximum common subgraph. The result of the experiment shows this method is able to yield better document retrieval results.

In this paper, we mainly study the problem of computing the k nearest neighbor over the encrypted data. We solve this question from two aspects. Firstly, we focus on the query efficiency. It is necessary to improve the user experience and save computing resources through reducing query time. We propose a k nearest neighbor query algorithm based on hierarchical Clustering. This algorithm uses the pruning rules to exclude most of the non-nearest neighbor data and improve query efficiency. Secondly, considered from data's security, we use the SSED algorithm which can safely compute distance between two encrypted data by using the properties of homomorphic encryption algorithm. Then the SSED algorithm is combined with the hierarchical clustering algorithm to realize the computation of k nearest neighbors over the ciphertext. The experiments show that the algorithm proposed in this paper has high query efficiency.

Uncertain data is inherent in various important applications and Top-[Formula: see text] query on uncertain data is an important query type for many applications. To tackle the performance issue of evaluating Top-[Formula: see text] query on uncertain data, an efficient optimization approach was proposed in this paper. This method can anticipate the tuples most likely to become Top-[Formula: see text] result based on dominant relationship analysis, greatly reducing the amount of data in query processing. When the database is updated, this method could determine whether the change affects the current query result, and help us to avoid unnecessary re-query. The experimental results prove the feasibility and effectiveness of this method.

BACKGROUND: Brain disorders are one of the top causes of human death. Generally, neurologists analyze brain medical images for diagnosis. In the image analysis field, corners are one of the most important features, which makes corner detection and matching studies essential. However, existing corner detection studies do not consider the domain information of brain. This leads to many useless corners and the loss of significant information. Regarding corner matching, the uncertainty and structure of brain are not employed in existing methods. Moreover, most corner matching studies are used for 3D image registration. They are inapplicable for 2D brain image diagnosis because of the different mechanisms. To address these problems, we propose a novel corner-based brain medical image classification method. Specifically, we automatically extract multilayer texture images (MTIs) which embody diagnostic information from neurologists. Moreover, we present a corner matching method utilizing the uncertainty and structure of brain medical images and a bipartite graph model. Finally, we propose a similarity calculation method for diagnosis. RESULTS: Brain CT and MRI image sets are utilized to evaluate the proposed method. First, classifiers are trained in N-fold cross-validation analysis to produce the best θ and K. Then independent brain image sets are tested to evaluate the classifiers. Moreover, the classifiers are also compared with advanced brain image classification studies. For the brain CT image set, the proposed classifier outperforms the comparison methods by at least 8% on accuracy and 2.4% on F1-score. Regarding the brain MRI image set, the proposed classifier is superior to the comparison methods by more than 7.3% on accuracy and 4.9% on F1-score. Results also demonstrate that the proposed method is robust to different intensity ranges of brain medical image. CONCLUSIONS: In this study, we develop a robust corner-based brain medical image classifier. Specifically, we propose a corner detection method utilizing the diagnostic information from neurologists and a corner matching method based on the uncertainty and structure of brain medical images. Additionally, we present a similarity calculation method for brain image classification. Experimental results on two brain image sets show the proposed corner-based brain medical image classifier outperforms the state-of-the-art studies.

Identification of ideal mid-sagittal line (iML) is important for image registration, brain segmentation, pathology detection and particularly for medical image classification. In this paper, iML extraction method based on scale invariant feature transform (SIFT) features is proposed for brain CT images. The method consists of an offline part and an online part. In the offline part, the iML feature points of training set is extracted by an auxiliary tool and an optimized matching template set is obtained by our feature fusion and filtering algorithms. In the online part, a matching point set is generated by matching SIFT features of test images to the offline template. Then the point set is refined by our pruning algorithm and iMLs of test images are fitted by the refined point set. Both real and simulated image data sets are used to verify the accuracy, robustness and execution efficiency of the algorithm. Experimental results show that, our method achieves good accuracy and efficiency in both real and simulation image sets, and performs better tolerance to rotation, noise, fuzzy and asymmetry in comparison with other existing algorithms.

many methods have been developed for corner detection and matching. However, these detection methods do not take the domain knowledge of brain medical images into account. They produce some useless corners and lose essential domain information. Moreover, existing corner matching methods do not consider the uncertainty and structure of brain medical images. And most of them are developed for 3D medical image registration, which are not applicable for 2D image classification. To address these problems, a corner detection method is firstly proposed based on hierarchical textures. Then, based on the uncertainty and structure of brain medical images, a corner matching method is developed to yield an initial and furthermore a maximum matched corner pair sequences. Finally, a similarity function is presented for classification. Experimental results show the proposed corner detection method outperforms the existing method and the classification results based on the maximum matched corner pair sequence are better than the state-of-the-art brain medical image classification methods.

In this work, an interactive visual system MICS is presented for large-scale brain CT image classification. Automatic feature extraction algorithms are added in MICS to improve system efficiency and classification accuracy. In visualization part, we designed an interactive feature extraction interface, enable users to extract and fine-tune image features according to specific requirements. In addition, all image features in database are visualized as dynamic charts in every phase of classification. These allow users to compare the current image with others in some specific feature and re-mark the possible misclassification. Finally, by series experiments and case studies, we verify the performance of the classification algorithm as well as the effectiveness and applicability of the visual design in MICS.

This paper proposed a Hadoop-based iterative sampling approximate aggregation query processing method. According to the user desire precision and the first sample data, we could compute the sample size to meet the user desired precision. In order to avoid the effects of data bias, this paper proposed a “layered sampling” method to ensure that the approximate aggregation result is statistically meaningful.

Medical images are one of the most important tools in doctors' diagnostic decision-making. It has been a research hotspot in medical big data that how to effectively represent medical images and find essential patterns hidden in them to assist doctors to achieve a better diagnosis. Several graph models have been developed to represent medical images. However, the unique structures of domain-specific images are not considered well to lose some essential information. Thus, aiming at brain CT images, we first construct a graph about the Topological Relations between Ventricles and Lesions (TRVL) and present the graph modeling process. Then we propose a method named Frequent Approximate Subgraph Mining based on Graph Edit Distance (FASMGED). This method uses an error-tolerant graph matching strategy that is accordant with ubiquitous noise in practice. Experimental results show that the graph modeling process is computationally scalable and FASMGED can find more significant patterns than current algorithms.

A number of brain computed tomography (CT) images stored in hospitals that contain valuable information should be shared to support computer-aided diagnosis systems. Finding the similar brain CT images from the brain CT image database can effectively help doctors diagnose based on the earlier cases. However, the similarity retrieval for brain CT images requires much higher accuracy than the general images. In this paper, a new model of uncertain location graph (ULG) is presented for brain CT image modeling and similarity retrieval. According to the characteristics of brain CT image, we propose a novel method to model brain CT image to ULG based on brain CT image texture. Then, a scheme for ULG similarity retrieval is introduced. Furthermore, an effective index structure is applied to reduce the searching time. Experimental results reveal that our method functions well on brain CT images similarity retrieval with higher accuracy and efficiency.