Hyperspectral facial imaging offers a promising avenue for analyzing cutaneous inflammation such as acne by capturing rich spectral-spatial information beyond conventional RGB modalities. However, the lack of publicly available pixellevel annotated datasets, limited histopathological validation, and the high computational burden of hyperspectral models hinder their clinical adoption. To address these challenges, we introduce MALSeg, the first benchmark dataset for hyperspectral acne lesion segmentation, containing 744 sub-images from 300 subjects across 121 spectral bands ($400-1000 ~\text{nm}$). Each image is densely annotated by dermatology experts and histopathologically validated, supporting reproducible algorithm development. To process high-dimensional spectral data efficiently, we apply a principal component-based tri-band selection strategy, mapping the most informative wavelengths into RGB space to reduce redundancy while preserving diagnostic features. Building on this dataset, we propose VML-UNet, a lightweight segmentation network that integrates a Vision Mamba-based state-space backbone for long-range context modeling with a lightweight UNet-style decoder for spatial detail extraction. A phased training strategy further enhances feature learning robustness. Extensive experiments on MALSeg demonstrate that VML-UNet achieves a mean Dice score of 90.86 %, outperforming classical UNet and spectral-driven baselines while enabling real-time inference. This work establishes a standardized foundation for hyperspectral dermatological research, bridging the gap between high-fidelity spectral imaging and practical clinical deployment in acne lesion analysis.

Existing deep learning methods for medical image segmentation and anatomical landmark detection often struggle with limited scalability and robustness, particularly in multi-task settings. To address these challenges, we propose MSL-Net, a lightweight and generalizable self-supervised framework for joint medical image segmentation and landmark detection. MSL-Net adopts an LR-ASPP-MobileNetV3 backbone for efficient feature extraction and introduces a Deeply Separable Task-Specific (DSTS) module to decouple task-specific representations, effectively mitigating interference between segmentation and landmark detection tasks. A novel self-supervised pretraining strategy based on random masking and reconstruction allows the model to learn rich spatial and temporal features from unlabeled image sequences. During fine-tuning, weak supervision is employed to generate pseudo-labels, enabling full-sequence training with minimal annotations. Experimental results on the EchoNet-Dynamic echocardiography dataset demonstrate strong performance, achieving a 93.54% Dice score and 0.803 OKS, with only 5.007M parameters and 0.246 GFLOPs. Cross-domain evaluations on the SDD2020 spine CT dataset further validate the model's generalization ability across imaging modalities and anatomical regions. These results highlight the potential of MSL-Net as an effective solution for real-time, multi-task medical image analysis with limited supervision.

Fetal heart rate (FHR) classification is essential for perinatal fetal health monitoring, but existing methods face challenges such as strong reliance on manual annotations, complex signal dynamics, and poor generalization across data domains. We propose TSCA-BYOL, a self-supervised representation learning framework with medical Knowledge for robust FHR signal classification. A Medical Constraint Augmentation (MCA) module generates complementary time- and frequency-domain views by injecting physiologically constrained Gaussian noise and applying band-limited spectral perturbations, preserving key pathological features. A Temporal-Spatial Convolutional Attention encoder (TSCA) combined with the BYOL framework learns invariant fetal cardiac signal representations from large-scale unlabeled data. The learned representations are transferred to a supervised classifier, achieving an AUC of 0.99 and 97% accuracy on the CTU-UHB benchmark, and 96% accuracy on a private home monitoring dataset, demonstrating strong robustness under domain shift and real-world noise conditions. TSCA-BYOL effectively integrates medical knowledge with self-supervised learning, reducing annotation dependency and providing an accurate, scalable solution for real-time fetal health monitoring.

Continuous cardiovascular monitoring is vital for managing heart disease. However, the clinical standard, Electrocardiography (ECG), is impractical for long-term and daily use. Ballistocardiography (BCG), which measures the body's mechanical forces, offers a non-invasive alternative, but reconstructing a high-fidelity ECG from it remains a challenge. To address this, we introduce BiM-Diff, a conditional diffusion model featuring a custom Bidirectional Mamba bottleneck. This architecture efficiently captures long-range, bidirectional dependencies in physiological signals with linear complexity, while a hybrid loss function and Classifier-Free Guidance (CFG) ensure spectral realism and strong conditional adherence. On a public dataset, BiM-Diff achieves state-of-the-art performance with a Pearson Correlation Coefficient of 0.984 and a Fréchet Distance of 0.085, clearly outperforming existing methods. Notably, this fidelity is achieved in only 25 diffusion steps, showing a substantial advantage in both accuracy and efficiency over standard diffusion models. By effectively combining diffusion models with statespace architectures, our work provides a robust solution for synthesizing precise ECG signals from BCG.

Optical Coherence Tomography (OCT) plays a vital role in the early detection and monitoring of tractional retinal lesions (TRL), providing high-resolution visualization of retinal structures. However, automated TRL diagnosis remains challenging due to complex lesion morphology, large low-entropy background regions, and the scarcity of high-quality labeled data. Existing Self-Supervised Learning (SSL) approaches often treat all image patches equally, making them sensitive to background noise and limiting their ability to capture fine-grained lesion features. To address these issues, we propose Clustering Hetero-geneous Masked Image Modeling (CH-MIM), a novel SSL frame-work tailored for OCT-based TRL analysis. Our method lever-ages a large-scale clinical dataset containing 11,861 OCT scans collected over five years, including 3,950 expert-annotated images across six TRL severity levels (TO- T5). CH - MIM introduces a Weighted Feature Space Clustering (WFSC) module to selectively mask high-entropy regions, effectively filtering out irrelevant background information. A heterogeneous progressive masking strategy combines binary, Gaussian, and Poisson noise masks to provide diverse, informative reconstruction tasks. Furthermore, a Consistency Regularization Module (CRM) enforces stable predictions across masking branches, improving representation robustness and transferability to downstream classification. Ex-tensive experiments demonstrate that CH - MIM achieves a top-l accuracy of 97.7% and top-5 accuracy of 99.8%, surpassing state-of-the-art supervised and self-supervised baselines. These results highlight the potential of CH - MIM as an effective pretraining strategy for automated TRL screening and its applicability to broader OCT-based retinal disease diagnosis.

The analysis of Cardiotocography (CTG) signals is often hindered by challenges such as limited data availability and label imbalance, which can undermine the performance of deep learning models. To address these issues, we present CTGDiff, a novel conditional diffusion model designed for generating synthetic Fetal Heart Rate (FHR) and Uterine Contraction (UC) signals. CTGDiff leverages both Phase-Rectified Signal Averaging (PRSA) spectrograms and UC as conditioning inputs for FHR, and integrates time encoding, condition generation from PRSA features, and residual blocks with dilated convolutions to capture both temporal dynamics and long-range dependencies. Extensive experiments, both qualitative and quantitative, demonstrate the model’s ability to synthesize high-quality CTG signals. In comparison with GANs and image-based diffusion models, CTGDiff achieves superior signal fidelity and distribution similarity for FHR, as indicated by metrics such as a 0.004 maximum mean deviation (MMD), 0.646 percent root mean square difference (PRD), 3.951 relative entropy (RE), and 0.291 Frechet distance (FD). Expert evaluations confirm that the model can generate both normal and abnormal CTG signals with high accuracy, conditioned on specific input data. These results underscore the potential of diffusion models for a wide range of applications in biomedical time series analysis, including signal synthesis, imputation, and noise reduction.

Accurate segmentation and landmark detection are essential for medical image analysis. However, existing methods often struggle with generalization and robustness, particularly under sparse annotation conditions. This study introduces SSP-MAEFNet, a novel self-supervised framework designed to address these limitations. The model integrates a lightweight backbone network, LR-ASPP-MobileNetV3, for efficient feature extraction, along with a Deep Separable Task-Specific (DSTS) module that mitigates feature entanglement between segmentation and landmark detection tasks. By leveraging a self-supervised random masking and reconstruction strategy, the framework effectively captures temporal and spatial patterns, enhancing its generalization across datasets with minimal supervision. Extensive experiments on the EchoNet-Dynamic dataset demonstrate that SSP-MAEFNet achieves state-of-the-art performance, with a Dice score of 93.54% and an OKS of 81.30%, outperforming existing models in both tasks. Its robust performance under sparse annotation highlights its potential for clinical integration, especially in resource-constrained environments.

Fetal heart rate (FHR) is a biomedical time series essential for fetal monitoring and clinical assessment. A major challenge in automated FHR analysis lies in accurately segmenting acceleration, deceleration, and baseline patterns while deriving a stable and clinically meaningful baseline. This study proposes MSEU-Net, a multi-scale segmentation network with an encoder-decoder architecture designed to capture both local variations and global trends in FHR signals. The model performs fine-grained segmentation of FHR patterns and applies a post-processing strategy using long- and short-term median filtering to estimate a continuous baseline. Experiments conducted on the CULF-DB dataset demonstrate that MSEU-Net achieves a baseline root mean square deviation (BL.R) of 3.20 bpm and consistently outperforms traditional methods in both segmentation and baseline estimation metrics. Ablation studies further validate the effectiveness of multi-scale feature extraction and loss function design. These results indicate that MSEU-Net provides a reliable foundation for automated FHR interpretation and supports improved clinical decision-making in fetal health monitoring.

In the age of digital media, secure image transmission has become a key issue. To address this, we theoretically derived a two-dimensional chaotic model, combining a Logistic graph and sine map (2D-LGSM), achieving a Lyapunov index of 89.8558, indicating strong chaotic behavior. This model generates chaotic sequences for image encryption. Bayesian optimization enhances reinforcement learning in exploration strategies, while Lagrange optimization aids in histogram matching. The secret image is embedded into the frequency domain of the cover image, with information extracted using a majority voting method. With three BPP embedding, the PSNR reaches 53.02, SSIM is 0.9993, and the key space remains 2512. The NPCR and UACI values are 99.69% and 33.46%, respectively, demonstrating high sensitivity. Compared to recent schemes, it shows superior steganographic capacity, quality, and security performance.

Underwater object detection (UOD) is crucial for monitoring marine ecosystems, underwater robotics, environmental protection, and autonomous underwater vehicles (AUVs). Despite progress, many models struggle under real-world conditions due to poor visibility, dynamic lighting, and domain shifts. Traditional methods like Faster R-CNN are computationally expensive, while YOLO-based models suffer in challenging underwater scenarios. The scarcity of large-scale annotated datasets further limits model generalization. To address these challenges, we introduce UOD-SZTU-2025, a new dataset of 3,133 high-quality underwater images, sourced primarily from video platforms. The dataset is used in EFCWM (Enhanced Feature Correction and Weighting Module) to extract and refine a feature material library for detection targets. We propose EFCWM-Mamba-YOLO, a lightweight, real-time detection model designed to enhance feature representation and adapt to diverse underwater environments. The EFCWM module incorporates domain adaptation for improved robustness. Additionally, a two-stage training strategy first trains on a source domain and fine-tunes with limited target domain samples to enhance generalization. Experiments show our approach surpasses existing lightweight UOD models in accuracy, real-time performance, and robustness. Our dataset, model, and benchmark establish a strong foundation for future UOD research. The dataset for EFCWM-Mamba-YOLO is available at https://github.com/wojiaosun/UOD-SZTU-2025.

Accurate analysis of Fetal Heart Rate (FHR) signal is often impeded by challenges such as data scarcity and label imbalance, which affect the reliability and robustness of deep learning models. To address these challenges, this study introduces FHRDiff, a novel diffusion model conditioned on Phase-Rectified Signal Averaging (PRSA) spectrograms for the creation of synthetic FHR signals. Our model integrates time encoding, condition generation from PRSA spectrograms, and residual blocks with dilated convolutions to effectively manage temporal dynamics and long-range dependencies. Extensive qualitative and quantitative experiments on FHR signal synthesis demonstrate the feasibility and effectiveness of FHRDiff. Compared to Generative Adversarial Networks (GANs) and image-based diffusion model, our method achieves the highest signal fidelity and distribution similarity, with key measures including 0.067 maximum mean deviation (MMD), 0.492 percent root mean square difference (PRD), 1.763 relative entropy (RE), and 0.160 Frechet distance (FD). Expert validation confirms the model’s capacity to accurately generate data for normal and abnormal FHR signals based on the paired condition. In addition, an ablation study was conducted to highlight that the spectrogram paired condition can guide the diffusion model to produce synthetics FHR signals with greater diversity compared to unconditional models. The results emphasize diffusion models’ potential for broad application in biomedical time series analysis, such as generation, imputation and noise removal.

The accurate diagnosis of multiple thoracic diseases through chest X-ray (CXR) images is a challenging yet crucial task in the medical field. Deep learning approaches have shown promise in assisting clinicians in this endeavor. In this paper, we introduce CvTGNet, a novel framework that leverages Convolutional Vision Transformer (CvT) and Graph Convolutional Network (GCN) to enhance CXR diagnosis. CvTGNet is designed to harness the super generalization capability of CvT, a hybrid architecture that combines Convolutional Neural Networks (CNNs) and Transformers. We employ GCN to explore the co-occurrence relationships among thoracic diseases, allowing the model to gain insights into intricate pathological connections that might be overlooked by traditional methods. This information guides the CvT model in making more accurate multi-label pathological classifications. We conducted extensive experiments on two prominent CXR datasets, ChestX-Ray14 and CheXpert, to evaluate the performance of CvTGNet. The results demonstrate that our proposed method consistently outperforms other state-of-the-art approaches in terms of Area Under the Receiver Operating Characteristic Curve (AUC-ROC) scores for multilabel pathological classification. We also conducted ablation experiments to understand the contribution of each component of CvTGNet. Overall, our CvTGNet framework presents a significant advancement in CXR diagnosis. By combining the strengths of CvT and CGN, we achieve remarkable accuracy in detecting multiple thoracic diseases from CXR images. This approach holds promise in enhancing medical diagnosis, enabling clinicians to make more informed decisions and improving patient outcomes.

In Fetal Heart Rate (FHR) analysis for intrauterine growth restriction (IUGR), accurate baseline determination is essential for effective monitoring and intervention. MSEU-Net, a multi-scale deep learning framework, significantly advances this effort by offering enhanced accuracy in baseline calculations, leveraging convolution blocks and a unique multi-scale extraction module. This innovative approach promises to improve prenatal care by enabling more accurate assessments of fetal well-being.

UAV gesture recognition, a novel human-computer interaction form, offers an intuitive approach to controlling UAVs in various environments. However, there is a lack of comprehensive datasets for AI-powered UAV gesture recognition. This paper contributes in several ways: (i) We introduce MD-UHGRD, a unique UAV static gesture dataset with 20, 000 images and annotations, collected from a diverse group of participants in different environmental conditions. This dataset is expected to bridge a significant gap in UAV gesture recognition algorithms. (ii) We propose SA-YOLO, a multifunctional UAV gesture recognition method that not only enables gesture recognition but also includes face and pedestrian tracking, optimizing UAV control in complex scenarios. SA-YOLO incorporates the Spatial Asymptotic Feature Pyramid Network (SAFPN), Scale Pyramid Pooling with Cross Stage Partial Networks Convolution (SPPCSPC), and Space-to-Depth Convolution (SPD-Conv). (iii) Extensive evaluation of SAYOLO on MD-UHGRD establishes it as a benchmark in this domain. Our method demonstrates high accuracy, processing speed, and a compact model size, achieving a 93.2% mean Average Precision (mAP) with 10.3 million parameters and 48 frames per second (FPS). Among competing models, SA-YOLO not only achieves the highest mAP but also maintains a balance in model size and FPS. The database and code are available at: https://github.com/ijcnn2024/SA-YOLO.

Ensuring fetal health during pregnancy is critically dependent on precise Fetal Heart Rate (FHR) monitoring. A major challenge in this area is the limited availability of labeled FHR data, which poses a barrier to developing reliable automated analysis systems. To address this gap, our study introduces CTGGAN, a novel method employing Generative Adversarial Networks (GANs) to create synthetic, high-quality FHR signals. Specifically, CTGGAN integrates self-attention and residual modules within a Conditional GAN framework, fine-tuned to replicate the complex patterns characteristic of FHR data accurately. A notable feature of CTGGAN is its effective loss function, which combines Wasserstein distance with a gradient penalty to ensure training stability and enhance the authenticity of the generated signals. In performance metrics, Our method demonstrates the highest signal fidelity and distribution similarity, across five key measures: 0.215 maximum mean deviation (MMD), 0.012 sliced Wasserstein distance (SWD), 4.821 percent root mean square difference (PRD), 5.621 relative entropy (RE), and 0.614 Frechet distance (FD). This advancement in generating realistic FHR data with CTGGAN addresses critical issues like data insufficiency and class imbalance, thus advancing the field of prenatal healthcare technology. The code for CTGGAN is available at https://github.com/ijcnn2024/CTGGAN.

Cardiovascular diseases (CVDs) are the leading cause of global morbidity and mortality, necessitating the precise and continuous monitoring of blood pressure for proactive management. Our study presents the AMRUNet: a novel network designed exclusively for PPG-only, noninvasive, cuff-less blood pressure estimation. The network innovates upon the U-Net architecture, integrating a MultiRes Block for detailed multi-scale feature fusion and a residual block to mitigate the issue of vanishing gradients. An attention mechanism is further employed to selectively enhance salient features within the PPG signal. Our PPG-only AMRUNet demonstrates exceptional performance in translating PPG data into accurate ABP waveforms, achieving mean absolute errors (MAE) that comply with the standards of both the British Hypertension Society (BHS) and the Association for the Advancement of Medical Instrumentation (AAMI). Our method demonstrates highest MAE for both systolic blood pressure (SBP) and diastolic blood pressure (DBP), achieving a 2.85 MAE for SBP and 1.79 MAE for SBP among competing models. The model’s proficiency in precisely estimating systolic and diastolic blood pressure, along with its ability to reconstruct continuous ABP waveforms, contributes to reliable and trustworthy medical decision-making systems. The code for AMRUNet can be accessible at https://github.com/ijcnn2024/AMRUNet.

The advancement of machine vision systems necessitates efficient and accurate signal reconstruction methods to enhance real-time perception and decision-making capabilities. This paper introduces a Generalized Backtracking Regularization Adaptive Matching Pursuit (GBRAMP) algorithm, designed to reconstruct signals within machine vision systems using compressed sensing techniques. The GBRAMP algorithm improves upon existing methods by incorporating regularization for enhanced atom selection and a backtracking approach to accurately estimate sparsity, addressing the limitations of traditional convex optimization, greedy, and Bayesian reconstruction algorithms. The paper provides a comparative analysis of the GBRAMP algorithm against other prominent reconstruction techniques. Experimental results validate the GBRAMP algorithm's improved performance in terms of both reconstruction accuracy and computational speed, making it a competitive solution for the next generation of machine vision systems.

Medical image segmentation is a critical component in a variety of clinical applications, facilitating accurate diagnosis and treatment planning. The Segment Anything Model (SAM), a deep learning architecture, has emerged as a promising solution to the challenges inherent in medical image segmentation. SAM’s superior zero-shot capability allows it to generalize effectively, even in the absence of task-specific segmentation samples. This unique characteristic broadens its application potential across various medical image modalities. This paper provides an in-depth review of SAM, focusing on its application in medical image segmentation. The review discusses the advantages of deep learning image segmentation over traditional methods, emphasizing the superior accuracy, efficiency, and automation that deep learning models offer. The paper also highlights the applications of SAM across various medical imaging modalities, demonstrating its versatility and adaptability. A taxonomy of SAM approaches in medical image segmentation is presented, categorizing them based on modality, dimension, organ, dataset, prompt, and performance. Despite the promising results of SAM, challenges remain in the field of medical image segmentation. The paper identifies these challenges and suggests potential directions for future research. In conclusion, this review aims to provide a comprehensive understanding of SAM and its potential to revolutionize medical image analysis and contribute to advancements in healthcare.

Obstetric ultrasound examination of physiological parameters has been mainly used to estimate the fetal weight during pregnancy and baby weight before labour to monitor fetal growth and reduce prenatal morbidity and mortality. However, the problem is that ultrasound estimation of fetal weight is subject to populations’ difference, strict operating requirements for sonographers, and poor access to ultrasound in low-resource areas. Inaccurate estimations may lead to negative perinatal outcomes. We consider that machine learning can provide an accurate estimation for obstetricians alongside traditional clinical practices, as well as an efficient and effective support tool for pregnant women for self-monitoring. We present a robust methodology using a data set comprising 4,212 intrapartum recordings. The cubic spline function is used to fit the curves of several key characteristics that are extracted from ultrasound reports. A number of simple and powerful machine learning algorithms are trained, and their performance is evaluated with real test data. We also propose a novel evaluation performance index called the intersectionover-union (loU) for our study. The results are encouraging using an ensemble model consisting of Random Forest, XGBoost, and LightGBM algorithms. The experimental results show an loU of 0.64 between predicted range of fetal weight at any gestational age from the ensemble model and that from ultrasound. Comparing with the ultrasound method, the estimation accuracy is improved by 12%, and the mean relative error is reduced by 3%.

Improving the accuracy and consistency of interpretation results for foetal monitoring has been an active research direction in both obstetrics and gynaecology. In this paper, we have developed computer-aided analysis systems for use both in hospitals and at home that incorporate automatic scoring functions to evaluate the foetal conditions in the cavity of the uterus. These systems can analyse any segment of data in a foetal monitoring record. Our novel systems can accurately identify the CTG patterns, such as FHR baseline, foetal movements, uterine contractions, accelerations and type of decelerations, thus making the interpretation results more accurate. There are two modes of scoring: automatic and manual, and the system consists of a number of popular scoring methods, including the Kreb's, Fischer, and improved Fischer scoring methods and the ACOG three-tier classification methods. According to clinical tests in hospitals, the systems have comparable accuracy to obstetricians' interpretations. Computerised interpretation thus provides a supplement to traditional analysis that could help obstetricians function more effectively.

Adverse incidents in the privacy of patients' medical records can result in multiple negative impacts. Effective mechanisms are needed to communicate the lessons from the incidents into the Information Security Management Systems (ISMS) so as to prevent similar incidents. The Generic Security Template (G.S.T.) has been developed to enhance current mechanism and has demonstrated significant benefits in communicating the lessons compared to the more conventional use of text-based incident reports. This paper extends the work to evaluate the G.S.T. in healthcare. A case study with healthcare professionals working in a China healthcare organization shows that, the G.S.T. can enhance the current mechanism in communicating the lessons with the ISMS.