Biomedical Signal Processing and Control最新文献

{"title":"RHA-Net: A Residual Hybrid Attention Network for low-dose CT denoising","authors":"Yaoyao Ma ,&nbsp;Jing Wang ,&nbsp;Chao Xu ,&nbsp;Yuling Huang ,&nbsp;Minghang Chu ,&nbsp;Zhiwei Fan ,&nbsp;Yishen Xu ,&nbsp;Di Wu","doi":"10.1016/j.bspc.2025.107888","DOIUrl":"10.1016/j.bspc.2025.107888","url":null,"abstract":"<div><div>Low-Dose Computed Tomography (LDCT), designed to reduce the potential risks associated with excessive X-ray radiation doses, is now attracting increasing interest in medical practice. However, it often contains substantial noise and artifacts that could impede clinical diagnosis. In this paper, we propose a Residual Hybrid Attention Network, referred to as RHA-Net, to enhance the denoising effect for LDCT images. The LDCT image is initially processed by the Dualistic Parallel Convolution module within the network. The resulting feature maps are then fed into the Residual Hybrid Attention (RHA) group, which primarily consists of the Multi-scale Feature Extraction (MSFE) module. This module preserves the original high-resolution features while generating complementary features at multiple scales. Within the MSFE module, the designed Global-Local Feature Extraction Block (GLB) and a new Adaptive Feature Fusion (AFF) module work together to effectively extract both global and local information. These components integrate multiple feature maps while adaptively recalibrating the weights of these feature maps across channel dimensions. Ultimately, the module outputs the denoised image. In the objective evaluation metrics of the Mayo test set, RHA-Net achieves average PSNR, SSIM, FSIM, and RMSE scores of 33.8232, 0.9339, 0.7684, and 5.3150, respectively, outperforming eight SOTA denoising algorithms. Additionally, in the out-of-distribution test on the head LDCT images from the LDCT-and-Projection-data dataset, RHA-Net achieves high scores across all four evaluation metrics, with PSNR improving by approximately 12.8% compared to Noise2Sim. Furthermore, RHA-Net demonstrates excellent denoising performance on real LDCT images from the Piglet dataset, fully showcasing its robust generalization capabilities.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"109 ","pages":"Article 107888"},"PeriodicalIF":4.9,"publicationDate":"2025-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143936308","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enhanced coronary artery segmentation and stenosis detection: Leveraging novel deep learning techniques","authors":"Abu Jaffor Morshedul Abedin ,&nbsp;Rusab Sarmun ,&nbsp;Adam Mushtak ,&nbsp;Mohamed Sultan Bin Mohamed Ali ,&nbsp;Anwarul Hasan ,&nbsp;Ponnuthurai Nagaratnam Suganthan ,&nbsp;Muhammad E.H. Chowdhury","doi":"10.1016/j.bspc.2025.108023","DOIUrl":"10.1016/j.bspc.2025.108023","url":null,"abstract":"<div><div>Coronary artery disease (CAD) is a significant global health concern, emphasizing the need for reliable and automated diagnostic solutions. This study proposes a novel deep learning framework aimed at improving both full artery segmentation and stenosis localization by incorporating Self-Organizing Neural Networks (Self-ONN). The DenseSelfU-Net model leverages DenseNet121 as an encoder and a Self-ONN enhanced decoder within a U-Net based architecture to achieve robust feature extraction and precise full artery segmentation, achieving an IoU of 82.52% and a Dice score of 90.35% on the ARCADE Challenge dataset. For stenosis localization, Self-ONN is integrated into key components of the Multi-Scale Attention Network (MA-Net), which includes the Multi-Scale Fusion Attention Block (MFAB) and the Position-wise Attention Block (PAB), capturing complex vascular patterns through both local and global dependencies and resulting in the DenseSelfMA-Net model. The DenseSelfMA-Net achieves Dice scores of 60.59% and 60.36% and IoU scores of 46.09% and 45.36% for the MFAB and PAB configurations, respectively on the ARCADE challenge dataset. These results demonstrate the effectiveness of Self-ONN in enhancing diagnostic precision and facilitating early CAD diagnosis, with promising implications for clinical practice.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"109 ","pages":"Article 108023"},"PeriodicalIF":4.9,"publicationDate":"2025-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143936294","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A prior embedding-driven architecture for long distance blind iris recognition","authors":"Qi Xiong ,&nbsp;Xinman Zhang ,&nbsp;Jun Shen","doi":"10.1016/j.bspc.2025.108048","DOIUrl":"10.1016/j.bspc.2025.108048","url":null,"abstract":"<div><div>Blind iris images, caused by unknown degradation during the process of iris recognition at long distances, often lead to decreased iris recognition rates. Currently, limited literature addresses this issue. To tackle this challenge, we propose a prior embedding-driven architecture for long-distance blind iris recognition. Our approach introduces a blind iris image restoration network named Iris-PPRGAN. To effectively restore the texture of blind iris images, Iris-PPRGAN incorporates a Generative Adversarial Network (GAN) as a Prior Decoder and a Deep Neural Network (DNN) as the encoder. To enhance iris feature extraction, we also developed a robust iris classifier by modifying the bottleneck module of InsightFace, referred to as Insight-Iris. Initially, a low-quality blind iris image is restored using Iris-PPRGAN, and the restored image is subsequently recognized through Insight-Iris. Experimental results on the public CASIA-Iris-distance dataset demonstrate that our method outperforms state-of-the-art blind iris restoration techniques. Specifically, the recognition rate for long-distance blind iris images improves from 80.77 % (without restoration) to 90.38 % after applying our method, reflecting an approximate ten-percentage-point improvement. These results, validated both quantitatively and qualitatively, underscore the effectiveness of our approach in addressing the challenges of long-distance blind iris recognition.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"109 ","pages":"Article 108048"},"PeriodicalIF":4.9,"publicationDate":"2025-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143936295","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DIO-REGNET: Macular edema detection using Dingo optimized deep Reg network","authors":"Gulfishan Firdose Ahmed ,&nbsp;Piyush Kumar Shukla ,&nbsp;Chukka Santhaiah ,&nbsp;Raju Barskar ,&nbsp;Noha Alduaiji ,&nbsp;Balamurali Pydi ,&nbsp;A. Chandrasekar","doi":"10.1016/j.bspc.2025.107941","DOIUrl":"10.1016/j.bspc.2025.107941","url":null,"abstract":"<div><div>Macular edema (ME) is a primary cause of blindness and loss of vision in people with visual retinal disorders. Deep learning (DL) algorithms benefit significantly from extensive and diverse datasets during training, but obtaining a sufficient amount of labeled data for macular edema is challenging. An insufficient dataset may result in overfitting, reducing the network’s capability to generalize the diverse cases. An optical coherence tomography (OCT) framework is utilized to solve the problem of diabetic macular edema (DME). Due to the complex nature of this condition and the saturation of healthcare in affluent nations, it is among the main factors that induce blindness. In this paper, a novel DIO-RegNet was introduced for the early recognition of the ME using DL techniques. The input OCT images are pre-processed by a Gaussian adaptive bilateral filter to enhance the image quality. The noise-free images are fed to the Modified DeepLabV3 + to segment the Macular area in retinal images. Then, the segmented Macular region is fed into deep learning-based RegNet for extracting the structural feature. Finally, the Dingo Optimization (DIO) algorithm is applied for the feature selection and classify the cases of macular edema. The proposed DIO-RegNet achieves a detection accuracy of 99.44 % for macular edema. Compared to Dense Net, Alex Net, and ResNet, RegNet achieves an accuracy rate of 96.72 %, 92.89 %, and 97.11 %, respectively. The DIO-RegNet improves overall accuracy by 2.44 %, 5.04 %, and 4.34 % over CNN, faster R-CNN, and VGG-16 CNN, respectively.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"109 ","pages":"Article 107941"},"PeriodicalIF":4.9,"publicationDate":"2025-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143936306","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Expert system supporting automatic risk classification and management in idiopathic membranous nephropathy based on rule sets and machine learning","authors":"Dawid Pawuś ,&nbsp;Szczepan Paszkiel ,&nbsp;Tomasz Porażko","doi":"10.1016/j.bspc.2025.107989","DOIUrl":"10.1016/j.bspc.2025.107989","url":null,"abstract":"<div><div>The diagnosis and management of idiopathic membranous nephropathy (IMN) is a complex clinical challenge due to the disease’s unpredictable progression and the varying responses to treatment. Traditional methods of risk stratification and treatment planning often rely on manual assessments, which can lead to inconsistent decision-making and suboptimal patient outcomes. To address this issue, we propose an expert system that leverages machine learning (ML) and artificial intelligence (AI) models and a knowledge-based approach to automate risk classification and treatment recommendations for IMN patients. This system aims to standardize and automate clinical decision-making, improve diagnostic accuracy, and enhance patient care through data-driven insights.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"109 ","pages":"Article 107989"},"PeriodicalIF":4.9,"publicationDate":"2025-05-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143931580","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A feasibility study of leveraging intermuscular coherence in EMG-driven neuromusculoskeletal modeling to improve muscle moment estimation","authors":"Emilie Mathieu ,&nbsp;Sylvain Crémoux ,&nbsp;David Gasq ,&nbsp;Philippe Pudlo ,&nbsp;David Amarantini","doi":"10.1016/j.bspc.2025.108004","DOIUrl":"10.1016/j.bspc.2025.108004","url":null,"abstract":"<div><div>Current musculoskeletal models often oversimplify the neural strategies underlying muscle activation, potentially leading to unsatisfactory estimates of muscle forces. Numerous studies in motor control have established that the central nervous system synchronizes muscle activation by sending a common drive to synergistic muscles, measurable through intermuscular coherence – the frequency correlation between two EMG signals. As interest grows in understanding how muscles synchronize during movement coordination, leveraging intermuscular coherence into musculoskeletal models represents an innovative approach. This could enhance the accuracy of muscle effort estimation and introduce a physiologically meaningful component of motor control. In this study, we introduce a new method that decomposes EMG signals into common and independent components, informed by intermuscular coherence, and integrates them into an EMG-driven model to estimate muscle moments. Using data from twenty-four healthy subjects performing horizontal upper limb extensions, we estimated moments of the four main muscles actuating the elbow and compared these estimations with those from a traditional EMG-driven model informed by full-wave rectified signal envelopes. Our results demonstrate that incorporating intermuscular coherence significantly enhanced kinetic data tracking and improved the robustness of muscle moment estimations against variations in model parameters, addressing a major limitation of traditional EMG-driven models. Furthermore, antagonist muscle moments were more accurately represented, resulting in more realistic co-contraction index values.</div><div>By integrating neural control strategies via intermuscular coherence into musculoskeletal models, the proposed approach offers a more accurate representation of muscle coordination. We recommend that future neuromusculoskeletal models incorporate intermuscular coherence to improve physiological realism of muscle effort estimations.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"109 ","pages":"Article 108004"},"PeriodicalIF":4.9,"publicationDate":"2025-05-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143931579","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Automating vessel segmentation in the heart and brain: A trend to develop multi-modality and label-efficient deep learning techniques","authors":"Nazik Elsayed ,&nbsp;Yousuf Babiker M. Osman ,&nbsp;Cheng Li ,&nbsp;Jiarun Liu ,&nbsp;Weixin Si ,&nbsp;Jiong Zhang ,&nbsp;Shanshan Wang","doi":"10.1016/j.bspc.2025.108028","DOIUrl":"10.1016/j.bspc.2025.108028","url":null,"abstract":"<div><div>Cardio-cerebrovascular diseases remain the leading cause of mortality worldwide, making accurate blood vessel segmentation essential for both scientific research and clinical applications. However, segmenting cardio-cerebrovascular structures from medical images is highly challenging due to factors such as thin or blurred vascular shapes, imbalanced vessel-to-background pixel distribution, and interference from imaging artifacts. These difficulties render manual or semi-manual segmentation methods time-consuming, labor-intensive, and prone to inter-observer variability. Consequently, there is an increasing demand for automated segmentation algorithms. This paper presents the first comprehensive survey of deep learning techniques for cardio-cerebrovascular segmentation, covering supervised, semi-supervised, and unsupervised approaches for both cardiac and cerebral vasculature. We review state-of-the-art methods, including U-Net, Generative Adversarial Networks (GANs), Graph Convolutional Networks (GCNs), transformer models, diffusion models such as Denoising Diffusion Probabilistic Models (DDPM), foundation models like Segment Anything Model (SAM) and the SAM-VMNet models, as well as hybrid approaches combining multiple models with effective fusion techniques. We discuss the strengths and limitations of these methods, emphasizing their clinical applicability. Our analysis identifies key challenges, including the reliance on annotated data and the limitations of single-modality approaches, which fail to fully leverage the rich information available from multi-modal imaging sources. We highlight the importance of label-efficient, multi-modal deep learning as a promising direction for improving segmentation accuracy and robustness. This survey provides valuable insights for researchers and clinicians, aiming to guide the development of next-generation tools for more accurate diagnoses and personalized treatment strategies for cardio-cerebrovascular diseases.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"109 ","pages":"Article 108028"},"PeriodicalIF":4.9,"publicationDate":"2025-05-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143931578","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Deep representation learning for Nuclear Magnetic Resonance spectral clustering","authors":"Wentao Hu ,&nbsp;Zichen Shao ,&nbsp;Yanchao Xu ,&nbsp;Linzhu Yu ,&nbsp;Zhihao Chang","doi":"10.1016/j.bspc.2025.107892","DOIUrl":"10.1016/j.bspc.2025.107892","url":null,"abstract":"<div><div>Nuclear Magnetic Resonance (NMR) spectroscopy is essential for molecular structure elucidation, drug discovery, and biomedical research, as it enables the identification of spectral patterns and compound similarities. However, traditional clustering methods struggle with robustness against chemical shift variations, peak intensity fluctuations, and noise, limiting their effectiveness in complex NMR datasets. To address these challenges, we propose a novel framework that combines an attention mechanism with a bidirectional long short-term memory autoencoder. This approach extracts robust, low-dimensional representations of NMR spectra by integrating adaptive local feature extraction with deep representation learning. Our method enables effective clustering across diverse spectral regions while preserving critical chemical information. Comprehensive experiments on both synthetic and real-world datasets demonstrate that this framework significantly outperforms conventional techniques in clustering accuracy and representation quality. These findings highlight its practical utility for enhancing NMR spectral analysis and molecular characterization.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"109 ","pages":"Article 107892"},"PeriodicalIF":4.9,"publicationDate":"2025-05-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143928529","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A convolutional neural network for automatic detection of sleep-breathing events using single-channel ECG signals","authors":"Hao Dong ,&nbsp;Haitao Wu ,&nbsp;Guan Yang ,&nbsp;Junming Zhang","doi":"10.1016/j.bspc.2025.107943","DOIUrl":"10.1016/j.bspc.2025.107943","url":null,"abstract":"<div><div>Sleep apnea-hypopnea syndrome (SAHS) is a common sleep-breathing disorder, and the development of an automated method for detecting hypopnea and apnea is crucial for early prevention and treatment. Clinicians vary in their treatment approaches for sleep hypopnea and sleep apnea, often regarding sleep apnea as more severe. However, most previous methods for detecting sleep-breathing events were binary classifications that merged hypopnea and apnea into a single category, which are inadequate for auxiliary diagnosis and accurate sleep quality monitoring today. In this study, we proposed a convolutional neural network (CNN) model called TriGNet, which aims to detect hypopnea and apnea events. TriGNet utilizes single-channel electrocardiogram (ECG) signals from the MIT-BIH polysomnography dataset to learn and differentiate between normal, hypopnea, and apnea events. We designed a 2D feature extraction component in the model, which processes 1D ECG signal data as 2D to obtain additional feature information. In addition, this study proposed a dynamic data regulation mechanism for capturing subtle variations among the three categories of ECG signals. The TriGNet model can directly learn features from ECG signals and effectively classify sleep-breathing events. The proposed method achieved an accuracy of 94.08%, a macro F1 score of 93.02%, and a Cohen’s kappa coefficient of 89.91% on the test set. Experimental results show that the TriGNet model proposed in this study achieves state-of-the-art (SOTA) performance. You can find our source codes at <span><span>https://github.com/MMMaoTS/TriGNet</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"109 ","pages":"Article 107943"},"PeriodicalIF":4.9,"publicationDate":"2025-05-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143928674","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"New accurate deep learning model for Diabetic Retinopathy detection utilizing sequential pre-processing and transfer learning","authors":"Caner Sen ,&nbsp;Selim Doganay ,&nbsp;Giyasettin Ozcan","doi":"10.1016/j.bspc.2025.108060","DOIUrl":"10.1016/j.bspc.2025.108060","url":null,"abstract":"<div><div>This study considers the detection of Diabetic Retinopathy (DR) using deep learning. DR affects 80% of diabetic patients worldwide and is the second leading cause of blindness. Many studies have shown that early diagnosis and treatment are critical to prevent disease progression. The contribution of this study is the development of an accurate DR detection algorithm and corresponding model, where hemorrhages were brought to a more apparent form by an efficient processing pipeline. To handle limited DR data resources and to make the visibility of bleeding in the eye more apparent, we have developed an efficient deep learning model by combining data augmentation, pre-processing, transfer learning, and adaptation of a compatible CNN. The employed dataset comprises fundus images of individuals, which are categorized into five stages of DR. For evaluation, comprehensive ablation studies are conducted on the model. Next, the developed model is evaluated against state-of-the-art algorithms and demonstrates promising results in key metrics. Particularly, the model yields 96.95% accuracy and introduces a false negative rate below 1%. Efficient metrics of the study minimize the risk of missed diagnoses and reduce the likelihood of severe vision loss in diabetic patients. Therefore, our model has the potential to contribute to clinical patient care.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"109 ","pages":"Article 108060"},"PeriodicalIF":4.9,"publicationDate":"2025-05-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143928528","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}