Biomedical Signal Processing and Control最新文献

{"title":"A cross-session non-stationary attention-based motor imagery classification method with critic-free domain adaptation","authors":"Shuai Guo ,&nbsp;Yi Wang ,&nbsp;Xin Zhang ,&nbsp;Baoping Tang","doi":"10.1016/j.bspc.2024.107122","DOIUrl":"10.1016/j.bspc.2024.107122","url":null,"abstract":"<div><div>Recent studies increasingly employ deep learning to decode electroencephalogram (EEG) signals. While deep learning has improved the performance of motor imagery (MI) classification to some extent, challenges remain due to significant variances in EEG data across sessions and the limitations of convolutional neural networks (CNNs). EEG signals are inherently non-stationary, traditional multi-head attention typically uses normalization methods to reduce non-stationarity and improve performance. However, non-stationary factors are crucial inherent properties of EEG signals and provide valuable guidance for decoding temporal dependencies in EEG signals. In this paper, we introduce a novel CNN combined with the Non-stationary Attention (NSA) and Critic-free Domain Adaptation Network (NSDANet), tailored for decoding MI signals. This network starts with temporal–spatial convolution devised to extract spatial–temporal features from EEG signals. It then obtains multi-modal information from average and variance perspectives. We devise a new self-attention module, the Non-stationary Attention (NSA), to capture the non-stationary temporal dependencies of MI-EEG signals. Furthermore, to align feature distributions between the source and target domains, we propose a critic-free domain adaptation network that uses the Nuclear-norm Wasserstein discrepancy (NWD) to minimize the inter-domain differences. NWD complements the original classifier by acting as a critic without a gradient penalty. This integration leverages discriminative information for feature alignment, thus enhancing EEG decoding performance. We conducted extensive cross-session experiments on both BCIC IV 2a and BCIC IV 2b dataset. Results demonstrate that the proposed method outperforms some existing approaches.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"100 ","pages":"Article 107122"},"PeriodicalIF":4.9,"publicationDate":"2024-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142658180","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 deep learning based approach for automatic cardiac events identification","authors":"Yuanshu Li ,&nbsp;Kexin Hong ,&nbsp;Xiaohu Shi ,&nbsp;Wei Pang ,&nbsp;Yubin Xiao ,&nbsp;Peng Zhao ,&nbsp;Dong Xu ,&nbsp;Chunli Song ,&nbsp;Xu Zhou ,&nbsp;You Zhou","doi":"10.1016/j.bspc.2024.107164","DOIUrl":"10.1016/j.bspc.2024.107164","url":null,"abstract":"<div><div>Visually identifying End-Diastolic (ED) and End-Systolic (ES) frames from 2D echocardiographic videos without electrocardiogram is time-consuming but a fundamental step in routine clinical practice for assessment of cardiac structure and functionality. In recent years, several algorithms such as segmentation-based methods and regression-based methods, have been proposed to automatically identify ED and ES frames for automated cardiac function assessment. However, these methods require high quality images or large datasets which are difficult to obtain. In this work, we propose the first classification-based method, which is combined with a three-step postprocess for ED and ES identification from 2-D echocardiographic videos. We explored a deep-learning based model for this task which has lightweight structure with 175 kb parameters (in hdf5 format) and has no large demand for the number of echocardiographic videos compared with regression-based methods. In particular, we propose a weights-shared convolutional neural network module as the backbone, which is for two adjacent echocardiographic video frames feature extraction; and the backbone is combined with a classification module, which predicts the two frames’ relationship for volume-time prediction. Based on this, ED and ES can be effectively identified. We trained and evaluated the proposed method on apical 4 chamber views and apical 2 chamber views. For both cardiac views, the accuracies are above 0.95 and the AUCs (area under the ROC) are above 0.99. For ultimate ED and ES prediction, all the precisions are above 97% and all average Frame Distances(aFDs) are less than 2 frame errors. Moreover, we demonstrate that our method has considerable generalizability.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"100 ","pages":"Article 107164"},"PeriodicalIF":4.9,"publicationDate":"2024-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142657666","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":"Semi-supervised spatial-temporal calibration and semantic refinement network for video polyp segmentation","authors":"Feng Li ,&nbsp;Zetao Huang ,&nbsp;Lu Zhou ,&nbsp;Haixia Peng ,&nbsp;Yimin Chu","doi":"10.1016/j.bspc.2024.107127","DOIUrl":"10.1016/j.bspc.2024.107127","url":null,"abstract":"&lt;div&gt;&lt;div&gt;Automated video polyp segmentation (VPS) was of vitality for the early prevention and diagnosis of colorectal cancer (CRC). However, existing deep learning-based automatic polyp segmentation methods mainly focused on independent static images and struggled to perform well due to neglecting spatial–temporal relationships among successive video frames, while requiring massive frame-by-frame annotations. To better alleviate these challenges, we proposed a novel semi-supervised spatial–temporal calibration and semantic refinement network (STCSR-Net) dedicated to VPS, which simultaneously considered both the inter-frame temporal consistency in video clips and intra-frame semantic-spatial information. It was composed of a segmentation pathway and a propagation pathway by use of a co-training scheme for supervising the predictions on un-annotated images in a semi-supervised learning fashion. Specifically, we proposed an adaptive sequence calibration (ASC) block in segmentation pathway and a dynamic transmission calibration (DTC) block in propagation pathway to fully take advantage of valuable temporal cues and maintain the prediction temporally consistent among consecutive frames. Meanwhile, in these two branches, we introduced residual block (RB) to suppress irrelevant noisy information and highlight rich local boundary details of polyp lesions, while constructed multi-scale context extraction (MCE) module to enhance multi-scale high-level semantic feature expression. On that basis, we designed progressive adaptive context fusion (PACF) module to gradually aggregate multi-level features under the guidance of reinforced high-level semantic information for eliminating semantic gaps among them and promoting the discrimination capacity of features for targeting polyp objects. Through synergistic combination of RB, MCE and PACF modules, semantic-spatial correlations on polyp lesions within each frame could be established. Coupled with the context-free loss, our model merged feature representations of neighboring frames to diminish the dependency on varying contexts within consecutive frames and strengthen its robustness. Extensive experiments substantiated that our model with 100% annotation ratio achieved state-of-the-art performance on challenging datasets. Even trained under 50% annotation ratio, our model exceled significantly existing state-of-the-art image-based and video-based polyp segmentation models on the newly-built local TRPolyp dataset by at least 1.3% and 0.9% enhancements in both mDice and mIoU, whilst exhibited comparable performance to top rivals attained through using fully supervised approach on publicly available CVC-612, CVC-300 and ASU-Mayo-Clinic benchmarks. Notably, our model showcased exceptionally well in videos containing complex scenarios like motion blur and occlusion. Beyond that, it also harvested approximately 0.794 mDice and 0.707 mIoU at an inference of 0.036 s per frame in endoscopist-machine competition, which ","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"100 ","pages":"Article 107127"},"PeriodicalIF":4.9,"publicationDate":"2024-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142658179","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":"Explainable AI-based method for brain abnormality diagnostics using MRI","authors":"Mohamed Hosny ,&nbsp;Ahmed M. Elshenhab ,&nbsp;Ahmed Maged","doi":"10.1016/j.bspc.2024.107184","DOIUrl":"10.1016/j.bspc.2024.107184","url":null,"abstract":"<div><div>Detecting brain abnormalities using magnetic resonance imaging (MRI) is a vital frontier in neurological research. Therefore, accurate methods are essential for guiding neurologists in diagnosing enigmatic disorders such as Alzheimer’s disease (AD) and brain tumors. These methods aid in the early detection and treatment of these formidable conditions. However, traditional techniques often suffer from high computational complexity and efficiency. Additionally, existing detection models lack the ability to explain their predictions, rendering them untrustworthy for clinicians. This study presents an explainable framework for automatic brain abnormality detection in MRI images. The methodology includes a robust preprocessing pipeline that ameliorates image relevance through image thresholding, morphological operations and adaptive edge detection using the AutoCanny algorithm. AutoCanny method automatically adjusts thresholds to ensure effective edge detection across different images. Then, the MRI images are fed to efficient vision transformer model (EfficientViT) for classification. EfficientViT features a memory-efficient sandwich layout, cascaded group attention module and optimized parameter reallocation. These innovations collectively enhanced the model efficiency in terms of memory usage, computational complexity and parameter optimization. Moreover, gradient-based Shapley additive explanations is employed to explain the EfficientViT model predictions. EfficientViT achieved the highest accuracy of 99.24%, 97.1%, 99.5% and 98.87% on the AD, Tumor1, Tumor2 and merged datasets, respectively. Furthermore, the proposed model outperformed longstanding deep learning techniques. These findings have significant implications for uncovering hidden information associated with brain abnormality as well as improving the diagnostic process and treatment planning. Our model can aid neurologists in the validation of manual MRI neurological disorders screenings.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"100 ","pages":"Article 107184"},"PeriodicalIF":4.9,"publicationDate":"2024-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142658271","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 deep temporal network for motor imagery classification based on multi-branch feature fusion and attention mechanism","authors":"Jinke Zhao,&nbsp;Mingliang Liu","doi":"10.1016/j.bspc.2024.107163","DOIUrl":"10.1016/j.bspc.2024.107163","url":null,"abstract":"<div><h3>Objective:</h3><div>In recent years, the Brain-Computer Interface (BCI) technology has witnessed rapid advancements. Motor Imagery (MI), as one of the BCI paradigms, has found extensive applications in domains such as rehabilitation, entertainment, and neuroscience. How to conduct effective classification of it has emerged as one of the primary research issues. Electroencephalography (EEG) serves as an essential tool for studying the classification of MI. However, the existing models are incapable of fully extracting effective motion information from the interfered electroencephalogram data, leading to the final classification effect falling short of the expected goals. In response to this problem, we propose a deep temporal network based on multi-branch feature fusion and attention mechanism. This network incorporates a combination of multi-branch feature fusion, feature expansion, attention, and temporal decoding modules.</div></div><div><h3>Methods:</h3><div>First, primary features of EEG signals are extracted using a multi-branch convolutional neural network, followed by feature fusion. Subsequently, feature augmentation and attention mechanisms are employed to reduce noise interference while highlighting essential MI intentions. Finally, a temporal decoding module is utilized to deeply explore temporal information in MI data and perform classification.</div></div><div><h3>Results:</h3><div>The model performance was tested on the BCI_IV_2a, BCI_IV_2b, and OPenBMI datasets using both subject-specific and subject-independent experimental methods. The model achieved significant performance improvements on all three datasets, achieving accuracy of 81.21%, 93.12%, and 75.9%, respectively, better than other baseline models.</div></div><div><h3>Conclusion:</h3><div>Experimental results indicate that the proposed model leverages deep learning techniques for the classification of different types of MI, providing a reference framework for the development of more efficient MI-BCI systems.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"100 ","pages":"Article 107163"},"PeriodicalIF":4.9,"publicationDate":"2024-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142658272","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":"2D/3D fast fine registration in minimally invasive pelvic surgery","authors":"Fujiao Ju ,&nbsp;Yuan Li ,&nbsp;Jingxin Zhao ,&nbsp;Mingjie Dong","doi":"10.1016/j.bspc.2024.107145","DOIUrl":"10.1016/j.bspc.2024.107145","url":null,"abstract":"<div><div>The 2D/3D rigid registration between preoperative 3D CT and intraoperative 2D X-ray is a crucial step in minimally invasive pelvic surgery. The deep learning-based 2D/3D registration methods address the inefficiencies of traditional approaches. However, the wide range of spatial transformation parameters and other complexities pose significant challenges for achieving accurate registration in a single step. Additionally, the stylistic differences between Digitally Reconstructed Radiographs (DRRs) used in training and real X-ray images limit the practical applicability of most methods. To overcome these challenges, we propose a 2D/3D fast registration framework comprising a coarse registration network, fine registration based on key point tracking and alignment, and domain adaptation. Coarse registration using plug-and-play attention is introduced to preliminarily estimate transformation parameters. Then we design a key point tracking network to match key points between different images, and leverage points alignment to achieve fine registration. To address the stylistic differences between DRR and X-ray images, we investigate a domain adaptation network. The experiments were conducted on DRR and X-ray images, respectively. Our method achieved a mean absolute error of 0.58 on DRR and a structural similarity of 78% on X-ray, outperforming baseline methods. Extensive ablation studies demonstrate that fine registration and domain adaptation significantly improve registration performance.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"100 ","pages":"Article 107145"},"PeriodicalIF":4.9,"publicationDate":"2024-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142652847","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":"2MSPK-Net: A nuclei segmentation network based on multi-scale, multi-dimensional attention, and SAM prior knowledge","authors":"Gongtao Yue ,&nbsp;Xiaoguang Ma ,&nbsp;Wenrui Li ,&nbsp;Ziheng An ,&nbsp;Chen Yang","doi":"10.1016/j.bspc.2024.107140","DOIUrl":"10.1016/j.bspc.2024.107140","url":null,"abstract":"<div><div>Refined nuclei segmentation is of great significance for diagnosing the pathological conditions of tumor tissues. Although existing encoder–decoder networks have achieved remarkable progress in nuclei segmentation tasks, practical applications still encounter obstacles, especially for challenging issues such as highly dense nuclei targets and the ambiguity of boundaries between inter-class features, resulting in unsatisfactory segmentation accuracy. In this work, a novel encoder–decoder architecture was proposed to address these issues. Specifically, we first proposed a multi-scale and multi-dimension attention module to capture the contextual dependencies between individual pixels and the overall pixels, where in cross-scale learning was achieved by fusing different scale feature information of the encoding layer. Secondly, we integrated the prior knowledge of SAM into nuclei images to enhance the network’s ability to distinguish fuzzy features. To the best of our knowledge, this was the first attempt to utilize the prior knowledge of SAM to optimize nuclei segmentation tasks. Furthermore, the network was guided to supplement missing detailed features through a reverse erasing strategy and cross-layer information flow. Comprehensive experiments illustrated that the proposed method achieved MIoU improvements of 1.26% and 0.94% on the MoNuSeg and TNBC datasets, respectively, over several SOTA methods, indicating its great potential as a backbone for cancer nuclei segmentation. Code: <span><span>https://github.com/ThirteenYue/2MSPK-Net</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"100 ","pages":"Article 107140"},"PeriodicalIF":4.9,"publicationDate":"2024-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142657663","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":"Extraction of apparent BOLD components in resting state fMRI signals by a novel method called “BOLD-filter”","authors":"Yul-Wan Sung ,&nbsp;Uk-Su Choi ,&nbsp;Seiji Ogawa","doi":"10.1016/j.bspc.2024.107151","DOIUrl":"10.1016/j.bspc.2024.107151","url":null,"abstract":"<div><div>Resting-state functional magnetic resonance imaging (rs-fMRI) is widely used for studying brain diseases and cognition. Unlike task-based fMRI (tb-fMRI), which employs a task protocol to elicit the BOLD response, the rs-fMRI signal is itself considered a BOLD response and is used to examine brain function related to brain diseases and cognition because explicitly identifying the BOLD component in the rs-fMRI signal is challenging. In this study, we propose a novel method called “BOLD-filter” which aims to extract apparent BOLD (aBOLD) components from the rs-fMRI signal before further processing for cognitive and clinical applications. We confirm that applying the BOLD-filter to real data enables us to identify the aBOLD components in the rs-fMRI signal. Under the sufficient condition for aBOLD, among voxels of the whole brain, 84 % exhibited aBOLD signals after applying BOLD-filter, compared to only 14 % before the BOLD-filter, with the BOLD-filter identifying aBOLD signals in six times as many voxels (p = 0.001). Additionally, we observed that utilizing these apparent BOLD components enhances the association of functional connectivity with age in our example. The potential of BOLD filter to ensure BOLD response in rs-fMRI signal analysis is expected to bring a new dimension to rs-fMRI studies in relation to the understanding of brain disorders and cognitive function.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"100 ","pages":"Article 107151"},"PeriodicalIF":4.9,"publicationDate":"2024-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142658178","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":"An efficient data mining technique and privacy preservation model for healthcare data using improved darts game optimizer-based weighted deep neural network and hybrid encryption","authors":"D. Dhinakaran ,&nbsp;L. Srinivasan ,&nbsp;S. Gopalakrishnan ,&nbsp;T.P. Anish","doi":"10.1016/j.bspc.2024.107168","DOIUrl":"10.1016/j.bspc.2024.107168","url":null,"abstract":"<div><div>In recent days, association rule mining techniques have been widely used in healthcare data to provide accurate records that are important to ensure the data privacy. However, making this information public leads to creating attacks on them. In this paper, a secure privacy-preservation scheme for healthcare data is implemented to protect the security of the information for disease prediction in the current healthcare applications. The health data is collected from the benchmark datasets. Initially, the data is encrypted using Fully Homomorphic Encryption and Hyperelliptic Curve Cryptography (FHE-HECC) for the privacy preservation process. This model is developed by combining the Fully Homomorphic Encryption (FHE) and Hyperelliptic Curve Cryptography (HECC). For this encryption, the optimal key is generated using the Improved Darts Game Optimizer (IDGO) leveraging the Darts Game Optimizer (DGO). In the case of data decryption, the above-mentioned cryptography is utilized. The optimally selected key encrypts the data with high security without any breaches. The stored encrypted data is monitored and the disease is recognized using the Weighted Deep Neural Network (W-DNN) method and here, Deep Neural Network (DNN) acts as the fundamental model. Finally, the privacy of health data is preserved and the type of disease is detected by the implemented model. The suggested model attained accuracy of 92.83 which is higher than the existing techniques like GRU with 86.97, RNN with 88.28, LSTM with 91.92, WDNN with 90.10, respectively. The key findings of the suggested approach Proved that it facilitates effective treatment to the patient.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"100 ","pages":"Article 107168"},"PeriodicalIF":4.9,"publicationDate":"2024-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142658294","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":"Efficient retinal exudates detection method using ELNet in diabetic retinopathy assessment","authors":"G. Sasi,&nbsp;A. Kaleel Rahuman","doi":"10.1016/j.bspc.2024.107162","DOIUrl":"10.1016/j.bspc.2024.107162","url":null,"abstract":"<div><div>Diabetic Retinopathy (DR) can be detected at earlier stage by detecting exudates in retinal fundus images. In this article, the exudates are detected and segmented using the proposed Enhanced LeNet (ELNet) classification method. The proposed exudates segmentation method consists of the following modules Retinal image classification and Exudates segmentation. In retinal image classification, the retinal images are data augmented and then ELNet classification architecture is used to classify the retinal image into either normal or abnormal. In exudates segmentation module, the exudates are detected and segmented using Kirsch edge detector. The performance of the exudate’s detection method is improved by detecting and eliminating the blood vessels in the retinal image before detecting the exudates. In this paper, Digital Retinal Images for Vessel Extraction (DRIVE) and Diabetic Retinopathy database (DIARETDB1) retinal image datasets are used for the detection of exudates in the retinal images. In this study, the proposed method showcases remarkable results, demonstrating a sensitivity of 99.31% and 99.31%, specificity of 97.44% and 95%, and an accuracy of 99.09% and 98.8% for the DRIVE and DIARETDB1 datasets, respectively. Exudates detection in both datasets without eliminating OD and retinal blood vessels, we observe similar accuracy rates, Average 96.5% for both datasets. However, when eliminating OD and retinal blood vessels, the accuracy significantly improved for both datasets, reaching approximately 99.2% average. The performance is analyzed and compared with other state of the art methods.</div></div>","PeriodicalId":55362,"journal":{"name":"Biomedical Signal Processing and Control","volume":"100 ","pages":"Article 107162"},"PeriodicalIF":4.9,"publicationDate":"2024-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142657560","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}