Irbm最新文献

{"title":"Multiscale Passive Mechanical Characterization of Rodent Skeletal Muscle","authors":"Philippe Pouletaut ,&nbsp;Yoann Tatarenko ,&nbsp;Mashhour K. Chakouch ,&nbsp;Meng Li ,&nbsp;Venus Joumaa ,&nbsp;John R. Hawse ,&nbsp;Walter Herzog ,&nbsp;Simon Chatelin ,&nbsp;Sabine F. Bensamoun","doi":"10.1016/j.irbm.2023.100800","DOIUrl":"https://doi.org/10.1016/j.irbm.2023.100800","url":null,"abstract":"<div><h3>Purpose</h3><p>To experimentally measure selected passive properties of skeletal muscle<span> at three different scales (macroscopic scale: whole muscle, microscopic scale: single skinned fiber, and submicron scale: single myofibril) within the same animal model (mice), and to compare a primarily slow-twitch fiber muscle (soleus) and a primarily fast-twitch fiber muscle (extensor digitorum longus, EDL) for each scale.</span></p></div><div><h3>Methods</h3><p>Healthy 3 months old wild-type C57BL6 mice were used. To characterize each scale, soleus (N = 11), EDL (N = 9), slow fibers (N = 17), fast fibers (N = 16), and myofibrils from soleus (N = 11) and EDL (N = 11) were harvested. Passive mechanical (ramp, relaxation) tests were applied at each scale to compare the passive properties (Young's modulus, static and dynamic stresses) within a given scale, across scales and between muscle types.</p></div><div><h3>Results</h3><p>The soleus and EDL showed significant passive mechanical differences at the macroscopic scale while no variation was observed between both tissues at the microscopic and submicron scales. The results highlight the importance of the scale that is used to mechanically characterize a multiscale tissue.</p></div><div><h3>Conclusion</h3><p>The present work will allow for a better understanding of the multiscale passive mechanical properties for two muscles with vastly differing physiological and metabolic properties. This study provides referent data to the body of literature that can be built upon in future work.</p></div>","PeriodicalId":14605,"journal":{"name":"Irbm","volume":"44 6","pages":"Article 100800"},"PeriodicalIF":4.8,"publicationDate":"2023-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49756363","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Identifying Laryngeal Neoplasms in Laryngoscope Images via Deep Learning Based Object Detection: A Case Study on an Extremely Small Data Set","authors":"Shijie Fang ,&nbsp;Jia Fu ,&nbsp;Chen Du ,&nbsp;Tong Lin ,&nbsp;Yan Yan","doi":"10.1016/j.irbm.2023.100799","DOIUrl":"https://doi.org/10.1016/j.irbm.2023.100799","url":null,"abstract":"<div><h3>Objectives</h3><p><span><span>Laryngoscopy is a medical procedure for obtaining a view of the human </span>larynx. It is challenging for clinicians to distinguish </span>laryngeal neoplasms<span> by human visual observation. Recent deep learning methods can assist clinicians in improving the accuracy of distinguishing. However, existed methods are often trained on large-scale private datasets, while other researchers and hospitals can neither access these private datasets nor afford to build such large-scale datasets. In this paper, we focus on identifying laryngeal neoplasms under the “small data” regime, which is more important for many small hospitals to investigate deep learning models for diagnosis.</span></p></div><div><h3>Material and methods</h3><p>We build an extremely small dataset consisting of 279 laryngoscopic images of different categories. We found that traditional deep learning models for image classification<span> cannot achieve satisfactory performance for small data, due to the great variability of recording laryngoscopic images and the small area of the neoplasms. To address these difficulties, we propose to employ object detection methods for this small data problem. Concretely, a Faster R-CNN is implemented here, which combines the DropBlock regularization technique to alleviate overfitting additionally.</span></p></div><div><h3>Results</h3><p>Compared to previous methods, our model is more robust to overfitting and can predict the location and category of detected neoplasms simultaneously. Our method achieves 73.00% overall accuracy, which is higher than the average of clinicians (65.05%) and the recent state-of-the-art classification method (65.00%).</p></div><div><h3>Conclusion</h3><p>The proposed method shows great ability to detect both the category and location of neoplasms and can be served as a screening tool to help the final decisions of clinicians.</p></div>","PeriodicalId":14605,"journal":{"name":"Irbm","volume":"44 6","pages":"Article 100799"},"PeriodicalIF":4.8,"publicationDate":"2023-09-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49702659","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Simplified Muscle-Recruitment Strategy During Walking in Parkinson's Disease People: A Time-Frequency Analysis of EMG Signal","authors":"Francesco Di Nardo ,&nbsp;Marco Romanato ,&nbsp;Fabiola Spolaor ,&nbsp;Daniele Volpe ,&nbsp;Sandro Fioretti ,&nbsp;Zimi Sawacha","doi":"10.1016/j.irbm.2023.100798","DOIUrl":"https://doi.org/10.1016/j.irbm.2023.100798","url":null,"abstract":"<div><h3>Objective</h3><p>Although gait analysis<span> has been widely adopted to describe Parkinson's disease (PD) dysfunctions during walking, few efforts have been made to understand muscle activity role. The current study aims to characterize lower-limb-muscle recruitment during walking in time-frequency domain, based on Continuous Wavelet Transform (CWT) analysis of surface-electromyography (sEMG) signal from lower-limb muscles.</span></p></div><div><h3>Materials and methods</h3><p>sEMG signals from Tibialis Anterior (TA), Gastrocnemius Lateralis (GL), Rectus Femoris (RF), and Biceps Femoris (BF) of 20 people with PD and 10 age-matched healthy controls (HC) were acquired during gait. sEMG signals were processed applying a CWT-based approach to assess the occurrence frequency (OF, i.e., the percentage of strides of each muscle activation occurrence) and the frequency content of each muscle activation (in Hz). These parameters are rarely quantified in PD.</p></div><div><h3>Results</h3><p>Compared to HC, people with PD displayed a significant decrease (p&lt;0.05) in median OF on RF, BF, and TA, indicating a tendency to reduce the global involvement of lower-limb muscles. No significant differences (p&gt;0.05) in OF were detected among muscle within the same population. No significant changes (p&gt;0.05) in frequency content were revealed in PD.</p></div><div><h3>Conclusion</h3><p>This analysis suggests that people with PD are characterized by a reduced recruitment of those muscles typically adopted to finely control body-segment motion and a concomitant increased recruitment of those muscles mainly involved in locomotion. No substantial alteration in recruiting muscle fibers is associated with PD. These findings suggest that people with PD are inclined to adopt simpler muscular-recruitment strategies during walking, compared to HC.</p></div>","PeriodicalId":14605,"journal":{"name":"Irbm","volume":"44 6","pages":"Article 100798"},"PeriodicalIF":4.8,"publicationDate":"2023-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49728762","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Design of Novel Artificial Anal Sphincter","authors":"MingHui Wang ,&nbsp;YuCheng Liao ,&nbsp;YunLong Liu ,&nbsp;BoLin Liu ,&nbsp;HongLiu Yu","doi":"10.1016/j.irbm.2023.100796","DOIUrl":"https://doi.org/10.1016/j.irbm.2023.100796","url":null,"abstract":"<div><h3>Objectives</h3><p><span><span>Artificial anal sphincter is considered to be a good method for the </span>treatment of severe </span>fecal incontinence, but there are problems of low biomechanical compatibility in clinical application. The purpose of this study is to design a novel artificial anal sphincter that can output stable loading force to solve the mechanical mismatch between artificial anal sphincter and intestinal tissue caused by excessive or too small local pressure on the intestine.</p></div><div><h3>Material and methods</h3><p><span>Aiming at the shortcomings of the existing artificial anal sphincter, a novel artificial anal sphincter with constant force mechanism is designed based on the C-shaped SMA sheet of Ti-55.9at%Ni and combined with the normal defecation mechanism of human body. In this paper, the chord length l of C-shaped SMA sheet is used as the design variable, and the artificial anal sphincter is optimized and verified initially by using </span>finite element analysis method.</p></div><div><h3>Results</h3><p>The results show that the artificial anal sphincter can achieve a constant force to clamp the intestine in a large displacement range when the chord length l of the C-shaped SMA is 13 mm, the chord height h is 1.5 mm, the width w is 2 mm, and the thickness <em>δ</em> is 0.2 mm.</p></div><div><h3>Conclusion</h3><p>The design of artificial anal sphincter with constant force loading mechanism has a good effect in solving the biomechanical compatibility problem between the implant device and the intestinal tissue. In addition, the designed artificial anal sphincter occupies a small space, which provides a new idea for future clinical application.</p></div>","PeriodicalId":14605,"journal":{"name":"Irbm","volume":"44 6","pages":"Article 100796"},"PeriodicalIF":4.8,"publicationDate":"2023-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49702656","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Effect of Local Network Characteristics on the Performance of the SSVEP Brain-Computer Interface","authors":"Pengfei Ma ,&nbsp;Chaoyi Dong ,&nbsp;Ruijing Lin ,&nbsp;Shuang Ma ,&nbsp;Huanzi Liu ,&nbsp;Dongyang Lei ,&nbsp;Xiaoyan Chen","doi":"10.1016/j.irbm.2023.100781","DOIUrl":"https://doi.org/10.1016/j.irbm.2023.100781","url":null,"abstract":"<div><h3>Objective</h3><p>For decades, a great deal of interest in investigating brain network functional connective features has arisen in brain-computer interfaces (BCIs) based on steady-state visual evoked potentials (SSVEPs). Traditional decoding algorithms, for example, canonical correlation analysis (CCA), only consider the inherent properties of each channel in terms of feature extraction for the single channel electroencephalogram (EEG) signal, with inadequate features that cannot fully utilize the information transmitted by the brain.</p></div><div><h3>Material and methods</h3><p>This paper proposes a fused feature extraction method, CCA-DTF, which combines CCA with a direct transfer function (DTF) to construct local brain network features with seven leads in the occipital region. First, the features extracted by the CCA algorithm were combined with these features extracted by DTF to analyze the EEG data from 20 subjects. Then, two methods, support vector machine (SVM) and random forest (RF), were used in constructing the classifiers for the four tasks classification of the SSVEP-BCI.</p></div><div><h3>Results</h3><p>The experimental results showed that incorporating local network features (extracted from DTF or Pearson correlation coefficient) can effectively improve the classification average accuracy and the information transfer rate (ITR) of SSVEP, not only for SVM but also for the ensemble method RF. In particular, CCA-DTF plus SVM obtained a 94.52% classification average accuracy and a 49.23 bits/min ITR in a time window of 2 seconds. The performance was 5.57% and 8.01 bits/min higher than those of traditional CCA plus SVM, respectively.</p></div><div><h3>Conclusion</h3><p>The proposed feature extraction method based on local network features is robust for improving the performance of SSVEP-BCI significantly, which has a perspective of being used in neural rehabilitation engineering field.</p></div>","PeriodicalId":14605,"journal":{"name":"Irbm","volume":"44 4","pages":"Article 100781"},"PeriodicalIF":4.8,"publicationDate":"2023-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49700807","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"On the Utility of Parents' Historical Data to Investigate the Causes of Autism Spectrum Disorder: A Data Mining-Based Framework","authors":"Zahid Halim ,&nbsp;Gohar Khan ,&nbsp;Babar Shah ,&nbsp;Rabia Naseer ,&nbsp;Sajid Anwar ,&nbsp;Ahsan Shah","doi":"10.1016/j.irbm.2023.100780","DOIUrl":"https://doi.org/10.1016/j.irbm.2023.100780","url":null,"abstract":"<div><h3>Objective</h3><p>Autism Spectrum Disorder (ASD) is acknowledged as a challenge that influences the learning ability of adolescents and also negatively impacts their families. Autism may be caused due to environmental exposure or genetically inherited disorder, however, no definitive or universally customary reasons are known. This makes the issue fairly challenging.</p></div><div><h3>Material and methods</h3><p><span>This work focuses on identifying the reasons of ASD utilizing computational methods. For this, data is collected that focuses on parental history for finding the trigged features by reviewing antenatal, perinatal, and infant hazard factors of ASD. Afterwards, ML techniques are applied on the collected instances to develop a predictive model and identify the reasons to ASD. While collecting the data, samples are obtained for ASD and non-ASD individuals both. A total of 115 features are obtained from each subject. The collected dataset has 47% samples of the subjects with ASD. Dimensionality reduction, and four feature selection methods are applied on the data to eliminate noise and least valued features. The data is verified using two clustering techniques, i.e., </span><em>k</em>-means and <em>k</em>-medoid. To validate the clustering results five clustering validation indices are used. Later, three classifiers, i.e. <em>k</em>-nearest neighbor (<em>k</em><span><span>-NN), Support Vector Machine (SVM), and </span>Artificial Neural Network (ANN) are trained to predict cases with ASD. The frequent items mining technique and the descriptive analysis of the clustered data are utilized to identify the factors that may cause ASD.</span></p></div><div><h3>Results</h3><p>The proposed framework enables to identify the features that may contribute towards ASD. Whereas, for the classification part, SVM classifier performs better than others do with an average accuracy of 98.34% in predicting the ASD cases.</p></div><div><h3>Conclusion</h3><p><span>The results identified stress as the dominant feature and environmental factors, like frequent use of canned food and plastic/steel bottles during </span>fertilization period that may contribute towards ASD.</p></div>","PeriodicalId":14605,"journal":{"name":"Irbm","volume":"44 4","pages":"Article 100780"},"PeriodicalIF":4.8,"publicationDate":"2023-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49700778","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The Effect of Mesh Orientation, Defect Location and Size on the Biomechanical Compatibility of Hernia Mesh","authors":"Wei He ,&nbsp;Fei Shen ,&nbsp;Zhiwei Xu ,&nbsp;Baoqing Pei ,&nbsp;Huiqi Xie ,&nbsp;Xiaoming Li","doi":"10.1016/j.irbm.2023.100777","DOIUrl":"https://doi.org/10.1016/j.irbm.2023.100777","url":null,"abstract":"<div><h3>Objectives</h3><p>Satisfactory biomechanical compatibility of implants<span> is important for obtaining desired tissue repair<span> efficiency. Here, we investigated the combined effects of three important influencing factors, mesh<span> orientation, defect location and size, on biomechanical compatibility of a typical anisotropic mesh by both computational simulation and animal experiment.</span></span></span></p></div><div><h3>Methods</h3><p>Numerical models of rabbits were developed based on CT images and material constitutive models obtained by uniaxial tests, during which two orientations, two defect locations and two defect sizes<span><span> were investigated. Corresponding pneumoperitoneum tests on rabbits and non-invasive measurements on the displacement of </span>abdominal wall surface were performed for validation.</span></p></div><div><h3>Results</h3><p>Numerical results showed that the displacement of abdominal wall was limited when the stiffest direction of mesh was parallel to the cranio-caudal direction, but the stress in suture area was greatly reduced. When the defect was located at the junction of different muscles, the strain distribution became uneven. In addition, for the defects with smaller size, difference between the results caused by different mesh orientations was smaller. Animal experimental results were in good agreement with the numerical results. Further simulations for a hypothetical mesh orientation showed that the meshes exhibited better biomechanical compatibility when their stiffest direction was consistent with that of oblique muscles for all four different defects.</p></div><div><h3>Conclusion</h3><p>The mesh orientation was the most influential factor and the proper orientation of the mesh was not necessarily consistent with the anisotropy of the defect tissue. In addition, the mesh design with asymmetric stiffness should be considered for defects at the junction of different tissues. Finally, it is possible to align the stiffest direction of the mesh with that of the defect tissue in repairing small defects to achieve better compliance. Our findings could provide some reliable and instructive guidelines for high-performance anisotropic meshes development and their appropriate selection and placement in surgery. And methods proposed in this study could be used to comprehensively and instructively evaluate the biomechanical compatibility of hernia meshes, predict their repair effect, and determine their appropriate positioning before they are put into clinical use.</p></div>","PeriodicalId":14605,"journal":{"name":"Irbm","volume":"44 4","pages":"Article 100777"},"PeriodicalIF":4.8,"publicationDate":"2023-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49700836","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"BCDNet: An Optimized Deep Network for Ultrasound Breast Cancer Detection","authors":"S.-Y. Lu ,&nbsp;S.-H. Wang ,&nbsp;Y.-D. Zhang","doi":"10.1016/j.irbm.2023.100774","DOIUrl":"https://doi.org/10.1016/j.irbm.2023.100774","url":null,"abstract":"<div><h3>Objectives</h3><p>Breast cancer is a common but deadly disease among women. Medical imaging is an effective method to diagnose breast cancer, but manual image screening is time-consuming. In this study, a novel computer-aided diagnosis system for breast cancer detection called BCDNet is proposed.</p></div><div><h3>Material and Methods</h3><p>We leverage pre-trained convolutional neural networks (CNNs) for representation learning and propose an adaptive backbone selection algorithm to obtain the best CNN model. An extreme learning machine serves as the classifier in the BCDNet, and a bat algorithm with chaotic maps is put forward to further optimize the parameters in the classifiers. A public ultrasound image dataset is used in the experiments based on 5-fold cross-validation.</p></div><div><h3>Results</h3><p>Simulation results suggest that our BCDNet outperforms several state-of-the-art breast cancer detection methods in terms of accuracy.</p></div><div><h3>Conclusion</h3><p>The proposed BCDNet is a useful auxiliary tool that can be applied in clinical screening for breast cancer.</p></div>","PeriodicalId":14605,"journal":{"name":"Irbm","volume":"44 4","pages":"Article 100774"},"PeriodicalIF":4.8,"publicationDate":"2023-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49700835","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The Impact of Missing Data on Heart Rate Variability Features: A Comparative Study of Interpolation Methods for Ambulatory Health Monitoring","authors":"Mouna Benchekroun ,&nbsp;Baptiste Chevallier ,&nbsp;Vincent Zalc ,&nbsp;Dan Istrate ,&nbsp;Dominique Lenne ,&nbsp;Nicolas Vera","doi":"10.1016/j.irbm.2023.100776","DOIUrl":"https://doi.org/10.1016/j.irbm.2023.100776","url":null,"abstract":"<div><h3>Objectives</h3><p><span>Heart rate variability (HRV) is a valuable indicator of both physiological and psychological states. However, the accuracy of HRV measurements taken by </span>wearable devices can be compromised by errors during transmission and acquisition. These errors can significantly affect HRV features and are not acceptable for precise HRV analysis used for medical diagnosis. This study aims to address this issue by investigating the effectiveness of four different interpolation methods (Nearest Neighbour - NN, Linear, Shape-preserving piecewise cubic Hermite - Pchip, and cubic spline) in tackling missing RR values in real-time HRV analysis.</p></div><div><h3>Materials and Methods</h3><p>In this study, HRV signals were obtained from Electrocardiograms (ECG) through automatic detection and manually corrected by a specialist, resulting in high-quality signals with no missing or ectopic peaks. To simulate low-quality data acquisition, values were iteratively deleted from each HRV analysis window. The deleted values were then replaced using four different interpolation methods. Time and frequency domain features were computed from both the original and reconstructed signals, and the Mean Absolute Percentage Error (MAPE) was used to compare these features.</p></div><div><h3>Results</h3><p>Results showed that as the percentage of missing values increased, some interpolation methods were more suitable for RR time-series with a greater number of missing data. Furthermore, the study suggests that the impact of interpolation on HRV features varied across different features and that SDNN is the least affected by interpolation. In the time domain, nearest neighbour interpolation gives the best results for up to 50% missing data. Beyond this threshold, it seems better not to use any interpolation for RMSSD. In the frequency domain however, the lowest errors of HRV feature estimation are obtained using linear or Pchip interpolation. To achieve maximum performance, it is recommended to adapt the interpolation method to both the percentage of missing values and the targeted HRV feature.</p></div><div><h3>Conclusion</h3><p>Results highlight the importance of choosing the appropriate interpolation method to accurately estimate HRV features in real-time analysis. Overall, the Pchip interpolation seems to yield the best results on most HRV features as it preserves the linear trend of the data while adding very light waves. The findings can be beneficial in the development of more precise and reliable wearable devices for real-time HRV monitoring.</p></div>","PeriodicalId":14605,"journal":{"name":"Irbm","volume":"44 4","pages":"Article 100776"},"PeriodicalIF":4.8,"publicationDate":"2023-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49700774","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"MCFA-UNet: Multiscale Cascaded Feature Attention U-Net for Liver Segmentation","authors":"Yuran Zhou ,&nbsp;Qianqian Kong ,&nbsp;Yan Zhu ,&nbsp;Zhen Su","doi":"10.1016/j.irbm.2023.100789","DOIUrl":"https://doi.org/10.1016/j.irbm.2023.100789","url":null,"abstract":"<div><h3>Objectives</h3><p>Accurate automatic liver segmentation has important value for subsequent tumor segmentation, diagnosis, and treatment. In this paper, a Multiscale Cascaded Feature Attention U-Net (MCFA-UNet) neural network model was proposed to solve the problem of edge detail feature loss caused by insufficient feature extraction in existing segmentation methods.</p></div><div><h3>Material and methods</h3><p>MCFA-UNet is a 3D segmentation network based on U-Net encoding and decoding structure. First, this paper proposes a multiscale feature cascaded attention (MCFA) module, which extracts multiscale feature information through multiple continuous convolution paths, and uses double attention to realize multiscale feature information fusion of different paths. Second, the attention-gate mechanism is used to fuse different levels of feature information, which reduces the semantic difference between coding and decoding paths. Finally, the deep supervision learning method was employed to optimize the network segmentation effect through the feature information of each hidden layer in the decoding path.</p></div><div><h3>Results</h3><p>MCFA-UNet was evaluated on LiTS and 3DIRCADb datasets. The Dice scores of 0.955 and 0.981 are obtained respectively. Compared with the baseline network, the segmentation accuracy is improved by 5% and 3.5%.</p></div><div><h3>Conclusion</h3><p>Experimental results show that MCFA-UNet has more accurate segmentation performance than baseline model and other advanced methods.</p></div>","PeriodicalId":14605,"journal":{"name":"Irbm","volume":"44 4","pages":"Article 100789"},"PeriodicalIF":4.8,"publicationDate":"2023-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49700808","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}