{"title":"[用于肌肉超声图像肌炎分类的轻量级卷积神经网络]。","authors":"Hao Tan, Xun Lang, Tao Wang, Bingbing He, Zhiyao Li, Yu Lu, Yufeng Zhang","doi":"10.7507/1001-5515.202301023","DOIUrl":null,"url":null,"abstract":"<p><p>Existing classification methods for myositis ultrasound images have problems of poor classification performance or high computational cost. Motivated by this difficulty, a lightweight neural network based on a soft threshold attention mechanism is proposed to cater for a better IIMs classification. The proposed network was constructed by alternately using depthwise separable convolution (DSC) and conventional convolution (CConv). Moreover, a soft threshold attention mechanism was leveraged to enhance the extraction capabilities of key features. Compared with the current dual-branch feature fusion myositis classification network with the highest classification accuracy, the classification accuracy of the network proposed in this paper increased by 5.9%, reaching 96.1%, and its computational complexity was only 0.25% of the existing method. The obtained results support that the proposed method can provide physicians with more accurate classification results at a lower computational cost, thereby greatly assisting them in their clinical diagnosis.</p>","PeriodicalId":39324,"journal":{"name":"生物医学工程学杂志","volume":"41 5","pages":"895-902"},"PeriodicalIF":0.0000,"publicationDate":"2024-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11527749/pdf/","citationCount":"0","resultStr":"{\"title\":\"[A lightweight convolutional neural network for myositis classification from muscle ultrasound images].\",\"authors\":\"Hao Tan, Xun Lang, Tao Wang, Bingbing He, Zhiyao Li, Yu Lu, Yufeng Zhang\",\"doi\":\"10.7507/1001-5515.202301023\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><p>Existing classification methods for myositis ultrasound images have problems of poor classification performance or high computational cost. Motivated by this difficulty, a lightweight neural network based on a soft threshold attention mechanism is proposed to cater for a better IIMs classification. The proposed network was constructed by alternately using depthwise separable convolution (DSC) and conventional convolution (CConv). Moreover, a soft threshold attention mechanism was leveraged to enhance the extraction capabilities of key features. Compared with the current dual-branch feature fusion myositis classification network with the highest classification accuracy, the classification accuracy of the network proposed in this paper increased by 5.9%, reaching 96.1%, and its computational complexity was only 0.25% of the existing method. The obtained results support that the proposed method can provide physicians with more accurate classification results at a lower computational cost, thereby greatly assisting them in their clinical diagnosis.</p>\",\"PeriodicalId\":39324,\"journal\":{\"name\":\"生物医学工程学杂志\",\"volume\":\"41 5\",\"pages\":\"895-902\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-10-25\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11527749/pdf/\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"生物医学工程学杂志\",\"FirstCategoryId\":\"1087\",\"ListUrlMain\":\"https://doi.org/10.7507/1001-5515.202301023\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q4\",\"JCRName\":\"Medicine\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"生物医学工程学杂志","FirstCategoryId":"1087","ListUrlMain":"https://doi.org/10.7507/1001-5515.202301023","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"Medicine","Score":null,"Total":0}
[A lightweight convolutional neural network for myositis classification from muscle ultrasound images].
Existing classification methods for myositis ultrasound images have problems of poor classification performance or high computational cost. Motivated by this difficulty, a lightweight neural network based on a soft threshold attention mechanism is proposed to cater for a better IIMs classification. The proposed network was constructed by alternately using depthwise separable convolution (DSC) and conventional convolution (CConv). Moreover, a soft threshold attention mechanism was leveraged to enhance the extraction capabilities of key features. Compared with the current dual-branch feature fusion myositis classification network with the highest classification accuracy, the classification accuracy of the network proposed in this paper increased by 5.9%, reaching 96.1%, and its computational complexity was only 0.25% of the existing method. The obtained results support that the proposed method can provide physicians with more accurate classification results at a lower computational cost, thereby greatly assisting them in their clinical diagnosis.