基于多损失混合对抗函数和启发式投影算法的逼真医学图像增强方法
TP391; 早期发现新冠肺炎可以及时医疗干预提高患者的存活率,而利用深度神经网络(Deep neural networks,DNN)对新冠肺炎进行检测,可以提高胸部CT对其筛查的敏感性和判读速度.然而,DNN在医学领域的应用受到有限样本和不可察觉的噪声扰动的影响.本文提出了一种多损失混合对抗方法来搜索含有可能欺骗网络的有效对抗样本,将这些对抗样本添加到训练数据中,以提高网络对意外噪声扰动的稳健性和泛化能力.特别是,本文方法不仅包含了风格、原图和细节损失在内的多损失功能从而将医学对抗样本制作成逼真的样式,而且使用启发式投影算法产生具有强聚集性和干扰性的噪声.这些样本被证明具有较强的抗去噪能力和...
Saved in:
| Published in | 数据采集与处理 Vol. 38; no. 5; pp. 1104 - 1111 |
|---|---|
| Main Authors | , , |
| Format | Journal Article |
| Language | Chinese |
| Published |
南京航空航天大学计算机科学与技术学院,南京 211106
01.09.2023
|
| Subjects | |
| Online Access | Get full text |
| ISSN | 1004-9037 |
| DOI | 10.16337/j.1004-9037.2023.05.009 |
Cover
| Abstract | TP391; 早期发现新冠肺炎可以及时医疗干预提高患者的存活率,而利用深度神经网络(Deep neural networks,DNN)对新冠肺炎进行检测,可以提高胸部CT对其筛查的敏感性和判读速度.然而,DNN在医学领域的应用受到有限样本和不可察觉的噪声扰动的影响.本文提出了一种多损失混合对抗方法来搜索含有可能欺骗网络的有效对抗样本,将这些对抗样本添加到训练数据中,以提高网络对意外噪声扰动的稳健性和泛化能力.特别是,本文方法不仅包含了风格、原图和细节损失在内的多损失功能从而将医学对抗样本制作成逼真的样式,而且使用启发式投影算法产生具有强聚集性和干扰性的噪声.这些样本被证明具有较强的抗去噪能力和攻击迁移性.在新冠肺炎数据集上的测试结果表明,基于该算法的对抗攻击增强后的网络诊断正确率提高了4.75%.因此,基于多损失混合和启发式投影算法的对抗攻击的增强网络能够提高模型的建模能力,并具有抗噪声扰动的能力. |
|---|---|
| AbstractList | TP391; 早期发现新冠肺炎可以及时医疗干预提高患者的存活率,而利用深度神经网络(Deep neural networks,DNN)对新冠肺炎进行检测,可以提高胸部CT对其筛查的敏感性和判读速度.然而,DNN在医学领域的应用受到有限样本和不可察觉的噪声扰动的影响.本文提出了一种多损失混合对抗方法来搜索含有可能欺骗网络的有效对抗样本,将这些对抗样本添加到训练数据中,以提高网络对意外噪声扰动的稳健性和泛化能力.特别是,本文方法不仅包含了风格、原图和细节损失在内的多损失功能从而将医学对抗样本制作成逼真的样式,而且使用启发式投影算法产生具有强聚集性和干扰性的噪声.这些样本被证明具有较强的抗去噪能力和攻击迁移性.在新冠肺炎数据集上的测试结果表明,基于该算法的对抗攻击增强后的网络诊断正确率提高了4.75%.因此,基于多损失混合和启发式投影算法的对抗攻击的增强网络能够提高模型的建模能力,并具有抗噪声扰动的能力. |
| Author | 王见 陈芳 成楚凡 |
| AuthorAffiliation | 南京航空航天大学计算机科学与技术学院,南京 211106 |
| AuthorAffiliation_xml | – name: 南京航空航天大学计算机科学与技术学院,南京 211106 |
| Author_FL | WANG Jian CHENG Chufan CHEN Fang |
| Author_FL_xml | – sequence: 1 fullname: WANG Jian – sequence: 2 fullname: CHENG Chufan – sequence: 3 fullname: CHEN Fang |
| Author_xml | – sequence: 1 fullname: 王见 – sequence: 2 fullname: 成楚凡 – sequence: 3 fullname: 陈芳 |
| BookMark | eNo9UE1LAlEUfQuDzPwPLdvMdN9780ZnGdIXCG1q00Zm5r2JBhmhIaJdUFCIpQYmWCSEtCuTFqJpv6b7pH_RRNHqcD4493IWSCqqRIqQJQomtTnPrYQmBbAMB3jOZMC4CcIEcFIk_a_Pk2wcH3iJaVs2dew02cPu-HN8jb2Ovupib6CHQ2xcYn-kq228mOrWK97UsNHHehMndV1t4XQwe2nrt9asc_51Opndd7H2js9PePeBZ3V8fMDJWN-OksAimQvccqyyf5ghu-trO4VNo7i9sVVYLRoxBe4YwpbMcpQXSMZdSwkr-Z2KvMeYVL6kvlQAPuWezAtIiB84SkgmbJeDp8DjPEOWf3uP3Shwo_1SWDk6jJKLpTj0wxO__DMGiGQK_g1EanjO |
| ClassificationCodes | TP391 |
| ContentType | Journal Article |
| Copyright | Copyright © Wanfang Data Co. Ltd. All Rights Reserved. |
| Copyright_xml | – notice: Copyright © Wanfang Data Co. Ltd. All Rights Reserved. |
| DBID | 2B. 4A8 92I 93N PSX TCJ |
| DOI | 10.16337/j.1004-9037.2023.05.009 |
| DatabaseName | Wanfang Data Journals - Hong Kong WANFANG Data Centre Wanfang Data Journals 万方数据期刊 - 香港版 China Online Journals (COJ) China Online Journals (COJ) |
| DatabaseTitleList | |
| DeliveryMethod | fulltext_linktorsrc |
| DocumentTitle_FL | Realistic Medical Image Augmentation by Using Multi-loss Hybrid Adversarial Function and Heuristic Projection Algorithm |
| EndPage | 1111 |
| ExternalDocumentID | sjcjycl202305009 |
| GroupedDBID | 2B. 4A8 92I 93N ADMLS ALMA_UNASSIGNED_HOLDINGS PSX TCJ |
| ID | FETCH-LOGICAL-s1039-56d249ebfd23a4e54202158b22decd1cde00c13bd850cdecf9e5d256a30be0b33 |
| ISSN | 1004-9037 |
| IngestDate | Thu May 29 04:00:12 EDT 2025 |
| IsPeerReviewed | false |
| IsScholarly | true |
| Issue | 5 |
| Keywords | 启发式投影法 adversarial attack multi-loss hybrid medical image augmentation attack transferability heuristic projection algorithm 攻击迁移性 医学图像增强 对抗性攻击 多损失混合 |
| Language | Chinese |
| LinkModel | OpenURL |
| MergedId | FETCHMERGED-LOGICAL-s1039-56d249ebfd23a4e54202158b22decd1cde00c13bd850cdecf9e5d256a30be0b33 |
| PageCount | 8 |
| ParticipantIDs | wanfang_journals_sjcjycl202305009 |
| PublicationCentury | 2000 |
| PublicationDate | 2023-09-01 |
| PublicationDateYYYYMMDD | 2023-09-01 |
| PublicationDate_xml | – month: 09 year: 2023 text: 2023-09-01 day: 01 |
| PublicationDecade | 2020 |
| PublicationTitle | 数据采集与处理 |
| PublicationTitle_FL | Journal of Data Acquisition & Processing |
| PublicationYear | 2023 |
| Publisher | 南京航空航天大学计算机科学与技术学院,南京 211106 |
| Publisher_xml | – name: 南京航空航天大学计算机科学与技术学院,南京 211106 |
| SSID | ssib023646196 ssib001102757 ssib000459638 ssib001164671 ssib006568634 ssib002264227 ssib036439733 ssib057620134 ssib023167944 ssib051372606 |
| Score | 2.3873603 |
| Snippet | TP391; 早期发现新冠肺炎可以及时医疗干预提高患者的存活率,而利用深度神经网络(Deep neural networks,DNN)对新冠肺炎进行检测,可以提高胸部CT对其筛查的敏感性和判读速... |
| SourceID | wanfang |
| SourceType | Aggregation Database |
| StartPage | 1104 |
| Title | 基于多损失混合对抗函数和启发式投影算法的逼真医学图像增强方法 |
| URI | https://d.wanfangdata.com.cn/periodical/sjcjycl202305009 |
| Volume | 38 |
| hasFullText | 1 |
| inHoldings | 1 |
| isFullTextHit | |
| isPrint | |
| journalDatabaseRights | – providerCode: PRVAON databaseName: DOAJ Directory of Open Access Journals issn: 1004-9037 databaseCode: DOA dateStart: 20230101 customDbUrl: isFulltext: true dateEnd: 99991231 titleUrlDefault: https://www.doaj.org/ omitProxy: true ssIdentifier: ssib006568634 providerName: Directory of Open Access Journals – providerCode: PRVEBS databaseName: Inspec with Full Text issn: 1004-9037 databaseCode: ADMLS dateStart: 20180901 customDbUrl: isFulltext: true dateEnd: 99991231 titleUrlDefault: https://www.ebsco.com/products/research-databases/inspec-full-text omitProxy: false ssIdentifier: ssib057620134 providerName: EBSCOhost |
| link | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwnV3Na9RAFA-1XryIouI3Cs5Jtk5mMsnMMdndUgQ9tVC8lM3HKkUq2PZgT4KCUqqtQi1UsSDFm9biobS2_jXOFv8GL773dppNrYIKIcy-eTPvze8lefOyMy-ed0WlulAq57VIFKoGHj-stXIta3B5qEimPChoI-2Nm-HQSHB9VI32HfpRWbU0PZUOZDO_3VfyP1YFGtgVd8n-g2XLToEAZbAvnMHCcP4rG7OmYmaQJTFrBnjWTaTEATNACZluYG2XkvhISSI8sBVnWlMVNDfEHDNDVRp4GkgxiiWcmAXTddcqpg71IDM-FpI6ll1zRZQGyYpY3KQOQaikqgi10gFrGlxdAQ2RUncaQv9JQvo0WBySrIQlNBwtSQRUCWaaTmhCAzShU55EVOfZFf0JB1AG5UZ4QAE61yGBpnugoW4RIVO-rUQC1GtQTbMYyn6vJkQEDQmIFSFO2MUVFhBkCGeN-CSy-opFyHINWfemoPYNwoyMGdepHciNCU5Dgy4ppDIQXSGqgKdxuPF-I5guZhFydk3nmEsMyIaaO-aunR0PjQPgF_WDSl6F2N7nYcW54aIjw7tJdva8n9SVu1xVXJnvvgvtpkW-c4oHXG4oKWvD-EDZ_QBiSPlwuelNM8rFn5Pj2fiD7C4ycUWbbw8LcMm88j7ExRr7nAPoI6LKX_Q-ZsarxDK4I1yI6qbpUIe93IcCUz5UNoHjhxNCv5drUtLUvBf7KF9GEOuX9QpVdKtU9gbqVgYiAtf-MH7aCzjRbk3crkxbh495R128eSnuPjyOe30zd054t-zK1ret53Z1ufNsxa6udzY27MJTu7bZmV2yT3Y6i5_syzm7sGbnX9jt-c7sot1Z3_241Pm8uLv8-PvD7d03K3bui_3w3r7-ah_N23dv7fZW59UmMJz0Rgabw_WhmvvISm2SVoGoMBeBKdJ2LmQLnsyBwChAp0LkRZb7WV5wnvkyzbXi8CNrm0LlECe1JE8Lnkp5yuufuDdRnMb0D1pneRqAWzBBy3Cdy7YoeFu2VbsNkJzxLjssxtxDdHLs14vh7F_wnPOO9O7S817_1P3p4gKEBlPpRbqEfgL5kM7Z |
| linkProvider | Directory of Open Access Journals |
| openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=%E5%9F%BA%E4%BA%8E%E5%A4%9A%E6%8D%9F%E5%A4%B1%E6%B7%B7%E5%90%88%E5%AF%B9%E6%8A%97%E5%87%BD%E6%95%B0%E5%92%8C%E5%90%AF%E5%8F%91%E5%BC%8F%E6%8A%95%E5%BD%B1%E7%AE%97%E6%B3%95%E7%9A%84%E9%80%BC%E7%9C%9F%E5%8C%BB%E5%AD%A6%E5%9B%BE%E5%83%8F%E5%A2%9E%E5%BC%BA%E6%96%B9%E6%B3%95&rft.jtitle=%E6%95%B0%E6%8D%AE%E9%87%87%E9%9B%86%E4%B8%8E%E5%A4%84%E7%90%86&rft.au=%E7%8E%8B%E8%A7%81&rft.au=%E6%88%90%E6%A5%9A%E5%87%A1&rft.au=%E9%99%88%E8%8A%B3&rft.date=2023-09-01&rft.pub=%E5%8D%97%E4%BA%AC%E8%88%AA%E7%A9%BA%E8%88%AA%E5%A4%A9%E5%A4%A7%E5%AD%A6%E8%AE%A1%E7%AE%97%E6%9C%BA%E7%A7%91%E5%AD%A6%E4%B8%8E%E6%8A%80%E6%9C%AF%E5%AD%A6%E9%99%A2%2C%E5%8D%97%E4%BA%AC+211106&rft.issn=1004-9037&rft.volume=38&rft.issue=5&rft.spage=1104&rft.epage=1111&rft_id=info:doi/10.16337%2Fj.1004-9037.2023.05.009&rft.externalDocID=sjcjycl202305009 |
| thumbnail_s | http://utb.summon.serialssolutions.com/2.0.0/image/custom?url=http%3A%2F%2Fwww.wanfangdata.com.cn%2Fimages%2FPeriodicalImages%2Fsjcjycl%2Fsjcjycl.jpg |