Michael Gadermayr , Maximilian Tschuchnig , Lea Maria Stangassinger , Christina Kreutzer , Sebastien Couillard-Despres , Gertie Janneke Oostingh , Anton Hittmair
{"title":"Improving automated thyroid cancer classification of frozen sections by the aid of virtual image translation and stain normalization","authors":"Michael Gadermayr , Maximilian Tschuchnig , Lea Maria Stangassinger , Christina Kreutzer , Sebastien Couillard-Despres , Gertie Janneke Oostingh , Anton Hittmair","doi":"10.1016/j.cmpbup.2023.100092","DOIUrl":null,"url":null,"abstract":"","PeriodicalId":72670,"journal":{"name":"Computer methods and programs in biomedicine update","volume":"3 ","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Computer methods and programs in biomedicine update","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2666990023000010","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
冷冻切片在手术干预过程中迅速生成。这使得外科医生可以在干预期间等待组织学检查结果,以便根据组织学结果做出手术内决策。然而,与石蜡切片相比,冷冻切片的质量通常会大大降低,导致诊断准确性降低。基于深度学习的图像转换技术促进了不同原生成像技术之间的虚拟转换,有可能将冷冻切片转换为虚拟石蜡切片。可以应用污点归一化来调整进一步不相等的图像特性。我们研究了基于深度学习的图像翻译、传统图像归一化和这些技术的组合对甲状腺癌症诊断计算机辅助决策支持系统的影响。对于分类,采用了基于卷积神经网络特征、k-means聚类和支持向量机的单词袋方法。虽然染色标准化导致总体分类准确率下降(0.703 vs 0.727),但图像翻译导致平均得分增加(0.770),图像翻译和归一化进一步提高了准确性(0.844),并明显缩小了与术后石蜡切片的差距(0.902)。基于深度学习的图像翻译被证明是提高计算机辅助诊断准确性的强大工具,其明显优于传统的染色翻译。这项工作为与专业病理学家一起进行研究提供了强大的动力,他们对冷冻切片和相应的改良切片进行分类,以研究在临床环境中是否达到了类似的效果。