利用全切片图像预测前列腺癌 TMPRSS2:ERG 融合状态的半监督式、基于注意力的深度学习技术

IF 4.1 2区医学 Q2 CELL BIOLOGY

Molecular Cancer Research Pub Date : 2024-04-02 DOI:10.1158/1541-7786.MCR-23-0639

Mohamed Omar, Zhuoran Xu, Sophie B Rand, Mohammad K Alexanderani, Daniela C Salles, Itzel Valencia, Edward M Schaeffer, Brian D Robinson, Tamara L Lotan, Massimo Loda, Luigi Marchionni

{"title":"利用全切片图像预测前列腺癌 TMPRSS2:ERG 融合状态的半监督式、基于注意力的深度学习技术","authors":"Mohamed Omar, Zhuoran Xu, Sophie B Rand, Mohammad K Alexanderani, Daniela C Salles, Itzel Valencia, Edward M Schaeffer, Brian D Robinson, Tamara L Lotan, Massimo Loda, Luigi Marchionni","doi":"10.1158/1541-7786.MCR-23-0639","DOIUrl":null,"url":null,"abstract":"Implications: Our study illuminates the potential of deep learning in effectively inferring key prostate cancer genetic alterations from the tissue morphology depicted in routinely available histology slides, offering a cost-effective method that could revolutionize diagnostic strategies in oncology.","PeriodicalId":19095,"journal":{"name":"Molecular Cancer Research","volume":" ","pages":"347-359"},"PeriodicalIF":4.1000,"publicationDate":"2024-04-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10985477/pdf/","citationCount":"0","resultStr":"{\"title\":\"Semi-Supervised, Attention-Based Deep Learning for Predicting TMPRSS2:ERG Fusion Status in Prostate Cancer Using Whole Slide Images.\",\"authors\":\"Mohamed Omar, Zhuoran Xu, Sophie B Rand, Mohammad K Alexanderani, Daniela C Salles, Itzel Valencia, Edward M Schaeffer, Brian D Robinson, Tamara L Lotan, Massimo Loda, Luigi Marchionni\",\"doi\":\"10.1158/1541-7786.MCR-23-0639\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Implications: Our study illuminates the potential of deep learning in effectively inferring key prostate cancer genetic alterations from the tissue morphology depicted in routinely available histology slides, offering a cost-effective method that could revolutionize diagnostic strategies in oncology.\",\"PeriodicalId\":19095,\"journal\":{\"name\":\"Molecular Cancer Research\",\"volume\":\" \",\"pages\":\"347-359\"},\"PeriodicalIF\":4.1000,\"publicationDate\":\"2024-04-02\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10985477/pdf/\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Molecular Cancer Research\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://doi.org/10.1158/1541-7786.MCR-23-0639\",\"RegionNum\":2,\"RegionCategory\":\"医学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"CELL BIOLOGY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Molecular Cancer Research","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1158/1541-7786.MCR-23-0639","RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"CELL BIOLOGY","Score":null,"Total":0}

引用次数: 0

摘要

前列腺癌（PCa）存在多种基因改变，其中最常见的是TMPRSS2:ERG基因融合，几乎影响到所有病例的一半。随着全切片图像（WSI）的可用性不断提高，本研究引入了一种深度学习（DL）模型，旨在从根治性前列腺切除术标本的H&E染色WSI中检测TMPRSS2:ERG融合。我们利用 TCGA 前列腺腺癌队列（该队列由来自 393 名患者的 436 个 WSIs 组成）开发了一个稳健的 DL 模型，该模型在 10 个不同的分区中进行训练，每个分区由不同的训练集、验证集和测试集组成。该模型的最佳表现是在训练期间 ROC 曲线下面积（AUC）达到 0.84，在 TCGA 测试集上达到 0.72。随后，该模型在一个由来自不同机构的 314 例 WSI 组成的独立队列中进行了验证，在预测 TMPRSS2:ERG 融合方面表现出色，AUC 为 0.73。重要的是，该模型能识别与TMPRSS2:ERG融合相关的高发组织区域，与融合阴性病例相比，这些区域的特点是肿瘤细胞含量更高，免疫和基质特征发生改变。多变量生存分析表明，这些形态特征与较差的生存结果相关，与格里森分级和肿瘤分期无关。这项研究强调了 DL 在从常规切片中推断基因改变并确定其潜在形态学特征方面的潜力，这些特征可能蕴含着预后信息。意义：我们的研究揭示了深度学习在从常规组织学切片中描述的组织形态有效推断关键前列腺癌基因改变方面的潜力，提供了一种可彻底改变肿瘤学诊断策略的经济有效的方法。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

查看原文本刊更多论文

Semi-Supervised, Attention-Based Deep Learning for Predicting TMPRSS2:ERG Fusion Status in Prostate Cancer Using Whole Slide Images.

Implications: Our study illuminates the potential of deep learning in effectively inferring key prostate cancer genetic alterations from the tissue morphology depicted in routinely available histology slides, offering a cost-effective method that could revolutionize diagnostic strategies in oncology.

求助全文

通过发布文献求助，成功后即可免费获取论文全文。去求助

来源期刊

Molecular Cancer Research 医学-细胞生物学

CiteScore

9.90

自引率

0.00%

发文量

280

审稿时长

4-8 weeks

期刊介绍： Molecular Cancer Research publishes articles describing novel basic cancer research discoveries of broad interest to the field. Studies must be of demonstrated significance, and the journal prioritizes analyses performed at the molecular and cellular level that reveal novel mechanistic insight into pathways and processes linked to cancer risk, development, and/or progression. Areas of emphasis include all cancer-associated pathways (including cell-cycle regulation; cell death; chromatin regulation; DNA damage and repair; gene and RNA regulation; genomics; oncogenes and tumor suppressors; signal transduction; and tumor microenvironment), in addition to studies describing new molecular mechanisms and interactions that support cancer phenotypes. For full consideration, primary research submissions must provide significant novel insight into existing pathway functions or address new hypotheses associated with cancer-relevant biologic questions.