Deep Learning Localizes and Identifies Polyps in Real Time With 96% Accuracy in Screening Colonoscopy.
Deep Learning Localizes and Identifies Polyps in Real Time With 96% Accuracy in Screening Colonoscopy.
复制标题
DOI：
10.1053/j.gastro.2018.06.037
复制
发表时间：
2018-10
期刊：
影响因子：
29.4
通讯作者：
Baldi P
中科院分区：
文献类型：
--
作者：
Urban G;Tripathi P;Alkayali T;Mittal M;Jalali F;Karnes W;Baldi P
结肠镜检查预防结直肠癌的益处取决于腺瘤的发现率(ADR)。ADR应反映腺瘤患病率，在筛查年龄人群中估计大于50%。然而，结肠镜检查发现腺瘤的比率从7%到53%不等。据估计，ADR每增加1%，间歇性结直肠癌的风险就会降低3-6%。需要新的策略来增加结肠镜检查中的不良反应。我们用卷积神经网络(图像分析的深度学习模型)测试了计算机辅助图像分析的能力，以改善息肉检测，这是ADR的替代品。我们设计并训练了深度卷积神经网络(CNN)来检测息肉，使用从2000多名患者的结肠镜筛查中收集的一组不同和具有代表性的8641张手标记图像。我们在20个结肠镜检查视频上测试了这些模型，总持续时间为5个小时。专家结肠镜检查专家被要求在9个去识别的结肠镜检查视频中识别所有息肉，这些视频是从存档的视频研究中挑选出来的，无论有没有CNN覆盖的好处。他们的发现与CNN的结果进行了比较，并以CNN辅助的专家审查为参考。当在人工标记的图像上进行测试时，CNN识别息肉的接收器工作特征曲线下面积(ROC-AuC)为0.991，准确率为96.4%。在对28个息肉被切除的结肠镜检查视频的分析中，4名专家审查者发现了另外8个没有CNN帮助的息肉，并在CNN的帮助下确定了另外17个息肉(总共45个)。所有摘除的息肉和经专家审查确认的息肉都由CNN检测到。美国有线电视新闻网的假阳性率为7%。在一组包含4088个独特息肉的8641张结肠镜图像中，美国有线电视新闻网识别息肉的交叉验证准确率为96.4%，ROC-AUC值为0.991。CNN系统可以使用带有当代图形处理单元的普通台式机在实时约束下很好地检测和定位息肉。该系统可以增加ADR和减少间歇性结直肠癌，但需要在大型多中心试验中进行验证。
The benefit of colonoscopy for colorectal cancer prevention depends on the adenoma detection rate (ADR). The ADR should reflect adenoma prevalence rate, estimated to be greater than 50% among the screening-age population. Yet the rate of adenoma detection by colonoscopists varies from 7% to 53%. It is estimated that every 1% increase in ADR reduces the risk of interval colorectal cancers by 3-6%. New strategies are needed to increase the ADR during colonoscopy. We tested the ability of computer-assisted image analysis, with convolutional neural networks (a deep learning model for image analysis), to improve polyp detection, a surrogate of ADR. We designed and trained deep convolutional neural networks (CNN) to detect polyps using a diverse and representative set of 8641 hand labeled images from screening colonoscopies collected from over 2000 patients. We tested the models on 20 colonoscopy videos with a total duration of 5 hours. Expert colonoscopists were asked to identify all polyps in 9 de-identified colonoscopy videos, selected from archived video studies, either with or without benefit of the CNN overlay. Their findings were compared with those of the CNN, using CNN-assisted expert review as the reference. When tested on manually labeled images, the CNN identified polyps with an area under the receiver operating characteristic curve (ROC-AUC) of 0.991 and an accuracy of 96.4%. In the analysis of colonoscopy videos in which 28 polyps were removed, 4 expert reviewers identified 8 additional polyps without CNN assistance that had not been removed and identified an additional 17 polyps with CNN assistance (45 in total). All polyps removed and identified by expert review were detected by the CNN. The CNN had a false-positive rate of 7%. In a set of 8641 colonoscopy images containing 4088 unique polyps the CNN identified polyps with a cross-validation accuracy of 96.4% and ROC-AUC value of 0.991. The CNN system can detect and localize polyps well within real-time constraints using an ordinary desktop machine with a contemporary graphics processing unit. This system could increase ADR and reduce interval colorectal cancers but requires validation in large multicenter trials.
登录
查看更多内容
影响因子：
5.8
作者：
Di Lena, Pietro;Nagata, Ken;Baldi, Pierre
通讯作者：
Baldi, Pierre
影响因子：
64.8
作者：
Esteva A;Kuprel B;Novoa RA;Ko J;Swetter SM;Blau HM;Thrun S
通讯作者：
Thrun S
DOI：
10.3174/ajnr.a5667
发表时间：
2018-07
期刊：
AJNR. American journal of neuroradiology
影响因子：
--
作者：
Chang P;Grinband J;Weinberg BD;Bardis M;Khy M;Cadena G;Su MY;Cha S;Filippi CG;Bota D;Baldi P;Poisson LM;Jain R;Chow D
通讯作者：
Chow D
DOI：
10.1056/nejmoa1309086
发表时间：
2014-04-03
期刊：
The New England journal of medicine
影响因子：
--
作者：
Corley DA;Jensen CD;Marks AR;Zhao WK;Lee JK;Doubeni CA;Zauber AG;de Boer J;Fireman BH;Schottinger JE;Quinn VP;Ghai NR;Levin TR;Quesenberry CP
通讯作者：
Quesenberry CP
影响因子：
14.4
作者：
Baldi, Pierre;Sadowski, Peter
通讯作者：
Sadowski, Peter