Graph temporal ensembling based semi-supervised convolutional neural network with noisy labels for histopathology image analysis.
Graph temporal ensembling based semi-supervised convolutional neural network with noisy labels for histopathology image analysis.
复制标题
DOI：
10.1016/j.media.2019.101624
复制
发表时间：
2020-02
影响因子：
10.9
通讯作者：
Yang, Lin
中科院分区：
文献类型：
--
作者：
Shi, Xiaoshuang;Su, Hai;Xing, Fuyong;Liang, Yun;Qu, Gang;Yang, Lin
虽然卷积神经网络在组织病理学图像分类方面取得了巨大的成功，但它们通常需要大规模的干净注释数据，并且对噪声标签敏感。不幸的是，标记大规模的图像是费力的，昂贵的和低可靠的病理学家。为了解决这些问题，在本文中，我们提出了一种新的基于自集成的深度架构，利用注释图像的语义信息，探索隐藏在未标记数据中的信息，同时对噪声标签具有鲁棒性。具体而言，所提出的架构首先通过使用指数移动平均（EMA）来聚合多个先前训练时期内的特征和标签预测，为训练样本的特征和标签预测创建集成目标。然后，同一类内的集合目标被映射到一个簇，使它们进一步增强。其次，利用一致性成本来形成不同配置下的一致性预测。最后，我们通过对包含数千张图像的肺癌和乳腺癌数据集进行广泛的实验来验证所提出的方法。它可以实现90.5%和89.5%的图像分类准确率分别使用20%的标记的患者在两个数据集上。该性能与所有标记患者的基线方法相当。实验也证明了它的鲁棒性的小百分比的噪声标签。
Although convolutional neural networks have achieved tremendous success on histopathology image classification, they usually require large-scale clean annotated data and are sensitive to noisy labels. Unfortunately, labeling large-scale images is laborious, expensive and lowly reliable for pathologists. To address these problems, in this paper, we propose a novel self-ensembling based deep architecture to leverage the semantic information of annotated images and explore the information hidden in unlabeled data, and meanwhile being robust to noisy labels. Specifically, the proposed architecture first creates ensemble targets for feature and label predictions of training samples, by using exponential moving average (EMA) to aggregate feature and label predictions within multiple previous training epochs. Then, the ensemble targets within the same class are mapped into a cluster so that they are further enhanced. Next, a consistency cost is utilized to form consensus predictions under different configurations. Finally, we validate the proposed method with extensive experiments on lung and breast cancer datasets that contain thousands of images. It can achieve 90.5% and 89.5% image classification accuracy using only 20% labeled patients on the two datasets, respectively. This performance is comparable to that of the baseline method with all labeled patients. Experiments also demonstrate its robustness to small percentage of noisy labels.
登录
查看更多内容
影响因子：
10.9
作者：
Cheplygina, Veronika;de Bruijne, Marleen;Pluim, Josien P. W.
通讯作者：
Pluim, Josien P. W.
DOI：
10.1093/bioinformatics/btw252
发表时间：
2016-06-15
期刊：
Bioinformatics (Oxford, England)
影响因子：
--
作者：
Kraus OZ;Ba JL;Frey BJ
通讯作者：
Frey BJ
影响因子：
4.6
作者：
Peikari M;Salama S;Nofech-Mozes S;Martel AL
通讯作者：
Martel AL
影响因子：
17.6
作者：
Quellec, Gwenole;Cazuguel, Guy;Lamard, Mathieu
通讯作者：
Lamard, Mathieu
影响因子：
3.8
作者：
GALEA, MH;BLAMEY, RW;ELLIS, IO
通讯作者：
ELLIS, IO