Weakly supervised learning for classification of lung cytological images using attention-based multiple instance learning.
Weakly supervised learning for classification of lung cytological images using attention-based multiple instance learning.
复制标题
DOI：
10.1038/s41598-021-99246-4
复制
发表时间：
2021-10-13
影响因子：
4.6
通讯作者：
Fujita H
中科院分区：
文献类型：
--
作者：
Teramoto A;Kiriyama Y;Tsukamoto T;Sakurai E;Michiba A;Imaizumi K;Saito K;Fujita H
在细胞学检查中，对可疑细胞的恶性程度和癌症类型进行评估。为了辅助这一点，我们之前提出了一种基于监督学习的自动化方法，将肺细胞学图像中的细胞分类为良性或恶性。然而，通常很难标记所有细胞。在这项研究中，我们开发了一种弱监督方法，用于细胞学图像中良性和恶性肺细胞的分类，该方法使用基于注意力的深度多实例学习（AD MIL）。肺细胞学标本的图像分成小片图像，保存在袋中。然后将每个袋标记为良性或恶性，并使用AD MIL进行分类，并计算注意权值的分布作为彩色图，以确认图像中存在恶性细胞。使用类alexnet卷积神经网络模型的AD MIL分类性能最好，准确率为0.916，优于监督学习。此外，基于注意力权重的整个图像的注意力图允许AD MIL聚焦于大多数恶性细胞。我们的弱监督方法基于监督学习，在没有复杂注释的情况下，以可接受的精度自动分类细胞学图像。
In cytological examination, suspicious cells are evaluated regarding malignancy and cancer type. To assist this, we previously proposed an automated method based on supervised learning that classifies cells in lung cytological images as benign or malignant. However, it is often difficult to label all cells. In this study, we developed a weakly supervised method for the classification of benign and malignant lung cells in cytological images using attention-based deep multiple instance learning (AD MIL). Images of lung cytological specimens were divided into small patch images and stored in bags. Each bag was then labeled as benign or malignant, and classification was conducted using AD MIL. The distribution of attention weights was also calculated as a color map to confirm the presence of malignant cells in the image. AD MIL using the AlexNet-like convolutional neural network model showed the best classification performance, with an accuracy of 0.916, which was better than that of supervised learning. In addition, an attention map of the entire image based on the attention weight allowed AD MIL to focus on most malignant cells. Our weakly supervised method automatically classifies cytological images with acceptable accuracy based on supervised learning without complex annotations.
登录
查看更多内容
影响因子：
2.4
作者：
Yan, Ke;Wang, Xiaosong;Summers, Ronald M.
通讯作者：
Summers, Ronald M.
影响因子：
8
作者：
Carbonneau, Marc-Andre;Cheplygina, Veronika;Gagnon, Ghyslain
通讯作者：
Gagnon, Ghyslain
影响因子：
10.9
作者：
Wang, Shujun;Zhu, Yaxi;Heng, Pheng-Ann
通讯作者：
Heng, Pheng-Ann
影响因子：
3.8
作者：
Teramoto, Atsushi;Fujita, Hiroshi;Tamaki, Tsuneo
通讯作者：
Tamaki, Tsuneo
DOI：
10.1109/jbhi.2020.2983730
发表时间：
2020-12-01
影响因子：
7.7
作者：
Wang, Xi;Tang, Fangyao;Heng, Pheng-Ann
通讯作者：
Heng, Pheng-Ann