Interpretable unsupervised learning enables accurate clustering with high-throughput imaging flow cytometry.

Interpretable unsupervised learning enables accurate clustering with high-throughput imaging flow cytometry.
复制标题

DOI:
10.1038/s41598-023-46782-w
复制
发表时间:
2023-11-23
期刊:
影响因子:
4.6
通讯作者:
--
中科院分区:
综合性期刊3区
文献类型:
--
作者:

文献摘要

参考文献

相似文献

高通量成像流式细胞术 (IFC) 的主要挑战是分析大量成像数据,尤其是在无法获得或难以获得真实标签的应用中。我们提出了一种无监督的深度嵌入算法,即基于深度卷积自动编码器的聚类(DCAEC)模型,可以在没有任何输入标签先验知识的情况下对无标签的 IFC 图像进行聚类。 DCAEC 模型首先将输入图像编码为潜在表示,然后基于潜在表示进行聚类。使用 DCAEC 模型,我们使用 3D IFC 图像和 3D DCAEC 模型实现了人类白细胞 (WBC) 聚类的平衡准确度为 91.9%,WBC/白血病聚类的准确度为 97.9%。最重要的是,尽管没有人类可识别的特征可以通过蛋白质定位来区分细胞簇,但我们证明融合的 DCAEC 模型可以从无标记 2D 透射和 3D 侧散射图像中实现 85.3% 的簇平衡精度。为了揭示神经网络如何识别超出人类能力的特征,我们使用梯度加权类激活映射方法来自动发现特定于簇的视觉模式。评估结果表明,自动识别的显着图像区域对于不同的簇具有很强的簇特定视觉模式,我们认为这是用于高通量 IFC 细胞分析的可解释神经网络的一大进步。
A primary challenge of high-throughput imaging flow cytometry (IFC) is to analyze the vast amount of imaging data, especially in applications where ground truth labels are unavailable or hard to obtain. We present an unsupervised deep embedding algorithm, the Deep Convolutional Autoencoder-based Clustering (DCAEC) model, to cluster label-free IFC images without any prior knowledge of input labels. The DCAEC model first encodes the input images into the latent representations and then clusters based on the latent representations. Using the DCAEC model, we achieve a balanced accuracy of 91.9% for human white blood cell (WBC) clustering and 97.9% for WBC/leukemia clustering using the 3D IFC images and 3D DCAEC model. Above all, although no human recognizable features can separate the clusters of cells with protein localization, we demonstrate the fused DCAEC model can achieve a cluster balanced accuracy of 85.3% from the label-free 2D transmission and 3D side scattering images. To reveal how the neural network recognizes features beyond human ability, we use the gradient-weighted class activation mapping method to discover the cluster-specific visual patterns automatically. Evaluation results show that the automatically identified salient image regions have strong cluster-specific visual patterns for different clusters, which we believe is a stride for the interpretable neural network for cell analysis with high-throughput IFCs.
DOI: 10.1073/pnas.2118068119
发表时间: 2022-02-22
影响因子: 11.1
作者:
Zhang Z;Tang R;Chen X;Waller L;Kau A;Fung AA;Gutierrez B;An C;Cho SH;Shi L;Lo YH
通讯作者: Lo YH
DOI: 10.1242/jcs.089110
发表时间: 2011-10-15
影响因子: 4
作者:
Hung, Mien-Chie;Link, Wolfgang
通讯作者: Link, Wolfgang
基于深度自动编码器的聚类
DOI: 10.3233/ida-140709
发表时间: 2014-01-01
影响因子: 1.7
作者:
Song, Chunfeng;Huang, Yongzhen;Wang, Liang
通讯作者: Wang, Liang
DOI: 10.1038/srep21471
发表时间: 2016-03-15
期刊: Scientific reports
影响因子: 4.6
作者:
Chen CL;Mahjoubfar A;Tai LC;Blaby IK;Huang A;Niazi KR;Jalali B
通讯作者: Jalali B
DOI: 10.1038/nmeth.2084
发表时间: 2012-06-28
期刊: NATURE METHODS
影响因子: 48
作者:
Eliceiri, Kevin W.;Berthold, Michael R.;Goldberg, Ilya G.;Ibanez, Luis;Manjunath, B. S.;Martone, Maryann E.;Murphy, Robert F.;Peng, Hanchuan;Plant, Anne L.;Roysam, Badrinath;Stuurmann, Nico;Swedlow, Jason R.;Tomancak, Pavel;Carpenter, Anne E.
通讯作者: Carpenter, Anne E.