Data for: Image-based Phenotyping of Disaggregated Cells Using Deep Learning

Abstract: The ability to phenotype cells is fundamentally important in biological research and medicine. Cur-rent methods rely primarily on fluorescence labeling of specific markers. However, there are many situations where this approach is unavailable or undesirable. Machine learning has been used for image cytometry but has been limited by cell agglomeration and it is currently unclear if this ap-proach can reliably phenotype cells that are difficult to distinguish by the human eye. Here, we show disaggregated single cells can be phenotyped with a high degree of accuracy using low-resolution bright-field and non-specific fluorescence images of the nucleus, cytoplasm, and cyto-skeleton. Specifically, we trained a convolutional neural network using automatically segmented images of cells from eight standard cancer cell-lines. These cells could be identified with an aver-age F1-score of 95.3%, tested using separately acquired images. Here we demonstrate the potential to develop an “electronic eye” to phenotype cells directly from microscopy images. Technical Info: 10X Fluorescent microscopy images of Trypsinized cells. Each Tiff image contains 6 different locations within a Greiner Sensoplate 96-well glass bottom imaging well. Channels are in order: Brightfield, Hoechst, SIR-Actin and Calcein Green. Images were taken on a Nikon TI2E with a DS-QI2 Camera.

摘要：细胞表型鉴定能力是生物学研究与医学领域的核心基础。当前的细胞表型鉴定方法主要依赖于特异性标志物的荧光标记，但诸多场景下该方法无法实施或并不适用。机器学习已被应用于图像细胞术，但受限于细胞聚集现象，且目前尚不明确该方法能否可靠地鉴定人类肉眼难以区分的细胞表型。本研究表明，利用细胞核、细胞质与细胞骨架的低分辨率明场（Brightfield）及非特异性荧光图像，可高精度完成解离后的单细胞表型鉴定。具体而言，研究团队使用来自8种标准癌细胞系的自动分割细胞图像训练了卷积神经网络（Convolutional Neural Network）；通过独立采集的图像进行测试时，该模型对细胞的识别平均F1分数（F1-score）可达95.3%。本研究证实了开发"电子眼"以直接通过显微图像实现细胞表型鉴定的可行性。
技术信息：胰酶消化细胞的10倍荧光显微图像。每张Tiff图像包含格瑞纳（Greiner）Sensoplate 96孔玻底成像板内的6个不同成像位点。成像通道顺序依次为：明场（Brightfield）、Hoechst染料（Hoechst）、SIR-肌动蛋白（SIR-Actin）与钙黄绿素绿（Calcein Green）。图像采集使用搭载DS-QI2相机的尼康（Nikon）TI2E显微镜完成。