Natural killer cell detection, quantification, and subpopulation identification on paper microfluidic cell chromatography using smartphone-based machine learning classification.
Natural killer cell detection, quantification, and subpopulation identification on paper microfluidic cell chromatography using smartphone-based machine learning classification.
复制标题
DOI：
10.1016/j.bios.2021.113916
复制
发表时间：
2022-03-15
影响因子：
12.6
通讯作者：
Yoon JY
中科院分区：
文献类型：
--
作者：
Zenhausern R;Day AS;Safavinia B;Han S;Rudy PE;Won YW;Yoon JY
自然杀伤（NK）细胞是防御病毒感染和癌症的免疫细胞，用于癌症免疫治疗。NK细胞的亚群包括CD 56 dim和CD 56 bright，其产生细胞因子或直接细胞毒性杀伤细胞。外周血中这些细胞的绝对数量和比例与适当的免疫功能有关。目前的细胞因子检测和NK细胞亚群比例的方法需要荧光染料和高度专业化的设备，例如，流式细胞术，因此在临床环境中需要快速细胞定量和亚群分析。在这里，基于智能手机的设备和双组分纸微流控芯片用于识别NK细胞亚群和炎症标志物。一个单位通过智能手机拍摄的视频测量流速，确定未稀释的血沉棕黄层血液样本中的细胞因子（IL-2）和总NK细胞浓度。另一个单流道单元使用抗CD 56纳米颗粒的差异结合在其长度上进行CD 56 dim和CD 56 bright以及细胞的空间分离。智能手机显微镜结合基于云的机器学习预测建模（利用随机森林分类算法）分析了流量数据和NK细胞亚群分化。细胞因子和细胞浓度的检测限分别为98 IU/mL和68个细胞/mL，细胞亚群分析显示准确度为89%。
Natural killer (NK) cells are immune cells that defend against viral infections and cancer and are used in cancer immunotherapies. Subpopulations of NK cells include CD56dim and CD56bright which either produce cytokines or cytotoxically kill cells directly. The absolute number and proportion of these cells in peripheral blood are tied to proper immune function. Current methods of cytokine detection and proportion of NK cell subpopulations require fluorescent dyes and highly specialized equipment, e.g., flow cytometry, thus rapid cell quantification and subpopulation analysis are needed in the clinical setting. Here, a smartphone-based device and a two-component paper microfluidic chip were used towards identifying NK cell subpopulation and inflammatory markers. One unit measured flow velocity via smartphone-captured video, determining cytokine (IL-2) and total NK cell concentrations in undiluted buffy coat blood samples. The other, single flow lane unit performs spatial separation of CD56dim and CD56bright and cells over its length using differential binding of anti-CD56 nanoparticles. A smartphone microscope combined with cloud-based machine learning predictive modeling (utilizing a random forest classification algorithm) analyzed both flow data and NK cell subpopulation differentiation. Limits of detection for cytokine and cell concentrations were 98 IU/mL and 68 cells/mL, respectively, and cell subpopulation analysis showed 89% accuracy.
登录
查看更多内容
影响因子：
14.8
作者：
Chung, Soo;Breshears, Lane E.;Yoon, Jeong-Yeol
通讯作者：
Yoon, Jeong-Yeol
影响因子：
1.2
作者：
Somanchi, Srinivas S.;Senyukov, Vladimir V.;Lee, Dean A.
通讯作者：
Lee, Dean A.
DOI：
10.1038/mto.2016.11
发表时间：
2016
期刊：
Molecular therapy oncolytics
影响因子：
--
作者：
通讯作者：
--
影响因子：
--
作者：
Shrirao AB;Fritz Z;Novik EM;Yarmush GM;Schloss RS;Zahn JD;Yarmush ML
通讯作者：
Yarmush ML
影响因子：
11.7
作者：
Ulep TH;Yoon JY
通讯作者：
Yoon JY