Rapid, label-free and low-cost diagnostic kit for COVID-19 based on liquid crystals and machine learning

We report a label-free method for detection of the SARS-CoV-2 virus in nasopharyngeal swab samples without purification steps and multiplication of the target which simplifies and expedites the analysis process. The kit consists of a textile which is agitated with liquid crystals (LC) and placed in a crossed polarized microscopy. The swab samples are placed on the LCs. In the presence of a particular biomolecule, the direction of LCs changes locally based on the properties of the biomolecule and forms a particular pattern. As the swab samples are not perfectly purified, image processing and machine learning techniques are employed to detect the presence of specific molecules or quantify their concentrations in the medium. The method can differentiate negative and positive COVID-19 samples with an accuracy of 96% and also differentiate COVID-19 from influenza types A and B with an accuracy of 93%. The kit is portable, simple to manufacture, convenient to operate, cost-effective, rapid and sensitive. The simplicity of the specimen processing, the speed of image acquisition, and fast diagnostic operations enable the deployment of the proposed technique for performing extensive on-spot screening of COVID-19 in public places.

本研究报道一种无标记检测方法，可直接对鼻咽拭子样本中的新型冠状病毒（SARS-CoV-2）进行检测，无需纯化步骤与靶标扩增，该方法简化并加快了分析流程。该检测试剂盒由浸润液晶（LC）的纺织品基底组成，将其放置于交叉偏振显微镜系统中，随后将拭子样本置于液晶基底之上。当样本中存在特定生物分子时，液晶的排布方向会根据该生物分子的特性发生局部改变，并形成特异性光学图案。由于拭子样本未经过严格纯化，本研究采用图像处理与机器学习技术，以检测样本中特定分子的存在与否，或定量其在介质中的浓度。该方法区分新型冠状病毒肺炎（COVID-19）阳性与阴性样本的准确率可达96%，区分新型冠状病毒与甲型、乙型流感病毒的准确率可达93%。该试剂盒具备便携、易制备、操作便捷、成本效益高、检测快速且灵敏度高等优势。样本处理流程简便、图像采集速度快、诊断操作高效，使得本技术可部署于公共场所，开展大规模的新型冠状病毒肺炎现场筛查工作。