High-content 2D light scattering flow cytometry for label-free classification of cervical carcinoma cells with deep learning

Two-dimensional (2D) light scattering has the capability for label-free single cell analysis. Recent development of flow cytometry has demonstrated the obtaining of high-content images. Here we demonstrate a flow cytometer for the obtaining of high-content 2D light scattering patterns of single cells. In our flow cytometer, single cells are flowing in a hydrodynamic focusing unit and their 2D light scattering patterns are recorded via a long working distance objective by using a high-speed complementary metal oxide semiconductor (CMOS) sensor. Big data of the 2D light scattering patterns from two types of cervical carcinoma cell lineage cells (HeLa and C33-A) are obtained with a rate of 60 frames per second. Deep learning is adopted for the classification of these two types of cells, and a high recognition accuracy is obtained. The results show that our high-content 2D light scattering flow cytometry together with deep learning can collect label-free single-cell information at high speed and has strong analytical capabilities, which may in future be used for early diagnosis of cervical carcinoma.