Low-latency label-free image-activated cell sorting using fast deep learning and AI inferencing.

Classification and sorting of cells using image-activated cell sorting (IACS) systems can bring significant insight to biomedical sciences. Incorporating deep learning algorithms into IACS enables cell classification and isolation based on complex and human-vision uninterpretable morphological features within a heterogeneous cell population. However, the limited capabilities and complicated implementation of deep learning-assisted IACS systems reported to date hinder the adoption of the systems for a wide range of biomedical research. Here, we present image-activated cell sorting by applying fast deep learning algorithms to conduct cell sorting without labeling. The overall sorting latency, including signal processing and AI inferencing, is less than 3 ms, and the training time for the deep learning model is less than 30 min with a training dataset of 20,000 images. Both values set the record for IACS with sorting by AI inference. . We demonstrated our system performance through a 2-part polystyrene beads sorting experiment with 96.6% sorting purity, and a 3-part human leukocytes sorting experiment with 89.05% sorting purity for monocytes, 92.00% sorting purity for lymphocytes, and 98.24% sorting purity for granulocytes. The above performance was achieved with simple hardware containing only 1 FPGA, 1 PC and GPU, as a result of an optimized custom CNN UNet and efficient use of computing power. The system provides a compact, sterile, low-cost, label-free, and low-latency cell sorting solution based on real-time AI inferencing and fast training of the deep learning model.

High efficiency sorting and outgrowth for single-cell cloning of mammalian cell lines.

Single-cell selection and cloning is required for multiple bioprocessing and cell engineering workflows. Dispensing efficiency and outgrowth were optimized for multiple common suspension (CHO ES, Expi293F, and Jurkat) and adherent (MCF-7, A549, CHO-K1, and HEK293) cell lines. Single-cell sorting using a low pressure microfluidic cell sorter, the WOLF Cell Sorter, was compared with limiting dilution at 0.5 cells/well to demonstrate the increased efficiency of using flow cytometry selection of cells. In this work, there was an average single cell deposition on Day 0 of 89.1% across all the cell lines tested compared to 41.2% when using limiting dilution. After growth for 14 days, 66.7% of single-cell clones sorted with the WOLF Cell Sorter survived and only 23.8% when using limiting dilution. Using the WOLF Cell Sorter for cell line development results in higher viable single-cell colonies and the ability to select subpopulations of single-cells using multiple parameters.