Identification of leader cells by filopodia in collective cell migration using computer vision.

Abstract

BACKGROUND Imaging of cells and cellular organelles has been of great interest among researchers and medical staff because it can provide useful information on cell physiology and pathology. Many researches related to collective cell migration have been established and leader cells seem to be the ones that regulate the migration, however, the identification of leader cells is very time-consuming. OBJECTIVE This study utilized computer vision with deep learning to segment cell shape and to identify leader cells through filopodia. METHODS Healthy Madin-Darby Canine Kidney (MDCK) cells cultured in a Polydimethylsiloxane (PDMS) microchannel device allowed collective cell migration as well as the formation of leader cells. The cells were stained, and cell images were captured to train the computer using UNet++ together with their corresponding masks created using Photoshop for automated cell segmentation. Lastly, cell shape and filopodia were filtered out using Filopodyan and FiloQuant were detected. RESULTS The segmentation of cell shape and the identification of filopodia were successful and produced accurate results in less than one second per image. CONCLUSIONS The proposed approach of image analysis would be a great help in the field of cell science, engineering, and diagnosis.