Single-Cell Radiation Response Scoring with the Deep Learning Algorithm CeCILE 2.0

Abstract

External stressors, such as ionizing radiation, have massive effects on life, survival, and the ability of mammalian cells to divide. Different types of radiation have different effects. In order to understand these in detail and the underlying mechanisms, it is essential to study the radiation response of each cell. This allows abnormalities to be characterized and laws to be derived. Tracking individual cells over several generations of division generates large amounts of data that can no longer be meaningfully analyzed by hand. In this study, we present a deep-learning-based algorithm, CeCILE (Cell classification and in vitro lifecycle evaluation) 2.0, that can localize, classify, and track cells in live cell phase-contrast videos. This allows conclusions to be drawn about the viability of the cells, the cell cycle, cell survival, and the influence of X-ray radiation on these. Furthermore, radiation-specific abnormalities during division could be characterized. In summary, CeCILE 2.0 is a powerful tool to characterize and quantify the cellular response to external stressors such as radiation and to put individual responses into a larger context. To the authors knowledge, this is the first algorithm with a fully integrated workflow that is able to do comprehensive single-cell and cell composite analysis, allowing them to draw conclusions on cellular radiation response.