Online Program

Detecting Contamination in Cell Cultures with ECIS Time Series Data (309566)

Keywords: time series,classification,multivariate time series,cell cultures,biophysics

An ongoing problem in research on cells in culture is the possibility that cells have been mislabeled or contaminated with other organisms. We explore an approach to detecting such contamination automatically using ECIS (Electric Cell-substrate Impedance Sensing) technology. ECIS is performed by passing an alternating current through a cell culture grown on an electrode, and measuring the impedance (and its components resistance and capacitance) across the culture. Repeating these measurements over time, and with different AC frequencies, creates a multivariate time series that reflects the growth, spreading, and confluence behavior of the cells. We show that mammalian cells infected with a mycoplasma organism do indeed display different ECIS output than uninfected cells, and that the distinction can be seen using simple, interpretable features based on the time series. Since laboratory conditions and natural variation can also result in differences in ECIS output, we also explore features that may be more robust to this variation. We examine additional cell behaviors, such as recovery from wounding, as well as more sophisticated time series methods.