Automatic Cell Counting From Microchannel Images

Cell counting is used in many fields such as disease diagnosis in medicine, and there are manual and automatic methods developed for this purpose. Because manual methods are slow and error-prone, automated cell counting methods such as flow cytometry have been developed. Flow cytometry methods are based on analyzing cells one by one by passing the cell-containing fluid through a microchannel. These methods allow obtaining cell sizes in addition to automatic cell counting. In this study, an image processing and object tracking based method for automated cell counting and analysis is presented. This method is based on recording images with a camera as the cells are passed through the microchannel. Cell counting is done by automatic detection and tracking of cells over recorded images. Image processing methods such as filtering, background modeling, segmentation, opening and closing are used for the detection of cells, and Kalman filter and Hungarian assignment algorithm are used for tracking. Cell analysis includes the average velocity of cells in the microchannel, cell sizes, and classification of cells by K-medoids method. Experiment results showed that the proposed method counts cells accurately under appropriate experimental conditions and can make size-based classification for cells having different sizes, therefore it can be used as a direct or reference method in cell experiments.