Counting cells is one of the most important and common applications in biology. At the same time, this is a highly challenging task if the cells are not fluorescently labelled or stained. Therefore, classical image processing methods can often not be used for brightfield images.
This solution is based on a pre-trained deep-learning network that detects the cells in one or more bright field microscopy images. Time series images as well as multi-well images are supported. The output is a heatmap that shows the number of cells per image in each well and time point (if applicable). The respective deep-learning network is trained on datasets from multiple microscopes with different resolutions/magnifications.