This repository covers the efforts to create a cell counting model that can effectively predict the number of DAPI stained cells in a Z-stack of images taken with a fluorescence microscope. The model consists of three components:

This component takes in each image of the Z-stack and returns a black and white mask image.

This component takes the produced mask image as an input and returns the number of cells as an Integer, and the cell locations as an array of Integers.

This component takes the locations of the cells in each image in the Z-stack and resolves the cell count in the complete volume. It does so by detecting count artifacts and using a recurrent setup where information about the cell locations from the previous slice are used to validate the cell locations in the next slice. If a particular location is present in multiple slices its chance of becoming a true cell increases.