Last but definitely not least, comes the software. Even with the most perfectly optomized immunoassay and the highest resolution microscope, our brains would not be able to see a difference between a photograph collected from a water sample with norovirus versus one without. The published paper used MATLAB or ImageJ for the image processing, and Excel for data analysis.
The first step is to turn your images into usable data. The raw images basically all look the same. They are a green circle with some small pinpricks of light all over the FOV. The image processing code goes through each image in tiny square sections and uses adaptive thresholding to eliminate background intensities. The final results is a black and white image showing only the bright particles (the antibody-antigen clumps).
Once you get an image with all the bright particles showing, you turn this into a long list of particle areas. In our paper we used the average sum of particles within a certain size. This value is then compared to values we previously got from other experiments, to estimate the virus concentration.