Generating confocal data from ideal signal

illus_confFrom left to right : ground truth (projected), mock confocal image (projected) and thresholded pixels.

A large issue in image analysis is to make sure that the segmentation and analysis procedure does not create artifacts or misrepresents the data in any way.  If one knows the “ground truth”, one can easily check the behaviour of his analysis program…

… However, it can be very tricky to know the ground truth, especially for the very dynamic and variable biological samples. This is why I came up with an extremely simple tool to generate “mock” experimental data. More specifically, one starts from a ground truth which is a set of points – see it as fluorophore coordinates – and ends up with a confocal microscopy image.

For this, the ground truth is 3D-convolved (using gaussian3filter from Max W.K. Law) and pixelized, and one can include fluorophore stochasticity, background fluorescence, and pixel noise.

The pixel noise distribution and signal value can be extracted directly from a sample image given by the user, which makes this tool extremely easy to use !

The code is now available : https://github.com/SergeDmi/ConfocalGN

This entry was posted in Code, Image analysis, Simulations. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *