A new detection algorithm for image analysis of single, fluorescence-labeled proteins in living cells.

A new algorithm is presented for the detection of single, fluorescence-labeled proteins in the analysis of images from living cells. It is especially suited for images with just a few (<1 per 10 microm2) fluorescence peaks from individual proteins with high background and noise (signal to background ratios as low as 2 and signal to noise as low as 10). The analysis uses the peaks over threshold method from extreme value theory and requires minimal assumptions on the underlying distributions. The significant advantage of the method over others is the rare occurrence to detect false positives. Some examples of simulated and real data are given as comparisons. The algorithm is implemented in MATLAB.