Concept and in-silico assessment of an algorithm for monitoring cytosolic fluorescent aggregates in cells.
Autophagy is an evolutionary conserved pathway, by which eukaryotic cells degrade long-living cellular proteins and intracellular organelles, to maintain a pool of available nutrients. Impaired autophagy has been associated to important pathophysiological conditions, and this is the reason why several techniques have been developed for its correct assessment and monitoring. Fluorescence microscopy is one of these tools, which relies on the detection of specific fluorescence changes of targeted GFP-based reporters in dot-like organelles in which autophagy is executed. Currently, several procedures exist to count and segment this punctate structures in the resulting fluorescence images, however, they are either based on subjective criteria, or no information is available related to them. Here we present the concept of an algorithm for a semi-automatic detection and segmentation in 2D fluorescence images of spot-like structures similar to those observed under induction of autophagy. By evaluating the algorithm on more than 20000 simulated images of cells containing a variable number of punctate structures of different sizes and different levels of applied noise, we demonstrate its high robustness of puncta detection, even on a high noise background. We further demonstrate this feature of our algorithm by testing it in experimental conditions of a high non-specific background signal. We conclude that our algorithm is a suitable tool to be tested in biologically-relevant contexts.