Fuzzy logic based detection of neuron bifurcations in microscopy images

Quantitative analysis of neuronal cell morphology from microscopic image data requires accurate reconstruction of the axonal and dendritic trees. The most critical points to be detected in this process are the bifurcations. Here we present a new method for fully automatic detection of bifurcations in microscopic images. The proposed method models the essential characteristics of bifurcations and employs fuzzy rule based reasoning to decide whether the extracted image features indicate the presence of a bifurcation. Algorithm tests on synthetic image data show high noise immunity and experiments with real fluorescence microscopy data exhibit average recall and precision of 90.4% and 90.5% respectively.