Gaussian mixtures for intensity modeling of spots in microscopy

In confocal microscopy imaging, the target objects are labeled with fluorescent markers in the living specimen, and usually appear as spots in the observed images. Spot detection and analysis is an important task for the biological studies from the observed images. However, while the spots have irregular sizes and positions due to the variant amount of objects on each spot, the quantitative interpretation of the labeled objects is still heavily reliant on manual evaluation. In this paper, a novel shape modeling algorithm is proposed for automating the detection and analysis of the spots of interest. The algorithm exploits a Gaussian mixture model to characterize the spatial intensity distribution of the spots, and optimizes the model parameters using split-and-merge expectation maximization (SMEM) algorithm. As a result, a large amount of target objects with uncertain shapes can be analyzed in a systematic way.