A novel method for counting subcellular structures labeled by green fluorescent protein

The direct observation and research on the subcellular structure is greatly improved by the advances of digital imaging devices and green fluorescent protein (GFP) labeling techniques at the molecular level. However, the digital imaging devices generate huge amount of data, which makes it an urgent task to develop fast, accurate and automatic data processing technologies for the biomedical research. In this paper, we propose a novel method for counting the number of subcellular structures labeled by GFP. First, the original GFP image is segmented into foreground and background by the bi-modal Gaussian distribution modeling. Then, the gray-scale reconstructive opening operation is applied to find the local maxima of the image. Finally, the number of the subcellular structures is counted by the connected component analysis on the local maxima. The proposed method is compared with other existing methods and the experimental results demonstrate that our approach is promising.