Proof of concept of an automatic tool for bioluminescence imaging data analysis

Bioluminescence Imaging (BLI) is an important molecular imaging tool to assess complex biological processes in vivo. BLI is a sensitive technique, which is frequently used in small-animal preclinical research, mainly in oncology and neurology. Tracking of labeled cells is one of the major applications. However, BLI data analysis for the segmentation of up-taking regions and their quantification is not trivial and it is usually an operator-dependent activity. In this work, a proof of concept of an automatic method to analyze BL images is presented which is based on a multi-step approach. Different segmentation algorithms (K-means, Gaussian Mixture Model (GMM), and GMM initialized by K-means) were evaluated and an adequate image normalization step was suggested to include the background bioluminescence in the data analysis process. K-means segmentation is the most stable and accurate approach for different levels of signal intensity.