LIVING CELLS STATE TRAJECTORY IN TIME-LAPSE MICROSCOPY

Tissue cells exhibit fascinating properties: they synthetize a huge variety of chemicals, gather and process information about their surroundings, communicate, move across environments or change their biochemical composition. Advances in microscopy and culturing techniques made possible to see the cells in action. The most visually interesting events in the life of a cell occur when it divides. Imaging of living cells and tissue is now common in many fields of the life and physical sciences. Time-lapse microscopy makes link of initial cell cycle position with final. We extracted the region of the cell in order to analyze information about the cell state from spatial structure of observed objects. For maximization of information gain we calculated the point information gain entropy density (PIE/points, Stys et al. 2011). We determine the information contribution of each data point to the object by calculation of difference in Renyi entropy between datasets containing and excluding the examined data point. This procedure we perform for representative set of alpha coefficients and obtain a set of point information gains (PIG). Sum of PIG values for certain alpha is the appropriate PIE/points. Set of PIE/points is a point in the tentative phase space which is unique for each image. Using statistical procedure of Principal Component Analysis (PCA) and Clustering Analysis we were able to separate the state space in regions occupied by similar images (Stys et al. 2012). The sequence of regions is an objectively determined cell state trajectory, which are stable for certain period of time. For the cell cycle, we found that regions in the cell state space may be identified with known terminology, i.e. we were able to find cell state corresponding to each of the clusters. Such detailed analysis is extremely computationally intensive; however, it might be of high value for rapid diagnostics in medicine, biotechnology and any other discipline utilizing cell biology results.