Mathematical Modelling for Automatic Cell Contractions Detection and Their Directions in Artificially Grown Human Cardiomyocytes

Mathematical models can be used in diversified problems. They are consumed for precise simulations (e.g., in digital twins) but also can be aligned with slightly different ideas - for example to predict directions and movements of the objects in videos or for pose estimation. In this work, the authors presented their own approach for evaluation of direction and the moment of contraction in artificially grown human cardiomyocytes (2D models based on microscopic images). Observation of these two parameters (and their variability) is especially important in the case of cardiomyocyte behavior analysis after application of new medication (it needs to be done real-time and analysis has to be performed constantly to detect the potentially dangerous changes). During the experiments the authors consumed their own dataset (50 videos, each no shorter than 25 s collected by Institute of Human Genetics, Polish Academy of Science - prof Tomasz Kolanowski Team) and used not only mathematical modelling but also image processing and analysis to obtain needed information. The results have shown that the proposed processing and analysis pipelines are precise and fast and can be practically used in real medical environment.