Automatic Hotspots Detection for Intracellular Calcium Analysis in Fluorescence Microscopic Videos

In recent years, life-cell imaging techniques and their software applications have become powerful tools to investigate complex biological mechanisms such as calcium signalling. In this paper, we propose an automated framework to detect areas inside cells that show changes in their calcium concentration i.e. the regions of interests or hotspots, based on videos taken after loading living mouse cardiomyocytes with fluorescent calcium reporter dyes. The proposed system allows an objective and efficient analysis through the following four key stages: (1) Pre-processing to enhance video quality, (2) First level segmentation to detect candidate hotspots based on adaptive thresholding on the frame level, (3) Second-level segmentation to fuse and identify the best hotspots from the entire video by proposing the concept of calcium fluorescence hit-ratio, and (4) Extraction of the changes of calcium fluorescence over time per hotspot. From the extracted signals, different measurements are calculated such as maximum peak amplitude, area under the curve, peak frequency, and inter-spike interval of calcium changes. The system was tested using calcium imaging data collected from Heart muscle cells. The paper argues that the automated proposal offers biologists a tool to speed up the processing time and mitigate the consequences of inter-intra observer variability