A Segmentation Based on Feature for LSCM Sequence Images

Region of interest (ROI) plays an important role in biomedical image analysis. The calcium ions laser scanning confocal microscope (LSCM) real-time optical sections of fluo-4 labeled rat ventricular myocyte show point distribution and are interfered by random noise. The fluorescent images have no continuous grayscale distribution and obvious edge. It is difficult to apply the methods which process static fluorescent images to our experiment. A self-adaptive segmentation is proposed based on spatial point pattern cluster features comparability in Markov field. The advantage of this method is that the ROI of calcium within all the disturbed optical sections can be properly segmented under a coincident standard. This method will benefit to other dynamic analysis researches using LSCM.