Quantitative Analysis Of Dynamic Character For Fluorescence Microscope Image Sequence And Identification Of Fusion Events

Technical advances in bright fluorescent probes and novel microscopy techniques have revolutionized the way subcellular dynamics are studied. Analyze the dynamic behavior of cell physiology and pathology after real-time recording of fluorescent intracellular substance has become a routine. Studies of subcellular dynamics generate ever-increasing amounts of image data. Handling these data and capturing the full spatiotemporal behavior on a quantitative level and understanding the underlying biological processes require many sequence image processing algorithm.We propose a quantitative dynamic character analysis framework for fluorescence microscope image sequence. The framework consists of three levels:the overall dynamic performance evaluation, detailed motion analysis based on particle tracking and dynamic event identification. We have also discussed the advantages, disadvantages and scopes of application of each method, and provided a guildline for how to choose the most appropriate method for certain application.The overall dynamic performance evaluation includes two methods:colocalization dynamic evaluation and difference image dynamic evaluation. In which the Difference of Gaussians template is applied to detect blob feature. Also we propose a fully automated mitochondrial segmentation algorithm. First we compute the local contrast value of each pixel. Then we use minimum cross entropy thresholding to get the final result. This algorithm does not require any parameter, and can automatically adapt to the uneven background of the fluorescence image. The segmentation results are smoother and better than traditional vessel segmentation algorithm, especially in the low contrast area. These dynamic evaluation methods were used to analyze the dynamic of GLUT4 vesicles in skeleton muscle cells and movement of mitochondria in Aβ25-35 induced PC 12 cell.For detailed motion analysis, we propose a particle tracking algorithm based on multidimensional assignment. We combine an Interacting Multiple Model (IMM) filter, multidimensional assignment, particle occlusion handling, and merge-split event detection in a single software analysis package. The main advantage of a multidimensional assignment is that both spatial and temporal information can be used by using several later frames as reference. The IMM filter, which is used to maintain and predict the state of each track, contains several models which correspond to different types of biologically realistic movements. It works especially well with multidimensional assignment, because there tends to be a higher probability of correct particle association over time. First the method generates many particle-correspondence hypotheses, merge-split hypotheses and misdetection hypotheses within the framework of a sliding window over the frames of the image sequence. Then it builds a multidimensional assignment problem (MAP) accordingly. The particle is tracked with gap-filling, and merging and splitting events are then detected using the MAP solution. The tracking method is validated on both simulated tracks and microscopy image sequences. The results of these experiments show that the method is more accurate and robust than other "tracking from detected features" methods in dense particle situations. The particle tracking method was used to analyze the dynamic behavior of GLUT4 vesicles in primary rat adipocyte and autophagic vacuoles in degenerating neurites.To automatically the fusion events from image sequence, we propose a running average difference method to remove the unrelated image background while preserving the character of fusion events, which is the increased intensity at the beginning of vesicle fusion. This background remove ability of this algorithm is better than method based on template matching. Also it is more suitable for detecting fusion with different styles. Test results show that the detection rate is better than template matching method. This method was used to analyze the fusion events of skeleton muscle cell under both basal and insulin stimulation state.Collectively, our present study achieved the following novel results:1) Fully automated mitochondrial segmentation algorithm, which can be adapted to image with varied contrast, for difference image dynamic evaluation; 2) Multipe dense particles tracking algrithm, which combines IMM filter, multidimensional assignment, particle occlusion handling, and merge-split event detection in a single tracking framework. The NP-hard multidimensional assignment problem were effectively solved by transforming it to general integer linear programming; 3) By using running average difference method, the unrelated image background was removed and the fusion events were automatically detected. The proposed dynamic analysis framework laid a solid foundation for biological mechanism research. It is suited for high-throughput image-based analysis and has a wide range of applications.