The Development Of Cell Image Analysis And Processing System

During the Life Science Research, the analysis of massive two-dimensional images, including how to precisely deal with and express the useful information, has become the main focus of many scientists. In the past, observable/recordable cellular information was limited with number and length through microscope. However, the information on area, shape, grey scale and complexity were hardly measurable. Meanwhile, some experiments may cost massive time and manpower due to large cell/sample number, and manual processing error is unavoidable. Therefore, a rather complete set of cell image processing software is required in order to increase the automation level of the experiments, achieve quantitative detection, ensure accuracy, improve work efficiency, and reduce experiment cost.In this research, a cell image analysis software is developed based existing research laboratories demand. Detailed work consists of the following aspects:1. Based on the image processing performance requirements and the system scalability, main structural framework for image processing system is designed, core data conversion and storage modules are established, and the common interface for image data input/output is built.2. Basic cellular image calculation and operation modules are constructed, in order to provide rudirnental functions such as color converting, space converting, scaling, rotating and moving. And ROI tools are developed to achieve free regional selection and analytical statistics.3. Enhancement module is constructed to solve the problems including inadequate contrast, low brightness of target image, fuzzy edge and fuzzy details, to improve visual effects, and to highlight specific information. For example, increase grey scale to increase the contrast between cells and background, smoothen image to remove noise, use histogram to statistically analyze and adjust the grey scale, enhance image to carry out linear grey-scale image detection, grey lattice and distribution based on the simple grey area.4. Modules for processing mathematical morphology and cell image segmentation are built. Due to different cellular type, shape and grey scale in the laboratory research, a common method for area measurement is proposed, in which the area of cell is measured precisely through noise reduction, cell edge detection, Otsu Act threshold segmentation, hole-filling, iterative corrosion, feature extraction and seed filling. 5. Cell image matching module is formed to achieve cell count under the condition of complex background.6. The feasibility of the system is verified through varies tests on grey-scale measurement of fluorescent cells, cell measurement, cell count, etc.Furthermore, this study presents the improvement of the existing ROI tools for users' convenience to a great extent. Moreover, through cell image analysis, an effective method to eliminate the noise is found, relatively correct segmentation results are gained, segmentation ability to the images of overlapping, adhesive cells and complicated background. In addition, image matching method is proposed to count cells. This method is an improved version from the existing approach, in which a nine-regional segmentation method is applied to enhance the matching operation efficiency. Meanwhile, matching strategy can be adjusted by users in order to make exact estimation under various measure conditions.