Cancer Cells In Pleural Fluid Micro-images Processing And Recognition Method Research

With the traditional method of medical microscopic images processing, the doctors have to extract the features manually through the microscope, and draw the conclusion according to the stylebook. In this way, it requires a heavy workload, and the diagnoses rely on experiences of the doctors excessively.In order to improve the above problems, this paper applied the computer image processing technology, fuzzy pattern recognition and fuzzy clustering analysis to the processing of the cancer cells in pleural fluid micro-images. Namely, based on the edge detection of the original micro-image, the paper recognises and extracts figures form the microscopic images of cancer cells in pleural fluid by combining the morphological knowledge and fuzzy pattern recognition. Optimizing the partition of the cells sample space through fuzzy clustering analysis method, and providing some reliable and concrete parameters.The edge detection of cell images is the key of the micro-images recognition, and the edge detection is an emphasis of the paper. The conventional edge detection is not suitable to the cancer cells in pleural fluid because of the great uncertainty of shape. Even shapes of the cells are not uniform in the same image, generally. This article applies the fuzzy edge detection to extract the edge information due to the uncertainty of the image as a result of the fuzziness.According to the extraction of cells figures, based on the edge detection the paper extracts 9 measurement parameters such as centroid, area, area rate of karyon and cytoplasm, optic density, cells and karyon shape factor according to the morphological knowledge and the medical knowledge of pathology expert.Then based on the 9 extracted parameters, the paper applies fuzzy pattern recognition method to recognise the cancer cells in pleural fluid. Establishing 6 fuzzy sets such as cells with increscent karyon, cells with deeply dyed karyon, cells whose area rates of the karyon and cytoplasm are excessive and so on and 6 membership functions.The paper partitions the sample space furtherly into 4 nomarl pattern, and realizes the recognition of the cancer cells in pleural fluid.Finally, the paper applies the fuzzy clustering analysis method based on the objective function to improve the partition os the cancer cells in pleural fluid sample space, furtherly.