The Recognition Technology Research For Cancer Cells In Microscopy Images Of Lung Tissue

The human is more and more intelligent in dealing with medical image processing,and its application in clinical diagnosis has become a hot research topic in the medicaland computer fields. Along with enhancement of diagnostic methods and deteriorationof the living environment, the causes of cancer is also changing, diagnosis and treatmentof diseases are facing many new challenges. In the clinical field, it is an especiallyimportant method for diagnosing lung diseases through analyzing lung tissuemicroscopy images.Ontology is derived from the field of artificial intelligence, which is used torepresent knowledge. In the field of computer, it refers to conceptual analysis andmodeling by the principle of ontology, which can abstract the things of objective worldinto a set of concepts and some theories and methods of the relationship amongconcepts. In a way, the ontology can achieve the sharing of knowledge, so that thesemantics of understanding in different areas.Compared with other traditional methods, fuzzy pattern recognition has its uniquecharacteristics. First, the identify standard is no definite measure, and fuzzyidentification can simulate the human thought process to some extent and need not veryclear numerical discrimination conditions. Second, it is an effective complement to thestatistical and syntax methods, which is built on the basis of the theory of fuzzymathematics. In short, the fuzzy recognition can simulate human brain analysis process,and allows recognizing objects containing interferential information. At the same time,it does not need a large number of samples.Based on the cancer cells of lung tissue in the microscope image, this paper, whichhas combined image processing, image recognition, ontology modeling, fuzzy theory toa process of image recognition, has simplified the process of image recognition andadvanced the idea of object-oriented image recognition. Then it has a certain referencevalue and significance in practice and theory. In this paper, the followings are my works:1) Learned the basics of image processing and programmed to achieve somealgorithms.2) By morphological edge detection and watershed algorithm to extract theboundary part of the cells in the target area, then calculated the circumference, area,elongation, of the cell region, round degree, the proportion of blue value in RGB. According to the extracted boundary, we have selected a reasonable threshold rangeafter a large number of experiments, which is based on the extraction of themorphological characteristics. In this way, we have judged preliminarily the cancerregions of the images.3) By the closed edge of digital information, which is extracted, I have proposed anobject-oriented image recognition method. And then, designed the specific steps of thismethod.4) Imported this method to the recognition of the microscope image of the lungtissue, and analyzed the texture features in extracted gray-scale image of lung tissue,which are secondary moment, entropy, contrast, and correlation of gray level co-occurrence matrix.5) By the texture features extracted, calculate the weighted Euclidean distance ofthe image, which need to recognize, and the texture characteristics of lung grayscaleimage in the feature library, whereby to determine which type of lung cancer.