The Design And Implementation Of Dynamic Library For The Automatic Detection Of WBC From Human Stool Based On Machine Vision

Human stool are the main secretions, about 75% of normal human stool is water, the other is solids. And the solids components mainly are of undigested food, digestive secretions and physical components, food decomposition products and bacteria. The stool of healthy human generally don’t have white blood cells or occasionally have some WBC(white blood cells). And when it has WBC, the WBC are neutrophils. There was no difference for the WBC having a integrated shape between in stool and in blood. If detected the change of abnormal shape and the number of white blood cells, it implies human body’s inflammation, the number and the shape of WBC in stool can be used as a judgment condition to judge in advance whether there is human body disease. And it also can provide direction for further testing and physician clues. Medical stool detection is very useful especially for the early diagnosis of gastrointestinal leukocytes and other diseases.Hospital stool WBC detection are artificial microscopy identification at the present stage, it is low efficiency, qualitative analysis and subjective factors, leading to can’t timely in the early detection of disease. At the same time, the large amount of Hospital stool WBC detection and the shortfall of related resources cause the nervous relationship of doctor-patient, bringing adverse social benefits. By expanding automatic detection technology research of WBC in stool, it can solve the problem of the nervous doctor-patient relationship, the bad detection precision and so on, having great academic and social value. And the development trend of intelligent medical equipment is also accord with modern medical, related fields has good prospects.Although the stool WBC automatic detection has significant value, but because the target pattern background is extremely complex and the shape of white blood cell is varied, it makes the stool WBC quantitative automatic detection is very difficult. Aiming at these problems, This thesis take research on the study of the following several aspects, and get good results, achieving the goal of medical practical use.a) This thesis statistically analyzes the characteristics of stool WBC, and it studies the related features. Through consulting relevant literature, analyzing the morphology characteristics of sample pattern of WBC, this thesis get the characteristics information needed to obtain design algorithm, this thesis has analyzed 5000 samples of stool WBC, and each of sample has 5 pictures of low power lens,15 pictures of high power lens.b) This thesis statistically analyzes the characteristics of background characteristics of the picture of stool WBC samples, and the characteristics of RBC(red blood cells) and other cells in the stool in order to correctly detect WBC in the complex background, red blood cells and to avoid interference with the WBC detection,avoid the interference from RBC or other cells. By these work, this thesis get the judgment information for designing WBC recognition algorithm, this thesis has analyzed 5000 samples of stool WBC, and each of sample has 5 pictures of low power lens,15 pictures of high power lens.c) This thesis designs the stool WBC quantitative automatic detection algorithm based on the white blood cell characteristics, stool samples pattern background information and the feature information of other cells in stool. When actually used in the hospital, it realized the target(masculine sensitivity and specificity were higher than 90%)in hospital stool WBC detection.d)this thesis develops a dynamic library for the automatic detection of WBC from human stool based on machine vision, and actually meet the use of related medical equipment in hospital.