Research Of The Application Of Data Mining In Computer Aided Diagnosis

These days,Computer Aided Diagnosis(CAD) is gradually becoming the focus in the study of the medicine area.The appearance and development of many CAD technologies is helpful to improving diagnostic precision and reducing diagnostic mistakes for clinic physicians.Data mining technique provides more development for CAD.In this paper,we discuss and study the application of CAD in laboratory medicine based on data mining.CAD acquires more important progress in medical imaging,but relatively less in laboratory medicine.In this paper,we study the application of CAD in laboratory medicine based on decision tree and fuzzy clustering analysis,and get the general application model.Blood corpuscle counting test has diagnostic significance for some diseases.This paper introduces mostly how to mine the correlative data using decision tree,form diagnostic model and use for CAD,so that it can provide the effective diagnostic information for clinic diagnosis.We use cupidity algorithm to stipulate dimension,which is to find useful property from beginning.We use entropy increment technology to find related properties.Decision tree is one of the applications of dada mining.In this paper,we first expatiate ID3 algorithm and C4.5 algorithm,then adopt classical ID3 algorithm to stipulate dimension for the diagnostic criterion of anemia based on blood corpuscle counting.A decision tree model can be got in this way,and we use the model in computer for aided diagnosis.Stomach disease is one of the diseases which threaten people's health.It depends on physician knowledge and clinic experience to diagnose stomach diseases in the past.In this paper,we provide a method of fuzzy clustering analysis with combination of stepwise discriminatory analysis for the CAD of stomach diseases.We first introduce knowledge about fuzzy clustering analysis and stepwise discriminatory analysis,and found the medical aided diagnosis model based on it. According to the analyses of 156 case of stomach cancer and the other stomach diseases,which are collected from the digesting department of the hospital,we construct fuzzy similar matrix and adopt the rightλ.Then we use stepwise discriminatory analysis model in the meaning of Bayes based on fuzzy clustering analysis,and keep significant factors in discriminate function and take out of the not significant ones.We use the model to predict and diagnose the stomach diseases.It produces appropriate diagnosis about 96.0%in the retrospective test of the 100 cases, and produces appropriate diagnosis about 91.1%in the following practical use of the 56 cases.The model is designed in VB program language,which realizes the CAD on stomach diseases.A application in laboratory medicine of CAD based on data mining is now on the initial stage,and we only discuss two methods of data mining in this paper.The other technique of data mining can also be applied in CAD,so we will keep on doing in the following work.