Font Size: a A A

Research Of Medical Image Classification Approach Based On Rough Sets And Association Rule

Posted on:2015-08-06Degree:MasterType:Thesis
Country:ChinaCandidate:S ChenFull Text:PDF
GTID:2298330422983965Subject:Computer technology
Abstract/Summary:PDF Full Text Request
With the rapid development of the computer network, we enter the digitalinformation age. All walks of our lives are informationized, which makes thesociety became a society of information explosion. Therefore, using a new type ofstrong data analysis tools to transform data into useful information and knowledgeautomatically and intelligently is badly needed. As the same as many other fields,biomedical also is under the influence of information science and technology. Theinformation sharing is not limited to the professional literature, which breakthroughthe limitation of time and space and has been extended to the open experiment andclinical database, experience, knowledge and so on. So we cannot deal with theamount of data by the traditional medical diagnosis.The development and application of data mining techniques can be appliedfor medical diagnosis, and improve the problem which traditional medicinediagnosis cannot be achieved. Data mining is very fit for disease diagnosis andtreatment of scientific decision, and can find the useful information and knowledgefrom the vast amounts of medical data. Therefore, in order to be better for thedecision of medical service and hospital management, researchers have attachedgreat importance to how to use data mining to look for patterns from a large amountof data the problem, and regarded as a very important research subject.Based on this background, in this paper, a new method is proposed, whichbased on rough set and association rule. First, Multiple reducts are generated byrough set, and weighted reduction is proposed to meansure the importance of therecucts. Second, The algorithm method, Association rule is used to generate a set ofrules for each reducts. Meanwhile, A new concept, rules importance is proposed tosort the rules, and a new data will be classified by matching the rules having beensorted. Due to the particularity of medical data, a reduction will miss someimportant information. So this article use multiple attributes reduction to handle. Inthe classification phase, sort of the rules can be used to match and classify more accurately and rapidly for the new data. The algorithm was applied to MIAS, whichis the standard data sets of Mammograms. The experimental results show that theefficiency and time of this algorithm which is applied to Mammogramsclassification is obviously better than the Apriori algorithm and single reductionalgorithm based on rough set, and it is a kind of effective and fast algorithm ofassociation rules.
Keywords/Search Tags:Data mining, Rough sets, Association rules, Mammograms
PDF Full Text Request
Related items