Research On Classification Of Mammograms Based On GEP-DT

Thanks to the rapid growth of technology and economic, people’s living standard has beensignificantly improved, but it also brings a large amount of issues, such as serious pollution,rising cancer rates, and so on. Breast cancer which ranks first on the list of women cancerseriously affects the health of women friends.” Early detection and early treatment” is theprimary means of improving breast cancer survival. Currently, the therapy is mainly taken, first,images are obtained by devices, e.g, X-rays, MRI, CT etc, and then doctors observe and identifywith the eyes, diagnosis illness. However, due to the skills of doctors, fatigue, and other factors,it often results in a high rate of misdiagnosis. Computer technology and the image processinghave achieved great development in rent years, information is popularized in hospital so thatmedical images can be accessed and managed. But, faced with such massive data, how toexplore the useful information, assist physician’s diagnosis, raise the diagnostic accuracy andworking efficiency. In view of this, Data Mining is applied to medical diagnosis,so data miningon medical image emerges. Many achievements on medical image data mining have been madeat home and abroad, however, owing to the complex characteristics, high-resolution,heterogeneous of medical image, we have yet got a satisfactory solution. So, data mining onmedical image still has great practical significance and broad prospects.From the perspective of data mining, this article focuses on the algorithm of decision treeand gene expression programming, proposes the Integrated Decision Trees Algorithm based onGEP (GEP-DT algorithm), and applies GEP-DT algorithm to diagnosis of breast X-ray image,constructs a breast X-ray image classifier based on GEP-DT. In order to raise the diagnosis rate,decrease the misdiagnosis and missed diagnosis rate.The main research work of this paper is as follows:(1)This article analyzed the breast cancer’s situation at the present stage, emphasized theimportance of the construction of CAD system, and introduced the knowledge of breast cancer,including the performance of breast lesions and related imaging technology screening tool,introduced the research status of breast X-ray image at home and abroad on image processingand mining algorithms.(2)Systematically expounded the theory knowledge of decision tree and gene expressionprogramming, analyzed the respective merits and drawbacks, according to integratethinking,combined with both advantages, this paper proposed the Integrated Decision TreesAlgorithm based on GEP(GEP-DT algorithm), the proposed GEP-DT algorithm was proved tohave a good effect through the experiment. (3)The GEP-DT algorithm was used in classification of Mammograms and a breast X-rayimage classifier based on GEP-DT was constructed. And focusing on the characteristics of X-rayimages of the breast, this paper analyzed the associated image processing technology, forinstance: image preprocessing (gray normalization, image denoising, image enhancement),image segmentation, feature extraction, etc. and got the image processing algorithms specific tobreast X-ray images. Finally, through the simulation experiments on the MIAS dataset,comparing classifiers’ accuracy, sensitivity, specificity and ROC, the proposed GEP-DTclassifier was proved to have good performance.