Study on Artificial Immune System Algorithm Applying to Ultrasound Breast Tumor Image Diagnosis

Because breast cancer ranks in the causes of cancer deaths among women are always to occupy a high position, the early diagnosis and cure is getting more and more important. However, early diagnosis requires an accurate and reliable diagnostic procedure that allows physicians to distinguish benign breast tumors from malignant ones. Hence, we need a well-designed CAD system to assist the inexperience physicians to avoid misdiagnosis and reduce the number of benign lesion biopsies without missing cancers. The accuracy of a CAD system is related to the data mining techniques. Data mining techniques have been used in classification problems. The decision trees (DT), artificial neural network (ANN), and support vector machine (SVM) are popular among them and can be applied to various areas. Nevertheless, different problems may require different parameters when applying DT, ANN and SVM. If parameters didn’t setting well, it will obtain unsatisfied result. On the other hand, a dataset may contain many features; however, not all features are beneficial for classification. The features may contain false correlations, which hinder the process of detecting process. Without feature selection, the classification accuracy rate may be worse due to the noises or too many dirty features. In most research either parameters turning or feature selection is used to improve classification accuracy rate. Therefore, the artificial immune system algorithm (AIS) is proposed to select the beneficial subset of features and to obtain the better parameters of DT, ANN and SVM, which will result in a better classification. The above data mining techniques has its own advantages and disadvantages and the suitability will influenced by the characteristic of problem. If these techniques can work together, it is expected that the better result can be obtained. This is so-called ensemble architecture. Therefore, this proposal is plan to use the ensemble architecture to further enhance the classification accuracy rate.

Project ID:PB9907-10780
External Project ID:NSC99-2221-E182-041