Breast tumor tissues classification using the modified cole-cole parameters with machine learning technique

摘要

Microwave techniques have been widely studied for assisting breast health diagnosis and treatment. The basis for introducing the microwave is that the dielectric properties of the tumor and healthy tissues are different. In order to understand the complex permittivities of the breast tissues comprehensively, we have conducted electromagnetic measurements on 296 excised breast tissue samples from 105 patients. In our previous work, the modified Cole-Cole model was proposed for fitting the measurement data. Here, a preliminary analysis of the model parameter distribution and the potential of using the parameters for breast tumor tissues classification are investigated. The support vector machine (SVM) technique is employed. The classification accuracy of 87.2% between tumorous and healthy tissues is achieved using a combination of 4 parameters.