A post-processing method for EISI to distinguish breast lesions

To increase the accuracy of Electrical Impedance Scanning Imaging (EISI) on inspecting breast diseases, reduce the false positive and false negative, we propose a method to use Independent Component Analysis (ICA) to extract features from the original admittance data of EISI equipments. Combined with Support Vector Machine (SVM) as classifier, the proposed method could inspect breast diseases and give nonuser-dependent diagnosis results. Experimental results suggest that the features extracted by ICA are effective in classifying breast diseases. To recognize lesions within breast under inspect which is malignant or benign, the proposed method could achieve a sensitivity of 74.2% and a specificity of 82.8%. Accuracy is calculated as 80%. Combined the proposed method with regular EISI inspecting process could help to increase the reliability of EISI, avert benign biopsies.