Computer-Aided Diagnosis Applied to US of Solid Breast Nodules by Using Principal Component Analysis and Image Retrieval

Abstract

This paper combines three useful textural features of ultrasound (US) images, i.e. block difference of inverse probabilities (BDIP), block variation of local correlation coefficients (BVLC) and auto-covariance matrix, to classify benign and malignant breast tumors. 1020 sonograms of region of interest (ROI) from 255 patients were used as case samples. Two-view sonogram (longitudinal and transverse view) and four different rectangular regions are utilized for each tumor analysis. The textural features always perform as a high dimensional vector. High dimensional vector is unfavorable to differentiate breast tumors in practice. The principal component analysis (PCA) is used to reduce the dimension of textual feature vector and then the image retrieval technique was utilized to differentiate between benign and malignant tumors. The proposed computer-aided diagnosis (CAD) system differentiates solid breast nodules with a relatively high accuracy in the US system and helps inexperienced operators avoid misdiagnosis.