Computer-Aided Diagnosis Applied to 3-D US of Solid Breast Nodules by Using Principal Component Analysis and Image...

Textural features have been shown to be valuable in tumor diagnosis. This study combines three practical textural features in ultrasound (US) images, i.e. spatial gray-level dependence matrices (SGLDMs), gray-level difference matrix (GLDM) and auto-covariance matrix, to identify breast tumor as benign or malignant. The textural features were extracted from 147 3-D ultrasound cases and each case composes a volume of interest (VOI). Usually, the larger region of interest (ROI) sub-image contains considerable textural information. Thus the feature vector extraction utilizes only the adjacent frames with the largest ROI sub-image. The textural features always perform as a high dimensional vector that is unfavorable to differentiate breast tumors in practice. The principal component analysis (PCA) is used to reduce the dimension of textural 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 imaging and helps inexperienced operators avoid misdiagnosis.