Characterization of Breast Tumors with NIR Methods using Optical Indices

This work describes the characterization efficiency of optical properties of breast tumors based on the features obtained using in vivo near-infrared (NIR) spectroscopy measurements. Three features, relative blood concentration, oxygen saturation and the size of the tumor, are used to diagnose benign and malignant tumors. The performance of the proposed set of features are evaluated by various classifiers using data acquired from 44 patients with malignant tumors, and 72 patients with benign tumors. The area under the receiver operating characteristics (ROC) curve of the scaled nearest mean classifier (NMSC) using the three features yields a value of 0.91 with a significance level of 0.05. Our results suggest that the features, relative blood concentration, and oxygen saturation can differentiate breast tumors with a relatively high precision.