Tissue classification by texture and spectral analysis of intracoronary ultrasound radio-frequency data

Imaging of vascular structures by intracoronary ultrasound allows in principal the recognition of different lesion types due to the echomorphology in the B-mode image. The subjective visual diagnosis is often difficult, especially the differentiation between thrombi and non-calcified plaque. The aim of this study was the extraction of features from the ultrasound radio-frequency signal for an objective characterization of coronary tissue. Methods of texture analysis and frequency analysis were used to differentiate red and white thrombi in vitro. Eight texture parameters of first and second order significantly differentiated red and white thrombi. The backscatter transfer function of red thrombus tissue was significantly lower than the backscatter transfer function of white thrombi between 15 and 25 MHz.