Tissue boundary roughness index for evaluating malignancy of cancer in ultrasound-computed tomography system

Abstract

Ultrasound-computed tomography (USCT) is a promising candidate for a radiation free, painless, and quantitative modality for breast cancer examination. We developed a USCT prototype with a ring-shaped transducer array (frequency: 1.5 MHz) that moves along the ring axis for scanning the entire breast. Distributions of speed and attenuation of ultrasound are computed from ultrasound transmitted through tissue. Although prior research shows that there is a correlation between malignancy and the speed of sound and attenuation values, in some cases both benign and malignant tumors have similar values. In this study, we propose to quantify the boundary roughness of a tumor from the spatial power distributions of reflected ultrasounds that are transmitted from different apertures of the ring transducer. In conventional ultrasound echography, the boundary roughness of a tumor is an important indicator of malignancy. However, roughness is judged by radiologists based on B-mode images and is a subjective index. We simulated the ultrasound reflection signal from boundaries of different roughness. The simulation result indicates that a spatial power concentration has a significant correlation with boundary roughness. We confirmed through gel phantom experiments that the index distinguishes rough and smooth boundaries, which is not evident in B-mode images.