Ultrasound Detection of High Risk Uveal Melanoma Using Non-Spherical Scattering Models

Abstract

Abstract: : Purpose: We have reported the use of ultrasound parameter imaging (UPI) of the underlying microfluctuations of tissue elasticity in ocular melanoma for diagnosis and for clinical staging based on correlation with extravascular matrix patterns that relate to tumor lethality. The theoretical scattering model underlying UPI is for calibrated spectrum analysis with focused transducers assuming weak scattering and a Gaussian spherical autocorrelation function (ACF) to describe the backscatter constituents. In this study we examine the use of alternate functions for describing the shape of scatterers given our histologic understanding of 3-D tumor microarchitecture. Methods: The Cornell-Iowa-Chicago digital RF and pathology database was used for this study. A cohort of 117 cases was reprocessed to obtain UPIs for ellipsoidal and planar scattering functions. UPIs were compared with digital histology with color coded, by observers and using image correlation techniques. Of particular interest were the effects of differing ACFs on the spatial distribution of acoustic concentration (CQ2) in the tumor periphery. Results: Statistically significant correlations were found between histologic and acoustic factors for all ACFs. An improvement in spatial localization of high risk features was seen for ellipsoidal ACFs compared with the spherical model. Conclusions: Incorporating a priori knowledge about the tissue scattering structure can modify the diagnostic capacity of UPI. The ellipsoidal model may provide better detection of the spatial distribution of scattering elements associated with high risk extravascular matrix patterns in the tumor periphery and improve identification of patients with high risk disease.

Keywords: melanoma • imaging/image analysis: clinical • imaging methods (CT, FA, ICG, MRI, OCT, RTA, S