Analysis of surface roughness in optical coherence elastography using a novel phantom

Abstract

Compression optical coherence elastography (OCE) has shown promise in detecting diseases, such as breast cancer, based on changes in the elasticity of tissue. While the mechanical models used in OCE assume a flat surface, tissue typically exhibits surface topography that can lead to inaccurate estimation of elasticity in OCE images (elastograms). The coupling of surface topography and heterogeneous mechanical properties in tissue make it challenging to study the relative impact of surface roughness on image formation. To address this, we developed mechanically homogeneous phantoms that accurately replicate the surface topography of human breast tissue. Surface topography of breast tissue specimens was captured in three-dimensional (3-D) optical coherence tomography (OCT) scans. Molds replicating the surface roughness of tissue were generated using 3-D modelling and 3-D printing. The phantoms were then fabricated in the molds using polydimethylsiloxane. OCT was used to validate the accuracy of the phantom surface topography and the phantoms were then used to determine the impact of surface roughness in compression OCE elastograms. Surface roughness phantoms were shown to accurately reflect the surface topography of six breast tissue specimens with a correlation of 0.96 ± 0.03. When imaged with a variant of compression OCE called quantitative micro-elastography, surface roughness resulted in a slight decrease in elasticity accuracy with a mean percentage error of 8.2 % for a flat phantom compared to 11.7 % for the roughness phantom, and a larger decrease in elasticity precision, or sensitivity, with a mean relative standard deviation of 9.4 % for the flat phantom compared to 21.2 %, for the roughness phantom. Our results show that contrast in elastograms moderately correlated with the phantom surface profile (r = 0.59, p < 0.001) and that surface roughness introduces spatially-varying elasticity that should be considered in the analysis of elastograms in compression OCE.