Spatial comparison between the H -scan format for classification of ultrasound scatterers and histology - Preliminary results using an animal model of breast cancer

Abstract

H-scan imaging is a new ultrasound (US) technique used to visualize the relative size of acoustic scatterers. An important issue is the sensitivity of H-scan US image analysis to subtle changes in scatterer sizes. To that end, the purpose of this study was to evaluate the sensitivity of the H -scan US image analysis to scatterer size with a direct comparison to histological measures made at the cellular level. Image data was acquired using a programmable US scanner (Vantage 256, Verasonics Inc) equipped with a 256-element L22-8v capacitive micromachined US transducer (CMUT, Kolo Medical). Plane wave imaging with 5 angles was performed at a center frequency of 15 MHz. To generate H-scan US images, parallel convolution filters based on nth-ordered Gaussian-weighted Hermite polynomial functions (i.e GH 2 , GH 6 , and GH 10 ) were applied to the radiofrequency (RF) data to measure the relative strength of the received signals. The filter outputs were then color coded to form the final H-scan US image display. In vivo H-scan US imaging was performed using human breast cancer xenograft in athymic mice. After euthanasia, excised tumor tissue underwent histological processing and the nuclei were stained. Digitized images were then used to compute two key features, namely, nuclear size and spacing, followed by correlation with the co-registered H-scan US data. The H-scan US images exhibited gross spatial patterns that were mirrored in the histology microscopy images. A significant relationship was found between the local H-scan US image intensity and physical measurements of both nuclear size and spacing (R2 > 0. 61, p < 0.001). Overall, preliminary results from use of preclinical model of breast cancer reveals that in vivo H-scan US images linearly correlate with physical measures of nucleus size and spacing as quantified from co-registered histologic images.