Po-topic V-01: Parametric imaging of spontaneous mammary tumors in rats for the purposes of tissue microstructure characterization

Abstract

The characterization of tissue microstructure using ultrasound backscatter can assist in the detection and classification of diseased tissues. Parametric images can be formed from tissue characterizations that show structure information below that resolved for conventional ultrasound. Eight retired breeders were acquired that had developed spontaneous mammary tumors. 2D B-mode images of the rat tumors were constructed from backscattered echoes. After scanning, tumors were dissected free from each rat, trimmed in the plane of ultrasound exposure, fixed in 10% neutral-buffered formalin, embedded in paraffin, sectioned at 5 μm, and stained with hematoxylin and eosin. Tumors were diagnosed microscopically as mammary gland fibroadenomas. Regions of interest (ROIs) were selected from the ultrasound B-mode images of the tumors and surrounding tissues. Average scatterer properties (effective scatterer size and acoustic concentration) were estimated from the backscattered RF signal used to make the B-mode images. Scatterer estimates were made by using least squares to fit a line to the measured form factor that was calculated from the backscattered power spectrum. The effect of noise in the signal from increasing attenuation with depth was to reduce the ability to make estimates of scatterer properties at greater depths. A weighting scheme was used to reduce the effects of noise and increase the ability to make accurate estimates to greater depths. Comparison of scatterer estimates between normal tissues (intercostal tissues) and tissues inside the tumors were made. On average, the estimated effective scatterer diameters inside the tumors were 30% larger at 107 μm than estimates of effective scatterer diameters outside the tumors averaging 82 μm. The average acoustic concentration estimated inside the tumor was 3.16 × 10−2/mm−3 as opposed to 0.746 mm−3 for outside the tumor. In all but one of the rats, there was a statistically significant difference (P < .05) between estimates of scatterer properties made inside the tumors and in surrounding healthy tissues. Feature analysis plots clearly showed distinction between the diseased tissues and the normal tissues. Enhanced B-mode images were constructed by superimposing colored pixels corresponding to the estimated average scatterer properties in different ROIs on the gray-scale B-mode images. The enhanced B-mode images highlighted differences between tissues inside and outside the tumors.