Low-Echo Sphere Phantoms and Methods for Assessing Imaging Performance of Medical Ultrasound Scanners

Abstract

Tissue-mimicking phantoms and software for quantifying the ability of human observers to detect small low-echo spheres as a function of depth have been developed. Detectability is related to the imager's ability to delineate the boundary of a 3-D object such as a spiculated tumor. The phantoms accommodate a broad range of transducer shapes and sizes. Three phantoms are described: one with 2-mm-diameter spheres (for higher frequencies), one with 3.2-mm-diameter spheres (for lower frequencies) and one with 4-mm-diameter spheres (for lower frequencies). The spheres are randomly distributed in each phantom. The attenuation coefficients of spheres and surroundings are nearly identical; thus, compromising shadowing or enhancement distal to spheres does not occur. Reproducibility results are given for pairs of independent data sets involving eight different combinations of scanner, transducer and console settings. The following comparison results are also reported: (i) only the selected frequency differs; (ii) transducers and scan parameters are nearly the same but manufacturers differ; (iii) ordinary B-scanning, spatial compounding and tissue harmonic imaging are addressed. The phantoms and software promise to be valuable tools for scanning system and setup comparisons and for acceptance testing.