Improved method for determining resolution zones in ultrasound phantoms with spherical simulated lesions.

Quantification of ultrasound (US) imager performance simulating human observers is addressed using size-dependent lesion signal-to-noise ratio (LSNR) analysis of images of spherical simulated lesions in phantoms. LSNR values obtained over a broad range of image depths can be used with a single detectability threshold to determine the depth range over which lesions of a given size and contrast are deemed to be detectable, yielding a performance metric. Optimal LSNR analysis requires a priori knowledge of lesion locations in the image so that LSNR values relate to the center of each lesion. Phantoms having a regular array of accurately positioned spherical simulated lesions are described, along with easily employed and robust software that accurately determines lesion locations in images, even when only a few are detectable by visual inspection. The software accounts for image spatial calibration inaccuracies and accommodates sector, curvilinear, etc., formats. The minimum number of equivalent lesions to yield an acceptable mean (counteracting speckle variations) also is addressed.