An Observer Model for Lesion Detectability in Contrast-Enhanced Digital Mammography.

Lesion detectability in contrast-enhanced digital mammography (CEDM) was studied using a model observer for task-based system evaluation. Digital mammography (DM) and CEDM images were simulated to include a realistic mammographic appearance and contrast enhancement based on simple physiological assumptions of iodinated contrast agent uptake. A Laguerre-Gauss Channelized Hotelling observer (LG-CHO) was implemented to compute the lesion detectability, d', according to the signal-to-noise ratio of the CHO test statistic. Iodine concentrations from 0.3 to 3 mg/ml in simulated lesions resulted in greater d' for log-subtracted CEDM images (acquisition technique of Jong et al. Radiology 2003) compared to full dose DM (standard screening DM acquisition technique). Doubling mean glandular radiation dose yielded only minimal relative gains in d' for all but the lowest lesion contrast agent uptake levels considered, suggesting that quantum noise was not the limiting factor for lesion detectability. [ABSTRACT FROM AUTHOR]