Scatter reduction by air gaps in mammography was investigated. We have experimentally demonstrated that, independently of the imaging geometry, scatter in air-gap mammography can be well described by a virtual source of scatter (VSS) model. This model postulates that scatter radiation originates from a virtual point source of scatter placed on the central axis between the x-ray source and the exit surface of a patient at distance delta and utilizes only two parameters: delta and (S/P)0. The (S/P)0 parameter represents scatter-to-primary ratio without an air gap and delta is the distance from the exit surface of a patient to the virtual source of scatter. We have experimentally determined the analytical form of the two independent parameters of the VSS model; delta exhibits a linear increase proportional to the radiation field size, does not depend on patient thickness, and is in the 10-30 cm range, while (S/P)0 increases with the field size as a power function and is in the 0.4-1.3 range. In the framework of the VSS model the selectivity, the contrast improvement factor, and the signal-to-noise improvement factor were employed to evaluate performance of air-gap mammography systems. We have demonstrated that selectivity of an air gap rapidly deteriorates at some well-defined critical value of scatter fraction that has profound consequences on air-gap performance. Assuming fixed patient exposure, the results shows that, if a contrast limited detection system (such as film/screen mammography) is used, an air gap system can outperform a grid system only if a very large source-to-patient (SPD) distance is utilized, which might be possible with new laser-based x-ray sources. For the noise limited detection systems (such as digital mammography) even a small SPD (70 cm) and a small air-gap (20 cm) system will outperform a grid system.

No abstract