A general framework for the assessment of scatter correction techniques in digital mammography

Scattered radiation negatively impacts radiographic imaging, with particular regard to mammography. In clinical practice, anti-scatter grids are exploited for this purpose; however, anti-scatter grids may also degrade the image quality, since they remove part of the useful primary radiation with the consequent increase of the dose to be administered to the patient. A suitable digital scatter correction method could tackle the limits imposed by such grids with a great impact on diagnosis. The main contribution of this study is the development of a general framework for the assessment of digital scatter correction techniques in mammography. To this aim, the formation process of both primary and scattered image is described on the basis of a systems-theory approach. Through a simulation of the radiological process, a reference model of the primary image is obtained and used as ground truth to compare the intensities of images obtained by applying a deconvolution-based digital scattering correction technique. Then, an experimental case study on breast phantom images is carried out to assess the scatter correction using different Point Spread Functions (PSFs) (Gaussian and Hyperbolic) with varying parameters values. A central issue was the identification of a spatially variant PSF to model the scattered radiation. The results demonstrate that the proposed approach enables the assessment and the comparison of different PSF kernels employed for scatter correction; in particular, our procedure shows that rather low relative errors are obtained ([−0.5;0.5]) for both the PSFs tested and that Gaussian ones are more sensitive to variations in their parameters.