Image quality assessment (IQA) protocol ensures that mammography equipment operates according to its design standards. IQA permits to detect alterations in the equipment that may impact negatively the interpretation of mammograms. The mammography accreditation phantom simulates the radio-graphic attenuation of an average-size compressed breast and contains structures that model very basic image characteristics of breast parenchyma and cancer. It is composed of a poly-methyl-methacrylate (PMMA) block 4.5 mm thick and a wax insert. The wax insert contains six disks, fibers, and calcifications. To pass the image quality standards for screen mammography, at least four fibers, three calcification groups, and three masses must be clearly visible by a human reader (with no obvious artifacts) at an average glandular dose of less than 2.5 mGy. Predicting human performance in quality control process is critical for task efficacy. In this paper, as a first step to predict automatically human performance in the recognition of structures, we analyze under different acquisition conditions the signal-to-noise ratio (SNR), the contrast-to-noise ratio (CNR) for the different types of structures present in a phantom MG image (PMGI) and the extraction of Natural Scene Statistics (NSS) from a PMGI.