Title: High frequency residues: a new set of signals for detectability studies of an X-ray imaging system

Abstract: A new set of signals for studying detectability of an x-ray imaging system is presented. The results obtained with these signals are intended to complement the NEQ results. The signals are generated from line spread profiles by progressively removing their lower frequency components and the resulting high frequency residues (HFRs) form the set of signals to be used in detectability studies. Detectability indexes for these HFRs are obtained using a non-prewhitening (NPW) observer and a series of edge images are used to obtain the HFRs, the covariance matrices required by the NPW model and the MTF and NPS used in NEQ calculations. The template used in the model is obtained by simulating the processes of blurring and sampling of the edge images. Comparison between detectability indexes for the HFRs and NEQ are carried out for different acquisition techniques using different beam qualities and doses. The relative sensitivity shown by detectability indexes using HFRs is higher than that of NEQ, especially at lower doses. Also, the different observers produce different results at high doses: while the ideal Bayesian observer used by NEQ distinguishes between beam qualities, the NPW used with the HFRs produces no differences between them. Delta functions used in HFR are the opposite of complex exponential functions in terms of their support in the spatial and frequency domains. Since NEQ can be interpreted as detectability of these complex exponential functions, detectability of HFRs is presented as a natural complement to NEQ in the performance assessment of an imaging system.