Low contrast detectability assessment with homemade software and dedicated phantom: evaluation of results and comparison with cdmam analysis

Abstract

Introduction Contrast detail curve, or LCD, is an important aspect of image quality. LCD curves are usually determined with the Artinis CDMAM phantom. Purpose This study proposed a different method to obtain contrast detail curves based on a unique image of a homemade phantom. Materials and methods Phantom was made up of an acetate sheet equipped with a central uniform aluminium region ( 3 × 3 cm 2 , thickness 0.49 mm) and an aluminium step wedge (thickness: 0.2–1 mm) for linear conversion pixel values/millimetres of aluminium. LCD was estimated through a statistical method analysing a square ROI on the uniform aluminium region with a dedicated Matlab program. For each details dimension, it was divided into a matrix of subROIs: contrast thresholds were defined as 3.29 the standard deviation of subROIs mean values, corresponded to a statistical significance of 95%. Results were then converted in terms of mmAl. Reproducibility and variability with different combinations of kV, mAs and AGD were evaluated. Reliability of the method was evaluated comparing results with CDMAM curves. The effect of human perception was considered through a NPWE model and compared with the results of 2AFC experiment. Results Analysis confirmed the reproducibility of the method. The comparison with CDMAM analyses showed a good agreement for each equipment considered in this study, confirming the good reliability of the method. The Human perception was also introduced with good results. Conclusion The statistical approach proved to be less time-consuming and in accordance with the automatic readout of CDMAM images, allowing to be adopted in image quality assurance protocols.