Applicability and repeatability in angiography systems of statistical method for the low contrast detectability

Abstract

Purpose The aim of this study is to evaluate the applicability and repeatability of the Low Contrast Detectability (LCD) statistical method in angiography systems [1] , [2] . Methods A phantom consisting of a square aluminum target and a 5-step-wedge (0.2–1 mmAl) was built according to Radice et al. [2] . Different acquisition protocols were used (low, normal and high dose fluoroscopy; abdomen and cerebtal fluorography), and measurements were repeated for several thicknesses of PolyMethylMethAcrylate (PMMA). A Matlab software automatically calculates LCD values in terms of Pixel Value (PV) and contrast threshold (Ct); it also converts PV in mmAl (Fig. 1), providing the LCD(mmAl) [2] . The short- and long-term repeatability were evaluated by repeating acquisitions five times for every session and monthly (for six months) respectively. Results The Coefficients of Variation CV(%) reported in Table 1 are always higher for LCD(mmAl) than for LCD(PV) and LCD(Ct), especially in case of long-term measurements which consequently are not repeatable. This difference is related to unstable conversion from PV to mmAl; indeed, the image noise makes difficult to select the steps of the wedge. To overcome the problem, it would be possible to extend the step-wedge up to 2 mmAl, to introduce the point (0,0) in the conversion curve and to use only the curves with R2 > 0.8. Conclusions The LCD statistical method is easy to implement and provides consistent results for LCD(PV) and LCD(Ct). However, the LCD(mmAl) values seem to be not repeatable because of the instability of PV-mmAl conversion; the suggested solutions can improve the conversion, leading to more robust LCD (mmAl) values.