The Effect of Reconstruction Kernel on Low-Contrast Detectability in ACR CT Phantom Images Evaluated using the Two-Alternative Forced Choice Method

Low-contrast detectability (LCD) is an important aspect of medical imaging. This study aimed to analyze the effect of reconstruction kernel variations on the LCD of ACR CT phantoms evaluated using the two-alternative forced choice (2-AFC) method. The ACR CT phantom was scanned using a Philips MX 16-slice CT scanner with four reconstruction kernels: SA, SC, EA, and EC. LCD analysis was performed using Module #2 of the ACR CT phantom, focusing on 5-mm diameter objects. Seven medical physicist observers with a minimum of three years of experience participated in the study. A total of 80 questions were administered, divided into four sets of 20 items. The results showed that the reconstruction kernel significantly affected LCD. The SA kernel produced the highest detection performance, achieving a percentage correct (PC) score of 93.6%, while sharper kernels, such as the EC kernel, exhibited decreased performance, with a PC score of 90.7%. Analysis of observer response variability showed no significant differences among observers (p = 0.468). In conclusion, the 2-AFC method proved effective for LCD assessment, and kernels with smoother reconstruction characteristics are recommended to improve LCD in low-contrast imaging.