Evaluation of CT dose-averaged detectability: a repeatability study

In diagnostic radiology, it is essential to ensure that image quality is sufficiently good to support accurate diagnosis and patient safety. This is in accordance with the ALARA principle of keeping radiation exposure as low as reasonably achievable - taking into account economic and societal factors. Further, the detection of subtle low-contrast differences is critical for early disease detection, particularly in cases such as liver metastases. Iterative reconstruction algorithms have improved medical imaging by reducing noise, but their non-linear nature challenges traditional assessment metrics. Model observers like the Channelized Hotelling Observer provide task-specific evaluations suitable for non-linear systems. This study employs a previously proposed method for determining the dose-averaged detectability in computed tomography (CT) systems, focusing on minimizing systematic errors from measurement processes and phantom specimen variations. We conducted two repeatability analyses: intra-specimen, using the same CT scanner and phantom, and inter-specimen, using different specimens of the same phantom type. The study utilized a GE Brightspeed 16 scanner. The analyses revealed significant variations among different specimens, necessitating corrections for consistent comparison. The main correction underscores the importance of experimental verification of actual contrast values for accurate determination of the figure of merit (FOM) with multiple phantom specimens.