Implementation and validation of a channelized hotelling observer model (CHO) in an AIFM task group: a national...

This study aims to prove the feasibility of objectively characterizing clinical XA protocols by implementing a model observer. The model observer's performance aligns with a human observer's and is described using a simple and comprehensive figure of merit (FOM). The practical implications of this study, which utilized the Leeds TO10 phantom to acquire 146 imaging datasets in a fixed setup measuring kerma rate from four manufacturers and seven XA models by thirteen hospitals, are significant. The datasets were divided and analyzed into three main protocol categories (cardiac, neurological, and vascular) acquired with the field of view (FOV) locally used in that hospital. A 40-channel Gabor CHO was employed to analyze the datasets and calculate the contrast detail (CD) curves. A new figure of merit (FOM) tailored for the present task was calculated, accounting for image quality and kerma rate. The FOM demonstrated our observer model's ability to describe the XA protocol's optimization. Short-term reproducibility of selected XA protocols was within 10%. Smaller FOVs lowered long-term reproducibility in terms of FOM because the position of the dosimeter increasingly influenced the automatic exposure parameters. This study demonstrates the feasibility of using a CHO model observer to assess an angiography system's quality using a CD paradigm. The insights gained from this study will be instrumental in developing tolerance requirements for future quality assurance guides, enhancing the quality of X-ray angiography protocols, and improving patient care.