Effect of Monoenergetic Reconstructions and Ultra-High-Resolution on the Detection of Small Endoleaks With Photon-counting Detector CT

BACKGROUND AND OBJECTIVES: Following endovascular repair of an aortic aneurysm, patients need regular surveillance using CT to exclude endoleaks. The optimal technical parameters for photon-counting detector CT (PCD-CT) have not yet been established in this clinical context. This study examines the impact of the ultra-high resolution (UHR) mode and monoenergetic reconstructions at various keV on the detection of small endoleaks in experimental settings. MATERIALS AND METHODS: An aneurysm phantom model mimicking medium- and large-sized patients was scanned using PCD-CT. Monoenergetic images (MEIs) were reconstructed with standard resolution at 40, 45, 50, 55, and 60 keV and with UHR at 45, 50, and 55 keV. Three independent blinded radiologists assessed 80 CT series containing 2880 simulated endoleaks with diameters of 2, 4, and 6 mm. The observers' ratings were analyzed and compared with the jackknife alternative free-response operational characteristics (JAFROC1) method by calculating a figure-of-merit for the performance ( A1 ). The Spearman R was calculated for the correlation between A1 and objective image quality. RESULTS: The best readers' performance was achieved with the 50 keV UHR, 55 keV standard, 55 keV UHR, and 60 keV standard series ( A1 =0.66 to 0.72) in the medium and with the 55 keV standard, 55 keV UHR, and 60 keV standard reconstructions ( A1 = 0.40 to 0.49) in the large phantom. UHR provided higher A1 than the standard reconstruction at 50 keV for 4 to 6 mm endoleaks ( A1 , 0.82 vs. 0.72; P =0.015 in medium and 0.42 vs. 0.30; P =0.028 in large phantom) and for 6 mm endoleaks ( A1 , 0.93 vs. 0.83; P =0.002 in medium and 0.53 vs. 0.36; P =0.013 in large phantom). In both phantoms, readers' performance correlated negatively with image noise (Spearman R =-0.83; P =0.01 in the medium and Spearman R =-0.98; P