The potential of a coronary artery-specific deep learning CT reconstruction algorithm for improvement in image quality of abdominal CT angiography with special reference to small arterial visibility

/Introduction: We sought to clarify the potential of Precise IQ Engine (PIQE) for the visualization of small arteries in abdominal computed tomography angiography (CTA).

CT images scanned with a 320-multidetector row CT were generated using four reconstruction methods: Adaptive Iterative Dose Reduction 3D (AIDR 3D), Forward projected model-based Iterative Reconstruction SoluTion (FIRST), Advanced intelligent Clear-IQ Engine (AiCE), and PIQE. Phantom study: We evaluated spatial resolution via the modulation transfer function (MTF) and the slit phantom assessment. We also measured the noise power spectrum (NPS) and CT number to evaluate image noise.

Abdominal CTA images from five renal donor candidates were used to evaluate the visibility of four small abdominal arteries per person. The differences in visibility scores evaluated by two radiologists were analyzed using Sheffe's multiple comparison method.

The 10 % values of MTF were 0.54 for AIDR 3D, 0.67 for FIRST, 0.66 for AiCE, and 0.66 for PIQE. In the slit phantom assessment, FIRST, AiCE, and PIQE showed no clear difference in attenuation profiling curves, whereas the maximum spatial resolution for AIDR 3D was the worst. The NPS of FIRST was the highest, whereas that of PIQE was the lowest for the low spatial frequency range. The standard deviation of CT number with PIQE was the lowest, whereas that of FIRST was the highest. The median visibility score of PIQE was significantly higher than those of AIDR 3D, FIRST and AiCE, respectively, for each radiologist (p < 0.05).

PIQE may improve the image quality of abdominal CTA.