Artificial intelligence-based full aortic CT angiography imaging with ultra-low-dose contrast medium: a preliminary...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Radiology

Background:

• Contrast medium (CM) dose reduction is crucial in CT angiography (ACTA) to minimize patient risks.
• Existing low-dose protocols may compromise image quality and diagnostic accuracy.
• Advanced algorithms are needed to enable ultra-low-dose CM protocols without sacrificing diagnostic performance.

Purpose of the Study:

• To evaluate the efficacy of the augmented cycle-consistent adversarial framework (Au-CycleGAN) algorithm in reducing CM dose for full ACTA.
• To assess the impact of the Au-CycleGAN algorithm on image quality and diagnostic accuracy at ultra-low CM doses.
• To determine if AI-enhanced ultra-low-dose CM ACTA can achieve diagnostic performance comparable to standard low-dose CM protocols.

Main Methods:

• A prospective study enrolled 150 patients undergoing ACTA for suspected aortic disease.
• Patients received both ultra-low-dose CM (ULDCM) and low-dose CM (LDCM) protocols.
• AI-based reconstruction using Au-CycleGAN was applied to ULDCM images, followed by comparison with LDCM images.

Main Results:

• AI-based ULDCM images demonstrated improved image quality scores compared to ULDCM and were comparable to LDCM.
• Quantitative metrics like SNR and CNR were superior in AI-based ULDCM images across all aortic locations.
• Diagnostic accuracy for aortic diseases was maintained, with no significant differences between AI-based ULDCM and LDCM groups.

Conclusions:

• The Au-CycleGAN algorithm enables a one-third reduction in CM dose for full ACTA while preserving diagnostic quality.
• AI-enhanced ULDCM imaging offers superior quantitative image quality parameters compared to both ULDCM and LDCM.
• This AI approach provides a safe and effective method for dose reduction in ACTA without compromising diagnostic capabilities.