Deep Learning-based Monoenergetic Imaging for Calcified Coronary Stenosis Assessment at Energy-integrating Detector...

Area of Science:

• Medical Imaging
• Artificial Intelligence in Radiology
• Cardiovascular Imaging

Background:

• Coronary CT angiography (CCTA) using energy-integrating detector (EID) CT is limited by blooming artifacts, especially with heavily calcified plaques.
• Accurate stenosis assessment in CCTA is crucial for diagnosing coronary artery disease.
• Photon-counting detector (PCD) CT offers improved image quality but requires hardware upgrades.

Purpose of the Study:

• To develop and evaluate the Deep learning-bAsed MONoenergetic imaging at Different energies (DIAMOND) framework.
• To generate virtual monoenergetic images (VMIs) from conventional EID CT.
• To reduce blooming artifacts and improve stenosis assessment in CCTA for heavily calcified plaques.

Main Methods:

• DIAMOND framework utilizes a U-Net architecture trained on retrospective EID CT and PCD CT data.
• The model was applied prospectively to patients with heavily calcified plaques undergoing EID CT and PCD CT.
• Quantitative analysis of percent diameter stenosis (PDS) was performed and compared across EID CT, DIAMOND, and PCD CT.

Main Results:

• DIAMOND effectively reduced blooming artifacts and improved lumen visualization, yielding image quality comparable to PCD CT.
• Average PDS decreased significantly from 35.65% (EID CT) to 25.19% (DIAMOND, P < .05), approaching PCD CT values (24.27%).
• DIAMOND reclassified stenosis severity in 42% of lesions, improving diagnostic accuracy without hardware changes.

Conclusions:

• The DIAMOND framework enables the generation of high-kiloelectron volt VMIs from single-energy EID CT.
• DIAMOND provides artifact-reduced coronary imaging and improved stenosis quantification for heavily calcified plaques.
• This deep learning approach offers a cost-effective alternative to PCD CT for enhanced CCTA analysis.