Deep Learning-Based Contrast Boosting in Low-Contrast Media Pre-TAVR CT Imaging

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Transcatheter aortic valve replacement (TAVR) requires accurate pre-procedural imaging.
• Patients with renal dysfunction often benefit from reduced contrast media (low-CM) protocols.
• Assessing image quality and measurement reliability in low-CM CT is crucial for TAVR planning.

Purpose of the Study:

• To evaluate the effectiveness of deep learning-based contrast boosting (DL-CB) in improving image quality for low-CM CT scans.
• To assess the impact of DL-CB on the reliability of aortic annular measurements in pre-TAVR CT.
• To determine if DL-CB can achieve diagnostic image quality comparable to standard-CM protocols without dual-energy CT.

Main Methods:

• Retrospective analysis of low-CM pre-TAVR CT scans in patients with renal dysfunction.
• Comparison of image quality metrics (CNR, SNR) between conventional, 50-keV, and DL-CB reconstructed low-CM images.
• Quantitative and qualitative assessment of aortic annular measurements and interobserver reliability using DL-CB images.

Main Results:

• DL-CB significantly improved contrast-to-noise ratio (CNR) and signal-to-noise ratio (SNR) compared to conventional and 50-keV reconstructions (P < .001).
• DL-CB achieved CNR and SNR comparable to standard-CM CT scans.
• DL-CB demonstrated high interobserver reliability for aortic annular measurements (ICC = .96) with minimal differences.

Conclusions:

• DL-CB effectively enhances image quality in low-CM CT for pre-TAVR assessment in patients with renal dysfunction.
• DL-CB provides reliable aortic annular measurements, crucial for TAVR planning.
• This deep learning technique offers a valuable alternative to standard-CM protocols, avoiding the need for dual-energy CT.