This retrospective study investigated the potential benefits of a deep learning-based noise reduction (DLD) technique in improving image quality and diagnostic performance of contrast-enhanced CT coronary assessment prior to transcatheter aortic valve implantation (TAVI). Two hundred patients with severe aortic stenosis who underwent CT examinations between October 2022 and April 2024 were included. Conventional and denoised images were compared using objective metrics—specifically signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR)—and a subjective 5-point Likert scale assessing sharpness, noise, contrast, and overall quality. Denoised images demonstrated significantly higher SNR and CNR, as well as reduced noise and superior overall quality (all P < 0.001). Across 800 vessels and 1,787 segments analyzed, DLD achieved an area under the curve of 0.90, an accuracy of 93.9%, a sensitivity of 85.7%, and a specificity of 94.7%, with no significant difference compared with standard images (P = 0.056). In conclusion, DLD significantly improves CT image quality for coronary evaluation prior to TAVI without compromising diagnostic accuracy.