Denoising parameter dependence of coronary artery depictability in compressed sensing magnetic resonance angiography

Radiol Phys Technol. 2024 Jun;17(2):375-388. doi: 10.1007/s12194-024-00787-w. Epub 2024 Mar 9.

Abstract

Using numerical indices and visual evaluation, we evaluated the dependence of coronary-artery depictability on the denoising parameter in compressed sensing magnetic resonance angiography (CS-MRA). This study was conducted to clarify the acceleration factor (AF) and denoising factor (DF) dependence of CS-MRA image quality. Vascular phantoms and clinical images were acquired using three-dimensional CS-MRA on a clinical 1.5 T system. For the phantom measurements, we compared the full width at half maximum (FWHM), sharpness, and contrast ratio of the vascular profile curves for various AFs and DFs. In the clinical cases, the FWHM, sharpness, contrast ratio, signal-to-noise ratio, noise level values, and visual evaluation results were compared for various DFs. Phantom image analyses demonstrated that the respective measurements of the FWHM, sharpness, and contrast ratios did not significantly change with an increase in AF. The FWHM and sharpness measurements slightly changed with the DF level. However, the contrast ratio tended to increase with an increase in the DF level. In the clinical cases, the FWHM and sharpness showed no significant differences, even when the DF level was changed. However, the contrast ratio tended to decrease as the DF level increased. When the DF levels of the clinical cases increased, the background signals of the myocardium, fat, and noise levels decreased. We investigated the dependence of the coronary-artery depictability on AF and DF using CS-MRA. Analysis of the coronary-artery profile curves indicated that a better image quality was achieved with a stronger DF on coronary CS-MRA.

Keywords: Acceleration factor; Compressed sensing; Coronary magnetic resonance angiography; Denoising factor; Image quality.

MeSH terms

  • Aged
  • Coronary Vessels* / diagnostic imaging
  • Female
  • Humans
  • Image Processing, Computer-Assisted / methods
  • Magnetic Resonance Angiography* / methods
  • Male
  • Middle Aged
  • Phantoms, Imaging*
  • Signal-To-Noise Ratio*