[Evaluation of Vessel Depictability in Compressed Sensing MR Angiography Using Numerical Phantom Model]

Area of Science:

• Medical Imaging
• Signal Processing
• Informatics

Context:

• Compressed sensing MRI (CSMRI) accelerates image acquisition by using under-sampled k-space data.
• CSMRI reconstruction involves complex iterative procedures based on data sparsity.
• Evaluating image quality (IQ), particularly thin blood vessel depictability in MR angiography (MRA), is challenging.

Purpose:

• To investigate the factors influencing image quality in compressed sensing MR angiography (CS-MRA).
• To quantitatively assess the depictability of thin blood vessels in CS-MRA using a numerical phantom.
• To elucidate the relationship between noise intensity, sparsifying transform, and vessel depictability in CS-MRA.

Summary:

• Numerical experiments were conducted using a cerebral artery phantom to evaluate CS-MRA image quality.
• Results indicate that vessel depictability is sensitive to noise intensity when using wavelet transform as the sparsifying transform.
• Reduced vessel depictability was observed at clinically relevant signal-to-noise ratio (SNR) levels.

Impact:

• Findings highlight the critical importance of SNR in CS-MRI for achieving adequate thin blood vessel depiction.
• Careful selection of data acquisition and reconstruction parameters is essential for optimizing CS-MRA performance.
• This research informs strategies for improving diagnostic accuracy in CS-MRA studies.