Compressed sensing MRI with variable density averaging (CS-VDA) outperforms full sampling at low SNR.

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Image Reconstruction
• Medical Physics

Background:

• Low signal-to-noise ratio (SNR) in MRI limits image quality, especially for high-resolution or specific contrast-weighted scans.
• Conventional MRI acquisition methods often require long scan times to achieve adequate image quality.
• Accelerated MRI techniques are crucial for improving patient comfort and throughput.

Purpose of the Study:

• To evaluate the effectiveness of combining k-space undersampling with variable density averaging (CS-VDA) for enhancing image quality in low-SNR MRI acquisitions.
• To compare the performance of the CS-VDA method against full k-space sampling in terms of image quality and scan time.
• To determine the optimal parameters for CS-VDA, particularly the distribution of averages within k-space.

Main Methods:

• Implementation of 3D Cartesian k-space prospective undersampling with a variable number of averages per k-space line.
• Retrospective analysis of fully sampled phantom MRI measurements to assess CS-VDA performance.
• Prospective acceleration of in vivo 3D brain and knee MRI scans using the CS-VDA technique.

Main Results:

• Both phantom and in vivo studies demonstrated improved image quality with CS-VDA compared to full sampling within the same scan time.
• CS-VDA with a higher number of averages in the center of k-space yielded the best image quality.
• Acquisitions showed increased anatomical detail and preserved soft-tissue contrast.

Conclusions:

• The CS-VDA approach significantly enhances image quality for inherently low-SNR MRI data.
• This novel method allows for improved imaging of challenging datasets, such as high-resolution or low-SNR contrast-weighted scans.
• CS-VDA offers a promising solution for achieving high-quality MRI with reduced scan times.