Multiresolution imaging using golden angle stack-of-stars and compressed sensing for dynamic MR urography

Abstract

Purpose: To develop a novel multiresolution MRI methodology for accurate estimation of glomerular filtration rate (GFR) in vivo.

Materials and methods: A three-dimensional golden-angle radial stack-of-stars (SoS) trajectory was used for data acquisition on a 3 Tesla MRI scanner. Multiresolution reconstruction and analysis was performed using arterial input function reconstructed at 1-s. temporal resolution and renal dynamic data reconstructed using compressed sensing (CS) with 4-s temporal resolution. The method was first validated using simulations and the clinical utility of the technique was evaluated by comparing the GFR estimates from the proposed method to the estimated GFR (eGFR) obtained from serum creatinine for 10 subjects.

Results: The 4-s temporal resolution CS images minimized streaking artifacts and noise while the 1-s temporal resolution AIF minimized errors in GFR estimates. A paired t-test showed that there was no statistically significant difference between MRI based total GFR values and serum creatinine based eGFR estimates (P = 0.92).

Conclusion: We have demonstrated the feasibility of multiresolution MRI using a golden angle radial stack-of-stars scheme to accurately estimate GFR as well as produce diagnostic quality dynamic images in vivo.

Level of evidence: 1 Technical Efficacy: Stage 3 J. MAGN. RESON. IMAGING 2017;46:303-311.

Keywords: compressed sensing; dynamic contrast enhanced MRI; glomerular filtration rate; multiresolution MRI; pharmacokinetic modeling.