Implementation of time-efficient adaptive sampling function design for improved undersampled MRI reconstruction

• Published in: Journal of magnetic resonance (1997) Vol. 273; pp. 47 - 55
• Main Authors: Choi, Jinhyeok, Kim, Hyeonjin
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 01.12.2016
• Subjects: Compressed sensing; MRI; Probability density function; Sampling function; Undersampling; Undersampling; Probability density function; MRI; Compressed sensing; Sampling function
• ISSN: 1090-7807; 1096-0856
• DOI: 10.1016/j.jmr.2016.10.006

[Display omitted] •A simple adaptive sampling function design is proposed for undersampling MRI.•An approximated energy distribution of an image in k-space is obtained in a prescan.•The proposed method was actually implemented on an MR scanner.•The proposed method was tested in Fourier transform and compressed sensing MRI.•The proposed method consistently outperforms the conventional approach. To improve the efficacy of undersampled MRI, a method of designing adaptive sampling functions is proposed that is simple to implement on an MR scanner and yet effectively improves the performance of the sampling functions. An approximation of the energy distribution of an image (E-map) is estimated from highly undersampled k-space data acquired in a prescan and efficiently recycled in the main scan. An adaptive probability density function (PDF) is generated by combining the E-map with a modeled PDF. A set of candidate sampling functions are then prepared from the adaptive PDF, among which the one with maximum energy is selected as the final sampling function. To validate its computational efficiency, the proposed method was implemented on an MR scanner, and its robust performance in Fourier-transform (FT) MRI and compressed sensing (CS) MRI was tested by simulations and in a cherry tomato. The proposed method consistently outperforms the conventional modeled PDF approach for undersampling ratios of 0.2 or higher in both FT-MRI and CS-MRI. To fully benefit from undersampled MRI, it is preferable that the design of adaptive sampling functions be performed online immediately before the main scan. In this way, the proposed method may further improve the efficacy of the undersampled MRI.