Alternate update of shifted extended keyholes (AUSEK): A new accelerating strategy for interventional MRI

Abstract

Real-time or near-real-time acquisition plays a key role in providing immediate image guidance for interventional magnetic resonance imaging (iMRI). However, the requirement of accurate needle tip localization has made several accelerating techniques, like Keyhole imaging or sliding window reconstruction, difficult to be applied to iMRI. The purpose of this work was to further explore the possible ways of applying view sharing techniques to iMRI. Inspired by Keyhole imaging, we present an easy-to-implement accelerating strategy called "Alternate update of shifted extended keyholes (AUSEK)". In this method, the keyhole views are not only extended but also shifted towards either high-frequency edge to form two divisions in k-space. The divisions which are mirrored to each other along the center are alternately updated following a reference scan. By using simulations and experiments, we demonstrate that AUSEK could effectively preserve the spatial resolution of the image, especially of the needle, at a temporal acceleration rate of about 2.5. AUSEK was implemented online in an open-configuration low-field MR imaging system.

A feature‐based convolutional neural network for reconstruction of interventional MRI

“…5 By adopting a radial sampling scheme to the keyhole method, in 2001 Shankaranarayanan et al proposed a radial keyhole sequence with improved temporal resolution and signal-to-noise ratio (SNR). 6 In 2017, Yu et al 7 modified the keyhole imaging with a shifted partial k-space acquisition towards the higher frequency edge. By taking advantage of the image sparsity, Hernando et al 8 incorporated compressed sensing (CS) into I-MRI.…”