Super Slice Interpolation For Generating Thin-Slice Images From Multichannel Multislice MRI Data

Abstract

This study aims to develop a super slice interpolation (SSI) method that generates thin-slice images from multichannel multislice images by exploiting the intra-slice coil sensitivity variations. SSI first calculates the thin-slice sensitivity maps by through-plane interpolation of the sensitivity maps computed from the acquired multislice images. It then reconstructs multiple thin-slice images from each acquired image using a through-plane regularized sensitivity encoding (SENSE) like procedure that consists of an initial SENSE reconstruction and denoising to set the prior information image, and subsequent regularized SENSE reconstruction. We evaluated SSI using multislice brain and abdominal images with typical slice thickness. SSI successfully separated each acquired image into two thinner ones without magnitude bias. Compared with the original thick-slice images, SSI revealed more anatomical details that were consistent with those in the separately acquired thin-slice images. SSI presents a novel slice interpolation approach to obtain thin-slice images from the multichannel thick-slice images.