Evaluation of a novel reconstruction method based on the compressed sensing technique: Application to cervical spine MR imaging

Compressed sensing-based reconstruction (CSR) is a new magnetic resonance (MR) image reconstruction method based on the compressed sensing (CS) technique. CSR suppresses ringing artifacts from truncated k-space sampling by estimating the high spatial frequency information required to support the acquired k-space data. CSR is intended to replace the existing zero-fill interpolation (ZIP) reconstruction. We investigated the usefulness of the CSR technique by obtaining sagittal T2-weighted images of the cervical spine and phantom images using CSR or ZIP. Our results indicated that the CSR technique reduces truncation artifacts compared to ZIP without prolonging the scan time or impairing image sharpness. •CSR technique allowed for reduction of truncation artifacts in cervical spine MR images.•CSR technique reduced striped artifacts without prolongation of the scan time.•CSR technique can reduce truncation artifacts while maintaining sharpness, in contrast to windowing with an apodizing filter.•The results of this preliminary study warrant further investigations of the utility of CSR in patients.