Item

Abstract

Purpose/Introduction: Over-discrete SENSE reconstruction with direct control of the shape of the spatial response function (SRF) [1] has been shown to reduce short- and far-reaching voxel bleeding. This makes it a suitable candidate for acceleration of non-lipid suppressed FIDMRSI, since lipid signals bleeding from the unsuppressed subcutaneous fat into the brain region become very problematic. Additionally, using the B0 map to correct for the frequency shifts on a sub-voxel level has been proven to provide SNR gain and line-shape improvement [2], which is especially beneficial for highly accelerated data. The aim of this work was to demonstrate the benefits of over-discrete SENSE reconstruction scheme along with the B0 correction on nonlipid suppressed ultra-short TR and TE 1H FID MRSI data at 9.4T. Subjects and Methods: High resolution (3.125mmx3.125mmx10 mm) single-slice ultra-short TR and TE non-lipid suppressed 1H FID MRSI [3–6] were acquired from the brains of healthy volunteers on a 9.4T Siemens whole-body human scanner with a TR of 300 ms. A high resolution scout anatomical image was used to extract the coil sensitivities using the ESPIRiT method [7], and high resolution B0 map was also acquired using a 2D GRE dual-echo sequence for B0 correction. The MRSI data was retrospectively undersampled by a factor of R = 2 9 2 and reconstructed with three schemes: (1) conventional SENSE, (2) over-discrete SENSE (OD-SENSE) with an over-discretization factor of 4, along with a Gaussian shape target matrix with the FWHM of the nominal voxel-size without and, (3) with B0 correction on a sub-voxel level (B0 OD-SENSE). Results: Figure 1 shows lipid contamination maps from the fully sampled along with the R = 4 times accelerated data reconstructed once with a conventional SENSE reconstruction scheme and once with the over-discrete scheme and direct control of the SRF The average SNR of the over-discrete reconstructed R = 2 9 2 SENSE data over the whole slice was 85.9 and 71.2, with and without the B0 correction, respectively. Discussion/Conclusion: In non-lipid suppressed MRSI, the lipid signals in the subcutaneous region are orders of magnitude larger than the metabolite signals. Therefore, any residual aliasing artifacts resulting from conventional SENSE reconstruction will overshadow the metabolites in the brain. The over-discrete reconstruction scheme greatly reduced this artifact as can be seen in Fig. 1. The over-discrete B0 correction resulted in 1.2 times gain in SNR which is in accordance with [2].