Search

The U-net is a deep-learning network model that has been used to solve a number of inverse problems. In this work, the concatenation of two-element U-nets, termed the W-net, operating in k-space (K) and image (I) domains, were evaluated for multi-channel magnetic resonance (MR) image reconstruction. The two-element network combinations were evaluated for the four possible image-k-space domain configurations: a) W-net II, b) W-net KK, c) W-net IK, and d) W-net KI. Selected four element (WW-nets) and six element (WWW-nets) networks were also examined. Two configurations of each network were compared: 1) each coil channel was processed independently, and 2) all channels were processed simultaneously. One hundred and eleven volumetric, T1-weighted, 12-channel coil k-space datasets were used in the experiments. Normalized root mean squared error, peak signal-to-noise ratio and visual information fidelity were used to assess the reconstructed images against the fully sampled reference images. Our results indicated that networks that operate solely in the image domain were better when independently processing individual channels of multi-channel data. Dual-domain methods were better when simultaneously reconstructing all channels of multi-channel data. In addition, the best cascade of U-nets performed better (p 0.01) than the previously published, state-of-the-art Deep Cascade and Hybrid Cascade models in three out of four experiments.

Background Smaller hippocampal volumes, as assessed by magnetic resonance imaging (MRI), and proton magnetic resonance spectroscopy ( 1 H-MRS) indexed alterations in brain metabolites have been identified in adults with major depressive disorder (MDD). Our group has found similar effects in MDD youth. However, this has not been studied in youth with treatment resistant MDD (TRD), nor has the interaction between regional N-acetyl-aspartate and volume deficits. N-acetyl-aspartate is an amino acid in the synthesis pathway of glutamate, and serves a marker of neuronal viability/number. Methods Fifteen typically developing youth (16–22 years of age; 7 males, 8 females) and eighteen youth with TRD (14–22 years of age; 8 males, 10 females) underwent 1 H-MRS and MRI on a 3 T scanner. A short echo PRESS protocol was used with voxels in the right and left hippocampi (6 mL each). Hippocampal volume was evaluated using FreeSurfer. Results Compared with the typically developing group, youth with TRD had lower concentrations of N-acetyl-aspartate in the left hippocampus (p=0.004), and a trend for smaller left hippocampal volume (p=0.067). In TRD subjects, hippocampal N-acetyl-aspartate was inversely correlated with left (r=−0.68, p=0.003) but not right hippocampal volume. Right hippocampal glutamate+glutamine was greater in TRD youth compared to typically developing controls (p=0.007). Conclusions These results suggest a neurochemical and structural deficit in the hippocampi of youth with TRD. These findings fit with the role of N-acetyl-aspartate in glutamate neurotransmission and the effect of glutamate on brain morphology.

To develop and evaluate a novel 3D Cartesian sampling scheme which is well suited for time-resolved 3D MRI using parallel imaging and compressed sensing.The proposed sampling scheme, termed GOlden-angle CArtesian Randomized Time-resolved (GOCART) 3D MRI, is based on golden angle (GA) Cartesian sampling, with random sampling of the ky-kz phase encode locations along each Cartesian radial spoke. This method was evaluated in conjunction with constrained reconstruction of retrospectively and prospectively undersampled in-vivo dynamic contrast enhanced (DCE) MRI data and simulated phantom data.In in-vivo retrospective studies and phantom simulations, images reconstructed from phase encodes defined by GOCART were equal to or superior to those with Poisson disc or GA sampling schemes. Typical GOCART sampling tables were generated in100ms. GOCART has also been successfully utilized prospectively to produce clinically valuable whole-brain DCE-MRI images.GOCART is a practical and efficient sampling scheme for time-resolved 3D MRI. It shows great potential for highly accelerated DCE-MRI and is well suited to modern reconstruction methods such as parallel imaging and compressed sensing.

Background Magnetic resonance imaging (MRI) has shown lower hippocampal volume in major depressive disorder (MDD). Preclinical and postmortem studies show that chronic stress and MDD may affect hippocampal subfields differently, but MRI spatial resolution has previously been insufficient to measure subfield volumes. Methods Twenty MDD participants (9 unmedicated and 11 medicated, both>6 months) and 27 healthy control subjects were studied. We used T2-weighted two-dimensional fast spin echo and T1-weighted three-dimensional magnetization prepared rapid acquisition gradient-echo sequences at 4.7 T to compare hippocampal subfield volumes at .09 μL voxel volume. Results Unmedicated MDD participants had a lower dentate gyrus volume than control subjects or medicated MDD participants and a lower cornu ammonis (CA1–3) volume in the hippocampal body subregion than control subjects. Conclusions Hippocampal volumes in unmedicated MDD showed evidence of localization to specific subfields and subregions, findings that appear, on the surface, consistent with preclinical evidence for localized mechanisms of hippocampal neuroplasticity. Strengths include in vivo measurement of entire hippocampal subfields and separation between unmedicated and medicated MDD. Limitations include power to control for multiple comparisons and that MRI landmarks approximate the subfields defined by cellular microstructure.