Purpose

The lateral geniculate nucleus (LGN) is a small (50-150 mm3) thalamic nucleus of the visual pathway processing motion. Precise depiction of LGN borders is needed to establish structural and anatomical alterations in ophthalmologic and neurodegenerative pathologies. Nevertheless, LGN volumes reported in the literature spread greatly. Therefore, we test the ability of 3D isotropic high resolution images with increased contrast to noise ratio (CNR) to improve LGN volume assessment against the clinical standard: proton density (PD) 2D TSE at 3T and the research standard: a motion corrected very high resolution 3D TFE at 7T.

Materials and Methods

First, the contrast between LGN and surrounding tissues in function of the MPRAGE inversion time (TI) was simulated to determine TIs for white matter (WM) anf grey matter (GM) signal nulling. Experimental validation was also performed.
Then, MPRAGE with WM signal nulled (WM0), MPRAGE with GM nulled (GM0), and PD images were obtained at 3T on 25 healthy subjects and at 7T on a subgroup of 14 healthy subjects selected prospectively, after approval by the ethics committee of the Canton of Zurich and written informed consent, with 32 channels receive head coils. The research gold standard developed within this study was a motion corrected segmented (7 dynamics) MPRAGE scan at 7T with maximized CNR and a resolution of 0.4x0.4x0.4mm3.
GM0 images were also subtracted from WM0 images to further increase contrast between LGN and surrounding tissues.
Three investigators segmented LGN independently and intraclass correlation coefficients (ICCs) were determined for volume and CNR quantification.

Results

The chosen TI provided optimal grey respective white matter signal nulling.
On coronal sections, at 3T and 7T, the LGN appeared as a small triangle in all MPRAGE images similarly as on PD images. The motion corrected 0.4x0.4x0.4mm3 segmented MPRAGE gave a very clear depiction of the LGN.
The CNR of the substracted WM0-GM0 images: 2.9±0.7 was higher than on other contrast: at 3T and 7T, and gave reliable estimations of the LGN volume and contrast at 3T and 7T. The CNR and the CNR after correction for voxel size and sqrt(acquisition time - TA) of the 3D TFE scan was higher than the PD 2D TSE at 3T and 7T. The 7T scanner had slightly higer CNR than the 3T scanner. Nevertheless, after correction for voxel size and sqrt(TA) the scanner effect disappeared, and the most efficient sequence was the WM0/GM0.
The LGN volumes determined at 3T were slightly smaller than the one determined at 7T (-7±2 mm3 p=0.003). The LGN volume was also smaller on the GM0 scan relative to the 3 other sequences (PD: -12±3, WM0: -9±1, WM0-GM0: -9±2 mm3 p<0.01). The correlation with the research standard, the motion corrected 0.4x0.4x0.4mm3 segmented MPRAGE, was significant (p<0.05) for all but one sequence and explained up to 50% of the variance. The mean LGN size was determined in the range 107±4 mm3. Male having a slightly bigger LGN. Age and laterality had consistent effect.
The raters’ ICCs were similar for all sequences for volume determination. For the CNR determination, the PD at 3T and the MPRAGE at 7T had the highest reliability. For LGN segmentation, all raters preferred the higher CNR and resolution of the MPRAGE scans. Nevertheless, thicker slices resulted in a good integrator and did not decrease the volume measurement accuracy – except for discretization noise due to the small number of slices covering the LGN for the PD scan at 7T.

Conclusion

High-resolution high-contrast 3D LGN images allow reliable LGN volume and signal determination. Our contrast optimization methods increased CNR and enabled thereby clinical applicability – also to other nuclei - and testing of clinical relevance.

Acknowledgements

No acknowledgement found.

References

No reference found.