Lateral geniculate nucleus volumetry at 3T and 7T: Four different optimized magnetic-resonance-imaging sequences evaluated against a 7T reference acquisition

Purpose: The lateral geniculate nucleus (LGN) is an essential nucleus of the visual pathway, occupying a small volume (60–160 mm3) among the other thalamic nuclei. The reported LGN volumes vary greatly across studies due to technical limitations and due to methodological differences of volume assessment. Yet, structural and anatomical alterations in ophthalmologic and neurodegenerative pathologies can only be revealed by a precise and reliable LGN representation. To improve LGN volume assessment, we first implemented a reference acquisition for LGN volume determination with optimized Contrast to Noise Ratio (CNR) and high spatial resolution. Next, we compared CNR efficiency and rating reliability of 3D Magnetization Prepared Rapid Gradient Echo (MPRAGE) images using white matter nulled (WMn) and grey matter nulled (GMn) sequences and its subtraction (WMn‐GMn) relative to the clinical standard Proton Density Turbo Spin Echo (PD 2D TSE) and the reference acquisition. We hypothesized that 3D MPRAGE should provide a higher CNR and volume determination accuracy than the currently used 2D sequences. Materials and methods: In 31 healthy subjects, we obtained at 3 and 7 T the following MR sequences: PD‐TSE, MPRAGE with white/grey matter signal nulled (WMn/GMn), and a motion‐corrected segmented MPRAGE sequence with a resolution of 0.4 × 0.4 × 0.4 mm3 (reference acquisition). To increase CNR, GMn were subtracted from WMn (WMn‐GMn). Four investigators manually segmented the LGN independently. Results: The reference acquisition provided a very sharp depiction of the LGN and an estimated mean LGN volume of 124 ± 3.3 mm3. WMn‐GMn had the highest CNR and gave the most reproducible LGN volume estimations between field strengths. Even with the highest CNR efficiency, PD‐TSE gave inconsistent LGN volumes with the weakest reference acquisition correlation. The LGN WM rim induced a significant difference between LGN volumes estimated from WMn and GMn. WMn and GMn LGN volume estimations explained most of the reference acquisition volumes' variance. For all sequences, the volume rating reliability were good. On the other hand, the best CNR rating reliability, LGN volume and CNR correlations with the reference acquisition were obtained with GMn at 7 T. Conclusion: WMn and GMn MPRAGE allow reliable LGN volume determination at both field strengths. The precise location and identification of the LGN (volume) can help to optimize neuroanatomical and neurophysiological studies, which involve the LGN structure. Our optimized imaging protocol may be used for clinical applications aiming at small nuclei volumetric and CNR quantification.