Magnetic resonance spectroscopy of the normal cerebellum: what degree of variability can be expected?

The objectives of this paper are (a) to establish the reliability of relative metabolite concentrations determined with (1)H-MR spectroscopy of the cerebellum using a method appropriate to the constraints of clinical radiology and (b) to record normal values for metabolites within the cerebellum and to look for differences in metabolite concentrations between the cerebellar hemispheric white matter and the superior vermis. 3-T (1)H-MR spectra were obtained from voxels positioned in the right cerebellar hemispheric white matter and the superior vermis in 55 healthy adults (mean age 41 years, range 20 to 77) using a single voxel PRESS sequence (TR/TE = 2,000/144 ms). One volunteer (male, age 34 years) was examined in six separate sessions over a period of 3 weeks. Reliability of intra- and inter-subject metabolite fitted area ratios was determined by evaluating coefficients of variance (%). Inter- and intra-subject coefficients of variance (%) in metabolite ratios were consistently lower in the vermis (4 to 11.6) compared to those of the hemisphere (7.2 to 14.3). Cho/Cr was significantly higher in the vermis (0.83 ± 0.10) compared to the hemisphere (0.76 ± 0.11) and NAA/Cho was significantly lower in the vermis (1.19 ± 0.12) compared to the hemisphere (1.35 ± 0.16). Low inter- and intra-subject variability can be achieved when using a (1)H-MR spectroscopy technique that is appropriate to the time constraints of clinical radiology. The regional variations of Cho/Cr and NAA/Cho within the hemisphere and vermis should be considered when performing studies of diseases, which may preferentially target a particular cerebellar location.