Characterization of macromolecular baseline of human brain using metabolite cycled semi-LASER at 9.4 T

Purpose/Introduction: Macromolecular resonances (MM) arise mainly from cytosolic proteins overlapping with metabolites influencing metabolite quantification.This effect becomes even more severe in the case of short echo times (TE) due to their fast T2 relaxation time [1].Additionally, MM can be valuable biomarkers for several diseases and pathologies. The objective of this study was the characterization of human brain macromolecular baseline at 9.4T exploiting all the advantages arising from the utilization of UHF. Subjects and Methods: An adiabatic inversion pulse was developed for metabolite nulling at 9.4 T utilizing double inversion recovery [2] in combination with MC semiLASER [3] and appropriate coil configuration. MM spectra were acquired from 7 volunteers from two different brain locations, occipital lobe and left cerebral white matter. For each brain region, volunteer’s MM baselines were averaged creating a MM template which was parametrized using 15 simulated Lorentzian lines within LCModel. Quantification, corrected for different tissue composition and T1 relaxation, was performed for each MM component for each volunteer. Results: Double inversion recovery scheme for metabolite nulling at 9.4 T. A) Diagram of the sequence used for the acquisition of macromolecular spectra. Two adiabatic inversion pulses (AFP) preceded the actual localization scheme (MC semiLASER) utilizing optimizing inversion times (TI1 = 2360 ms and TI2 = 625 ms). B) Simulation of the effect of the double inversion recovery scheme in subplot A on the longitudinal magnetization for several T1 recovery times. T1 of macromolecules is assumed to be 400 ms and for other metabolites higher. Double inversion recovery scheme result in an efficient suppression of longer T1 metabolites. For inversion times TI1 and TI2,the fast T1 recovered MM resonances reach about 60 of their maximum magnetization while the longitudinal magnetization of metabolites with longer T1 relaxation are relatively suppressed (Fig. 1). Parametrization of smoothed and averaged MM baselines for two brain locations: (left) occipital lobe and (right) left cerebral white matter. Totally 15 MM components are observed. The final number of MM peaks included in the model were recognized visually and also utilizing previous reported MM resonances in the literature. However, in this study 15 MM peaks were used to fit adequately the cubic spline-smoothed averaged MM baseline templates (Fig. 2). For both brain regions, the same number of MM compartments was used. Quantification results of 15 MM for two different brain areas, occipital lobe and left cerebral white matter, from 7 volunteers using LCModel. (Up) Concentrations levels in arbitrary units (a.u), normalized using water reference scans and corrected for GM, WM and CSF percentages and the different T1 relaxation of the water compartments. (Down) CRLB values in a.u calculated as: concentration level 9 CRLB (). Quantification results did not show any statistically significant difference in MM concentration levels between occipital lobe and white matter left cerebral (Fig. 3). The only exception was the M1 peak, where there was a trend for potential difference (no statistically significant level after multiple comparisons correction). Discussion/Conclusion: The averaged MM baselines for both brain locations do not show substantial differences from a visual inspection (Fig. 2).Moreover, MM spectra among different volunteers exhibit similar patterns except for macromolecules M2, M3, and M4 which is also confirmed from the quantifications results (Fig. 3).Noteworthy, the two MM peaks at 2.57 ppm (M7) and 2.74 ppm (M8) are easily observable. These peaks appear in MM spectra of other studies [1, 4–7] however, they have not been reported. M7 and M8 are most likely due to the beta-methylene protons of aspartyl-groups within cellular proteins [4,]. Quantification results (Fig. 3) show that for both brain regions there are not significant differences, except a small trend for M1