Deuterium metabolic imaging in the human brain at 9.4 Tesla with high spatial and temporal resolution

Abstract

Purpose: To present first highly spatially resolved deuterium metabolic imaging (DMI) measurements of the human brain acquired with a dedicated coil design and a fast chemical shift imaging (CSI) sequence at an ultrahigh field strength of B₀ = 9.4 T. ²H metabolic measurements with a temporal resolution of 10 min enabled the investigation of the glucose metabolism in healthy human subjects.

Methods: The study was performed with a double-tuned coil with 10 TxRx channels for ¹H and 8TxRx/2Rx channels for ²H and an Ernst angle 3D CSI sequence with a nominal spatial resolution of 2.97 ml and a temporal resolution of 10 min.

Results: The metabolism of [6,6'-²H₂]-labeled glucose due to the TCA cycle could be made visible in high resolution metabolite images of deuterated water, glucose and Glx over the entire human brain.

Conclusion: X-nuclei MRSI as DMI can highly benefit from ultrahigh field strength enabling higher temporal and spatial resolutions.

Keywords: 9.4 Tesla; DMI; Deuterium; Human brain; Oral glucose administration; TCA cycle.

Fig. 1

Coil design and corresponding ²H B₁ distribution of a phantom measurement. The first row (A) shows the design of the ²H/¹H double-tuned RF coil. Below the transmit field B₁⁺images for the ²H channels of the double resonant ²H/¹H phased array coil measured at a phantom are shown (B; length of the phantom 18 cm; axis of 19 cm x 15 cm). Below, ²H signal amplitude images of the same phantom are shown (C). The signal amplitudes are given in [i.u.].

Fig. 2

Measurement of T₁ relaxation times. Shown are the measured amplitudes of the water resonance with fitted relaxation time curves for all measured volunteers (n = 4) (A) as well as a representative spectrum acquired at a single time point from one volunteer (B). The T₁ measurement was performed with a non-localized inversion-recovery sequence.

Fig. 3

Non-localized ²H MRS with high temporal resolution. For two different volunteers, signal amplitudes with a temporal resolution of 2 min were measured with a non-localized FID pulse acquire sequence. Detectable ²H metabolites are water, glucose and glutamate/glutamine (Glx). In addition, a fourth peak is visible which can be assigned mainly to lipid contaminations. A shows the amplitude-time curves for the four resonances for all three volunteers. The presented errors are calculated as Cramer-Rao lower bounds. B shows exemplary spectra for different time points for the volunteer.

Fig. 4

In vivo SNR distributions. Seven different volunteers were measured with the MRSI protocol. Shown are the corresponding SNR maps for the natural abundant water signal for different transversal slices for all volunteers. SNR was calculated as the fitted signal amplitude in the time domain divided by the standard deviation of the time domain signal from a voxel outside of the human head.

Fig. 5

Exemplary in vivo spectra. Four different spectra from a single volunteer from positions indicated in the anatomical image in the left upper corner are presented. The shown spectra were measured approx. 90 min after administration of ²H-labeled glucose.

Fig. 6

Temporal resolved DMI of a single volunteer and sagittal, coronal and transversal direction. The first rows show the white matter, CSF and gray matter fractions without (A) and with (B) accounting for the point-spread function (PSF). In C, the DMI maps for water, glucose and Glx are shown. All metabolite images were normalized by the natural abundant water signal. At the left side, the begin times of the DMI measurement are stated. The images were zero-filled to two times the original resolution. The amplitudes are given in relative units [i.u.]/ [i.u.] = 1.

Fig. 7

Temporal resolved DMI for different volunteers. Shown are the ²H metabolic maps of water, glucose and Glx relative to the water reference measurement for two different volunteers and different time points (skipping every other; the starting time of the ²H MRSI measurement are given). The images were zero-filled to two times the original resolution. The amplitudes are given in relative units [i.u.]/ [i.u.] = 1.

Fig. 8

²H metabolic signal uptake in GM and WM. The left side (A) shows the signal uptake spectra for one volunteer summed over all voxel with either WM or GM content. The left side (B) shows the measured amplitudes from the summed tissue specific voxel for WM and GM for all volunteers as well as fitted curves for all volunteers (n = 7) individually as well as fitted through all data points (bold). For water, a linear model used. For glucose, a bi-exponential fit was applied (f(t) = A₁ exp(- k₁ t)) – A₂ + (A₂ – A₁) exp(-k₂ t)) and for Glx, a mono-exponential function was fitted to the data points (g(t) = A₁ - (A₁ – A₂) exp(-k t)).

Fig. 9

Difference spectra of tissue with a high content of white matter (WM) and gray matter (GM) for the different measured time points after the oral administration of deuterated glucose relative to the reference measurement before the administration of labeled glucose. The third image shows the relative difference in the uptake of the different ²H-labeled metabolites between different voxels with a high GM and WM content at different time points. The difference spectra are summed spectra over all volunteers (n = 7).

Fig. 10

Signal evolution of the resonance at 1.4 ppm. The upper row (A) shows the estimated signal amplitudes of the 1.4 ppm resonance for averaged voxels over from inside the brain for all volunteers (n = 7) and different time points as well as exemplary averaged spectra for different time points from a single volunteer. The lower row (B) shows the corresponding signal amplitudes and spectra for tissue with a high skull tissue fraction.