Measurements of GABA and GSH with Magnetic Resonance Spectroscopy using HERMES and MEGA-PRESS


Report by

Gunnhild Ager-Wick


Magnetic resonance spectroscopy has been used to measure the concentration of GABA and GSH in healthy volunteers. They used two different spectral editing methods for finding this concentration. One method is called MEGA-PRESS where you send in specific pulse frequencies that either effect GABA or GSH and give the concentration of one of the two biochemical substances. HERMES use another technique where one can get the concentration of both GABA and GSH at the same time.

For this assignment, we will look at a collection of datasets from HERMES and MEGA-PRESS, and plot histograms with error bars using C++. The goal will be to see if the two methods give a good result with a small deviation. The dataset used in this assignment is from a bachelor thesis "MR measurements of GABA and GSH with new spectroscopy method: HERMES" written by Lars Sandnes and Tobias Heggeli.

The zip-file of this directory with all source files

Table of Contents

1. Introduction
2. Methods
3. Results
4. Discussion and Conclusion

1. Introduction

When one need information about an object, a picture may not always suffice. Images can be useful for identifying what something may be, but spectroscopy gives us the understanding of what an object actually consists of. Spectroscopy is by definition the study of how electromagnetic radiation interacts with matter. It was first used to study visible light from prisms and later to look at absorption lines from gasses, but in the 1950s one could by using a strong magnet, coils and radio waves detect changes in the frequency of the atomic nucleus due to their chemical bonds. This led to magnetic resonance spectroscopy, which is the MR-technique that presents information about biochemical compounds in an object. MRS has because of this become the clinical assessment of conditions such as epilepsy, multiple sclerosis and cancer (Jansen, Backes, Nicolay, & Kooi, 2006).

Figure 1.1: An example of sectroscopy;
Analysis of white light by dispersing
it with a prism.

1.1 Magnetic Resonance Imaging

Magnetic Resonance Imaging is an imaging technique that uses a strong magnetic field, around 1.5-3Tesla, that make a nucleus with a spin different to 0, like protons, align parallel to the exterior magnetic field. When the protons align with the outer magnetic field they start to resonate with a frequency that is called the Larmor frequency. The Larmor frequency is given by the equation below.


Where f0 is the frequency of the proton given in MHz, Y is the gyromagnetic ratio measured in MHz/Tesla and B0 is the external magnetic field. If we would send radiation with the same frequency as the proton then the proton would absorb this energy. This absorption gives the proton enough energy to flip away from the external magnetic field lines before it again loses its energy and start wobbling around the external field line like before.
Figure 1.2: A spinning proton precesses
around an applied magnetic field.

The time it takes for this relaxation to occur is called relaxation time. In MRI we have two different relaxation times: T1 and T2, where T1 describes how long it takes for 63% of the original signal to be restored in the z-direction while T2 describes how long it takes before there are only 37% of the flipped signal in the xy-plane. Different tissue has different relaxation time, and it is then possible to see different types of tissue in a T1-weighted and T2-weighted image.

Figure 1.3: T1- and T2-relaxation after a 90 degree rf pulse.

In addition to the external magnetic field, there are also three orthogonal gradient coils that can manipulate the field strength so that it varies linearly. In that way, the resonance frequency is never the same and one knows where the frequency comes from.

1.2 Magnetic Reconanse Spectroscopy

Magnetic resonance Spectroscopy (MRS) is similar to MRI, but instead of creating a picture we get a spectrum of the biochemical information in tissue. This allows for examination of individual molecules within a voxel, which means that one can study the biochemistry of disease processes without invasive procedures like biopsies (Brown and Semelaka, 2003, p. 181). MRS is based on the fact that different molecules are surrounded by different electron clouds that affect the resonance frequency of the protons in the molecule. This change in frequency is called a chemical shift and is measured in parts per million (ppm). The chemical shift values we get from the protons are measured up against a reference frequency of tetramethylsilane (TMS) that then has a chemical shift of 0. One can then measure the chemical shift of a tissue from the formula:

Chemical shift = (f-fref)*10^6/fref

Where fref is the reference frequency and f the resonance frequency to the proton in question. This makes it possible to separate substances from each other and make a spectrum. Figure 1.4 shows how an MR spectrum may look like. Each top represents one or more molecules, and the area under the graph give the consentrtion of the molecule.

Figure 1.4: A typicle MR spectrum form a voxel.

There are several ways of recording a spectrum from a voxel. But they all have in common that they still use gradients and RF pulses. Gradients in the MR machine must turn off their magnetic field so that it does not disturb the frequency of the individual protons, otherwise they are still used for picking ut frequencies from the protons in a tissue. GABA (gamma-Aminobutyric acid) and GSH (Glutathione) are to substances that both have a very low concentration that overlaps with other chemicals in the spectrum which makes it hard to detect. The methods MEGA-PRESS and HERMES, are specialised methods for finding the concentration of these substances, where MEGA-PRESS first measures the concentration of GABA and then GSH, while HERMES measures both substances at the same time. To be able to do this one wants to use a reference frequency that is stable in the human body. Two reference frequencies often used is creatine (Cr) and water (H2O).

The reason why one would want to know the concentration of GABA and GSH is that it can predict possible illnesses. GABA is the most important inhibitory neurotransmitter in the central nervous system in the brain, and its concentration has a connection to several mood disorders like schizophrenia (Dwyer, 2013). GSH, on the other hand, is an important antioxidant. Oxidative stress, which antioxidants fight against, appears to be connected with the loss of neurons during the progression of neurodegenerative diseases like Alzheimer's- and Parkenssondesis (Elsevier Science Ltd, 2000). It can therefore be important to investigate concentrations of GABA and GSH to get a diagnosis of a disease correctly.

2. Methods

This experiment where performed at Haukeland university hospital. The MR-machine used was a 3Tesla GE Discovery MR750 with 32 channel head coils to scan ten healthy volunteers ageing 23+-3 years. Single voxel spectroscopy, with dimensions 25mm*33mm*25mm, was positioned on the left side of the anterior cingulate cortex (ACC) in each participant.

Figure 2.1: T1-weighted picture of voxel in front of the left cortex

The volunteers first underwent anatomical imaging needed for finding a correct placement of the spectroscopy voxel. When the voxel was in place, the spectral editing method HERMES (TE/TR=80/2000ms) where first used for concurrent measurements of GABA and GSH, then measuring again but with MEGA-PRESS for GSH (TE/TR=131/1800ms) and later GABA(TE/TR=68/1800). The datasets where later analysed with a program called Gannet that measured the concentration of the different molecules in the voxel.

3. Results

The result from the scanning is presented in table 3.1 and table 3.2. The first table show the concentration of GABA and GSH compared with Cr in the voxel, and their respective mean values. An error in the scanning of GABA and GSH with H2O as a reference frequency lead to some higher valus of GABA and GSH, but the values where still calculated and their result is found in table 3.2.

GABA/Cr and GSH/Cr
Table 3.1: Results from GABA/Cr and GSH/Cr for
measurements with MEGAPRESS and HERMES

Table 3.2: Results from GABA/H2O and GSH/H2O for
measurements with MEGAPRESS and HERMES

To find the concentration of GABA and GSH one would use a program that fit the curves of the collected data with a reference dataset. The area under the graph would give the concentration. It is possible to calculate how well the reference dataset fit together with the collected data and estimate the error between them. The calculations of errors are presented in table 3.3.

GABA/Cr and GSH/Cr
Table 3.3: Fit error between collected data and reference
data baced on the creatine levels

The histograms in figure 3.1 shows the values of the dataset from table 3.1 and table 3.2 together with the uncertainty of each value taken from table 3.3 and the results based on the H2O-levels, which is not added in a table above because of an error in the sampling. This is to better illustrate the uncertainty of each value for this assignment. To find the uncertainty of each value the fit error had to be calculated for all volunteers and put into text files. The histograms were then created using c++ scripts that opened each text file.

Figure 3.1: Comparing the results from GABA/Cr, GABA/H2O, GSH/Cr and GSH/H2O for meassurements with MEGAPRESS and HERMES

Here are the scripts used for this project:


Here are the text files:


4. Discussion and Conclusion

Measurements of GABA and GSH overlap very well for both simultaneous measurements in HERMES and singular measurements in MEGA-PRESS. But from table 3.3 one can see that the GSH measurements with MEGA-PRESS have a bigger error than with HERMES, which could indicate that HERMES is a better method for measuring GSH values. This difference in error could also originate from the settings of T2, where it was set to be bigger for MEGA-PRESS then for HERMES. One could, therefore, conclude that both methods give good results, but that more measurements are needed for a final evaluation of which spectral editing method is best suited for measurements of GABA and GSH.


Jansen, J. F., Backes, W. H., Nicolay, K., & Kooi, M. E. (2006). 1 H MR spectroscopy of the brain: absolute quantification of metabolites. Radiology, 240(2), 318-332. doi: 10.1148/radiol.2402050314

Brown, M. A., Semelka, R. C. (2003) MRI Basic Principles and Applications. New Yersey: John Wilwy & Sons, Inx., Hoboken.

Elsevier Science Ltd (2000). Metabolism and functions of glutathione in brain. Hentet fra:

Dwyer, G. (2013) Quantitative measurement of GABA with clinical MR-Spectroscopy, Department of Biomedicine, University of Bergen

Refsnes Data (2018) w3schools [Internet] Avalible from: [12. June 2018]