Chapter 3. Overview of experimental techniques and methods
Magnetic resonance imaging
Components of the following section have been reported in shorter form as a book chapter in Magnetic Resonance Spectroscopy: Tools for Neuroscience Research and Emerging Clinical Applications edited by C. J. Stagg and D. L. Rothman, Turner and Gant, Chapter 2.2 – The Biochemistry of Creatine. Copyright © Elsevier (2014); 91-103.
Nuclear magnetic resonance characterises a phenomenon whereby radio frequency waves are both absorbed and then emitted by nuclei when placed in a magnetic field. First described by Rabi (Rabi et al., 1938; Rabi et al., 1992) and further extended by Bloch and Purcell (Bloch, 1946; Purcell et al., 1946), it was recognised that certain nuclei will resonate when placed in a strong magnetic field and applied with radiofrequency energy. As magnetic properties differ between nuclei and the environment that they find themselves in, their behaviour when placed in an external magnetic field provides a very sensitive measure of their chemical environment. As such, MRI allows for the detection of certain nuclei and provides information regarding the chemical composition of scanned tissue.
Magnetic resonance imaging permits the non-invasive measurement of the structure, function, and chemical composition of physiological samples. Structural MRI, functional MRI, and MRS encompass the neuroimaging techniques that have been used throughout this thesis to assess structural and functional aspects of neural tissue. In this section fundamental magnetic resonance (MR) principles are reviewed as they apply to conventional MRI. Similar concepts apply to other forms of MRI but the material covered here is not exhaustive. The review focuses on the acquisition and analysis of spectroscopy data as it is a key technique and skill developed throughout this thesis.
Principles of MR
Obtaining the MR signal
Certain nuclei within tissue possess tiny magnetic moments and when in the presence of an external magnetic field, undergo a rotational motion known as precession (Bloch, 1946; Purcell et al., 1946). Precession is related to the magnetic moment of the nucleus and its intrinsic spin, a fundamental property of nuclei that contain an unpaired number of protons or neutrons (Hendee and Morgan, 1984; Santarelli, 2005). When in the presence of an external magnetic field (B), precession is generated by the coupling of the nuclei’s intrinsic spin together with the force of the applied external field in an attempt to align the magnetic moment of the nucleus with B, where energy is at a minimum (Santarelli, 2005; Storey, 2006). The frequency at which the nucleus precesses about the magnetic field is called the Larmor frequency which is proportional to the strength of the external magnetic field. In a stronger external magnetic field, more magnetic energy is available to pull the nuclei into alignment, resulting in a greater tendency for nuclei to align to B and creating a tighter frequency of precession. The net magnetic field of all nuclei within a given volume of tissue equates to the sum of all of the individual nuclei magnetic moments, called the nuclear magnetisation. This creates an oscillating magnetic field with a net magnetisation parallel to B, also known as the longitudinal direction measured in the z-coordinate defined by the conventional Cartesian coordinate system (Hendee and Morgan, 1984; Santarelli, 2005; Storey, 2006).
The net magnetic field that is experienced by a nucleus is influenced by the external magnetic field that it is placed in and the smaller magnetic fields surrounding the nucleus. If the nucleus is part of a chemical compound, the surrounding electrons tend to shield the nucleus from the external magnetic field. This results in a slight change in the precessional frequency of the nucleus that is characteristic of the molecular group in which it resides and contributes to the specific frequency of the MR signal produced by the nucleus, known as the chemical shift (Storey, 2006; Juchem and Rothman, 2014).
Measuring the MR signal
In order for the signal encoding the specific nuclei to be detected, the nuclear magnetisation must be tipped out of alignment with the longitudinal plane. This is achieved via the application of a secondary magnetic field in a transverse plane, known as a radio frequency (RF) or B1. The RF must match the Larmor frequency in order to transfer energy to the nuclei and tip it out of alignment with B into the transverse plane (measured in x- and y-coordinates; Santarelli, 2005; Storey, 2006). As the energy from the B1 field is absorbed by the nuclei, the amplitude of its transverse magnetisation gradually increases. The RF is applied in a short intense burst so that when the B1 field is terminated, the energy is emitted from the nuclei, the nuclear magnetisation returns back into alignment with B, and the transverse magnetisation decays (Hendee and Morgan, 1984). This excitation of the nuclei and their subsequent synchronous precession produces a rotating magnetic field which is detected by coils tuned to receive the RF. As the magnetisation rotates into the transverse plane, the magnetic flux through the coil oscillates inducing a small alternating voltage – the MR signal, which is proportional to the transverse component of the nuclear magnetisation. The maximum amount of signal that can be measured is attained when all the longitudinal magnetisation is transferred into the transverse plane (Hendee and Morgan, 1984; Santarelli, 2005; Storey, 2006).
The signal does not persist indefinitely. Through interactions with neighbouring nuclei and molecules, the transmitted energy is lost from the nuclei which eventually returns to equilibrium with B, a process termed relaxation (Hendee and Morgan, 1984; Santarelli, 2005; Storey, 2006). Longitudinal or spin-lattice relaxation, T1, describes the time that it takes for the nuclear magnetisation to return to the longitudinal plane and is related to the rate of energy loss. It is a measure of the degree of longitudinal magnetisation present and is referred to as the recovery of longitudinal magnetisation. Transverse or spin-spin relaxation, T2, describes the time it takes for the MR signal to disappear. It is a measure of the degree of transverse magnetisation and is referred to as the signal decay in the transverse plane. Transverse relaxation occurs more rapidly than longitudinal relaxation due to a loss of phase coherence amongst the individual nuclei spins, causing the nuclei to precess at slightly different rates and generating separate components of spin-spin relaxation. The component of dephasing that is related to intrinsic microscopic factors, such as molecule size and tissue type, produces the distinct dephasing timescale termed T2. Magnetic field inhomogeneities can create dephasing over a larger spatial scale, this timescale is termed T2* (Hendee and Morgan, 1984; Santarelli, 2005; Storey, 2006). The signal acquired, and subsequently the values of the longitudinal and transverse relaxation times, depends on the molecule and tissue that the nuclei are contained in and can be used to recreate images of scanned tissue.
Conventional and structural MRI
Conventional or structural MRI produces a high resolution image of a scanned region which aids in the assessment of internal anatomy. Considering tissue is made up of nuclei, the MR properties mentioned above can be exploited to non-invasively reproduce an image of the tissue structure.
Image generation and reconstruction
Hydrogen is the most common nuclei used for imaging due to its excellent sensitivity to MR methods and high concentration in biological tissues (Mitchell and Cohen, 2004; Santarelli, 2005; Storey, 2006). The MR signal produced from hydrogen originates predominately from water, which constitutes a large portion of the body’s tissues. The discrepancies in relaxation time among tissue types that contain differing amounts of hydrogen provides an important source of signal contrast used to generate images that map the anatomy of tissue – the foundation to structural MRI. Biological fluids, such as blood and CSF, tend to have long T1 and T2 relaxation times as a result of the relatively unrestricted motion of water molecules and more efficient retention of energy induced by the RF pulse. In contrast, solid tissue exhibits shorter relaxation times as the water component in these tissue types is in frequent contact with macromolecules. This allows for more efficient transmission of energy between the protons and their environment, such as that which occurs in grey matter and even more so in white matter (Hendee and Morgan, 1984; Mitchell and Cohen, 2004; Storey, 2006). An assessment of the signal emitted from these different types of tissue demonstrates that there is sufficient contrast in the relaxation times to obtain morphological depiction in soft tissue.
The generation of an image is achieved by both slice-selective excitation of nuclei and encoding spatial information into the phase and frequency of the signal. The application of a precise RF pulse enables the excitation of specific nuclei that lie within the plane of the induced magnetic field. This process ensures that only nuclei in any one slice of tissue are excited at any one time. During data acquisition, magnetic field gradients are applied to encode positional information into the frequency and phase of the signal emitted, which aids in further localising the signal within each excited slice. The time between RF excitation and acquisition of the data is known as the echo time (TE). The repetition time (TR) defines the time between each RF excitation. These procedures are repeated with the phase-encoding direction incremented from one repetition to the next in order to collect data with localised spatial information from the entire field of view (FOV) and reconstruct an image of the tissue of interest. The combination of repeated RF pulses and signal acquisitions coordinated with magnetic field gradients are known as an imaging sequence (Hendee and Morgan, 1984; Mitchell and Cohen, 2004; Santarelli, 2005; Storey, 2006).
The contrast between tissue types can be manipulated by the imaging sequence and parameters used to acquire the signal. Acquisition parameters that are under the control of the operator and can be utilised to modify image contrast include the timing parameters of the pulse sequence and the flip angle induced by the RF pulse, which quantifies the degree to which net magnetisation is tipped away from the longitudinal axis during excitation. The value of such parameters determines the degree to which T1, T2, or T2* relaxation time is emphasised in the acquired image, producing T1-, T2-, or T2*-weighted images, respectively (Mitchell and Cohen, 2004; Storey, 2006).
Gradient-echo and spin-echo sequences are the most common pulse sequences used. A gradient-echo sequence involves the delivery of a single RF for each TR, with the data acquired during the subsequent decay of the signal. In gradient-echo sequences, the value of TE is usually a few milliseconds, with TR generally several milliseconds long. In a spin-echo sequence, two RF pulses are delivered in each TR, with data acquired during the spin-echo when the spins are refocused following the second RF pulse. Here, TE typically ranges between a few to hundreds of milliseconds while TR is generally very long, in the order of hundreds of milliseconds to several seconds. Gradient-echo and spin-echo sequences are optimised to produce T2*- and T2-weighted images respectively, with both sequences capable of providing T1-weighting. Typically, T1-weighting is produced when images are acquired with short TR as the signal amplitude that is acquired is higher for tissues with short T1 relaxation times. In contrast, images acquired with long TE generate T2– or T2*-weighted images as the signal amplitude at acquisition is higher for tissues with long T2 or T2* relaxation times (Mitchell and Cohen, 2004; Santarelli, 2005; Storey, 2006).
Hardware necessary for MR data collection includes a primary magnet, gradient and shim coils, RF coils, and a computer system. The primary magnet is the central component of an MR system as it generates the B magnetisation, the strength of which is proportional to the static magnetic field. Magnetic field strength is measured in tesla (T), with most clinical scanners capable of generating field strengths of 1.5 or 3 T. Homogeneity of the B field is maintained by gradient and shim coils. Inhomogeneities may exist from manufacturing imperfections in the system, metallic structures in the building, and susceptibilities in the sample itself. It is vital that any inhomogeneities are addressed through a process called shimming. Shimming is achieved by producing compensatory magnetic fields in the gradient and shim coils to correct for spatial variations in the main magnetic field. Gradient coils also allow for volume selectivity when scanning, and encode spatial information into the signal allowing an image to be acquired. Radiofrequency coils produce the B1 field to transmit energy and excite nuclei. They also contain receiver coils that detect and process the signal emitted from excited nuclei. Finally, a computer is necessary to facilitate the acquisition and process the results (Hendee and Morgan, 1984; Storey, 2006).
Considerations and safety
A range of acquisition parameters contribute to the versatility of MRI, with the choice largely dependent on the purpose of the scan. As mentioned previously, timing parameters and the chosen flip angle will dictate the contrast of the image produced by virtue of emphasising either T1-, T2– or T2*-weighted relaxation times of the scanned tissue. Spatial parameters such as the size of the FOV, the thickness of each slice that is acquired, and the voxel dimensions govern the resolution of the image and can be manipulated to enhance the visual discrepancy between anatomical structures when distinct definition is necessary. Furthermore, all of these parameters influence the amount of signal that is detected above background noise, known as the signal-to-noise ratio (SNR). Image SNR can be controlled independently by the number of averages or repetitions of data acquisition that are made. However, although increasing the number of excitations enhances the amount of signal that is detected, it in turn lengthens scan time (Mitchell and Cohen, 2004; Storey, 2006).
Magnetic resonance techniques are considered safe and pose no risk to the health of people, animals or biological samples provided that primary safety precautions are observed. Potential health risks are largely associated with the high-intensity static magnetic field, however the generation of small electric fields and possible heating of scanned tissue could also potentially occur (Hendee and Morgan, 1984; Storey, 2006). The external magnetic field used to align nuclei and generate the image is extremely strong. As such, it is necessary to consider that ferromagnetic objects will be attracted to the field and have the potential to become projectiles when in the vicinity of the scanner. The strong magnetic field is also capable of disrupting the operation of electronic devices, such as pace makers, that may be introduced into the scanning environment. Furthermore, introduced metallic objects can spoil the homogeneity of the magnetic field and subsequently degrade the image that is produced. As such, ferromagnetic materials must be considered prior to scanner use. It is essential that all such objects are removed from the scanning environment to reduce risk, with screening for metallic and electronic implants considered a crucial procedure prior to entering the scanning environment. Nerve stimulation may be induced as a result of small transitory voltages within tissue from changes in magnetic field gradients. However, the potential for this is minimal as most scanners have internal constraints on the rate at which gradients can be changed. Induced currents from magnetic field gradients and radiofrequency pulses have the potential to generate heat in the scanned tissue. Though, the heat generated is not significantly greater than that produced naturally via basal metabolism or heavy exercise and as such does not strain physiological mechanisms of heat loss (Hendee and Morgan, 1984). It is common for clinical and experimental MRI scanners to contain alarms that the patient can ring from within the scanner if they feel uncomfortable, immediately alerting the experimenters to any such issues and stopping the scanning procedure.
Currently, there are two conditions considered as strict contraindications to MRI: pregnancy and dependence on a cardiac pacemaker. There are no known effects of magnetic field exposure to a growing fetus, however limits are in place to avoid any unknown risks that may be involved. Recent research suggests that if standard precautions are met, MRI is safe for both the mother and fetus (Patenaude et al., 2014). The second contraindication is motivated by the potential for the external magnetic field to disrupt the operation of cardiac pacemakers. However, the frequency of device failures recorded at the common clinical field strength of 1.5 T has been reported as similar to standard device failure rates (Cohen et al., 2012) suggesting that the presence of the magnetic field may not be as detrimental as initially considered.
The assessment and mapping of brain activity and function can be measured via two types of techniques: those that localise the underlying electrical activity of the brain, such as electroencephalography and magnetoencephalography, and those that are sensitive to local physiological or metabolic consequences of altered brain function. Functional MRI methods make up the latter category and are sensitive to changes in regional blood perfusion, blood volume, or blood oxygenation, all of which accompany neuronal activity. The assessment of blood oxygenation has been used in the current work and will be the topic of this section.
Principles of fMRI
The physiological basis of BOLD fMRI is associated with changes in blood oxygenation. Neuronal activity is reliant on energy in order to function. During times of enhanced neuronal activity, substrate delivery for energy metabolism is augmented via increased blood flow – known to reflect local signalling activities (Attwell and Iadecola, 2002). In order to meet the required demand, blood flow increases to a greater extent than is necessary to provide more oxygen and glucose for the increased energy production. As a result, oxygen extraction decreases with greater neuronal activity (Matthews and Jezzard, 2004; Storey, 2006).
A reduction in oxygen extraction leads to an increase in the ratio of oxy- to deoxyhaemoglobin in the circulation throughout regions of neuronal activation. The origin of the associated BOLD response or fMRI signal change arises from the different magnetic properties of haemoglobin-carrying oxygen (oxyHb) and deoxygenated haemoglobin (deoxyHb). Deoxygenated haemoglobin is slightly paramagnetic relative to neural tissue, while oxyHb is isomagnetic (Pauling and Coryell, 1936). Minute distortion of the local magnetic field results from blood rich in oxyHb.
Conversely, the local magnetic field of capillaries and veins carrying deoxygenated blood tends to be relatively distorted (Ogawa et al., 1990; Turner et al., 1991; Storey, 2006). These microscopic field inhomogeneities associated with the presence of deoxyHb cause an interference to the MR signal as a result of dephasing among the water protons, and subsequent shortening of the T2* relaxation time. When oxygen extraction falls as a result of enhanced blood flow to regions with increased neuronal activity, T2* becomes longer and the MR signal intensity is measurably larger relative to that recorded from a baseline resting state (Kwong et al., 1992; Ogawa et al., 1992; Matthews and Jezzard, 2004).
Echo-planar imaging (EPI) is sensitive to magnetic field inhomogeneities and as such is the most common imaging sequence used to collect BOLD fMRI data, which allows for acquisition of data from the whole brain in a few seconds. Rapid acquisition sequences are vital for functional imaging as they capture real-time changes in the haemodynamic response to brain activation throughout the entire brain volume. This occurs within the space of seconds (typically 1 – 6 seconds) depending on the timing parameters used, allowing for a synchronous measurement (Turner et al., 1998). Using a single RF pulse, T2* relaxation data are rapidly acquired generating a whole volume image, known as single-shot EPI (Mitchell and Cohen, 2004). As data are collected quickly, the spatial resolution that can be gained within a regular amount of time is considerably lower than that acquired with structural imaging – generally on the order of 4 mm3 (Matthews and Jezzard, 2004).
The change in the intensity of the BOLD signal that occurs with brain activation is extremely small in relation to background noise. As a result, it is essential that a large number of averages are collected in order to maximise the amount of signal collected and enhance the SNR. Typically, these images are acquired while the participant performs a task that cycles between one or more defined states which require a change in brain function, such as rest or baseline and task performance. The time course of the signal in each voxel is then correlated to the time course of the task in order to identify those voxels in the brain that demonstrate changes associated with the brain function that is being examined. As such, BOLD fMRI measures relative signal intensity changes associated with different cognitive states (Matthews and Jezzard, 2004).
Table of contents
List of Figures
List of Tables
List of Abbreviations
Chapter 1. Introduction
Chapter 2. Review of the literature
2.1. The biochemistry of Cr
2.2. Creatine supplementation
2.3. Hypoxia in the CNS
2.4. Traumatic brain injury
Chapter 3. Overview of experimental techniques and methods
3.1. Magnetic resonance imaging
3.2. Non-invasive brain stimulation
3.3. Neuropsychological assessment
3.4. Dietary Cr supplementation
3.5. Experimental hypoxia
Chapter 4. Comparative quantification of dietary supplemented neural creatine concentrations with 1H-MRS peak fitting and basis spectrum methods
Chapter 5. Absolute quantification of neural metabolites using magnetic resonance spectroscopy
Chapter 6. Acute hypoxic gas breathing severely impairs cognition and task learning in humans
Chapter 7. Creatine supplementation enhances corticomotor excitability and cognitive
performance during oxygen deprivation
Chapter 8. Effects of anodal tDCS on corticomotor plasticity during acute hypoxia and creatine supplementation
Chapter 9. A feasibility study assessing the effects of creatine supplementation on cognition
during hypoxia in mild traumatic brain injury
Chapter 10. General discussion
GET THE COMPLETE PROJECT
Effect of dietary creatine supplementation and hypoxia on motor and cognitive function