In addition to the structural and functional imaging approaches described above, MR provides a quantitative, noninvasive approach for measuring endogenous and possibly introduced neurochemicals. Similar to the hydrogen nucleus (proton) that is the source of the MRI signal, certain other naturally occurring nuclei also have a magnetic resonance signal. Of particular biological significance are carbon (¹³C), nitrogen (¹⁵N), and phosphorus (³¹P) nuclei. As described in the MRI section, resonant nuclei induce a signal in the scanner detector. In contrast to the spatial information encoded in MRI, MRS focuses on the discrete resonant frequencies of specific nuclei, which are dependent on the applied magnetic field strength. Each nuclear isotope has a characteristic frequency range. For example at 9.4 T, hydrogen (protons) resonate around 400 MHz. In comparison, carbon resonates around 100 MHz. In contrast to the anatomic data generated from MRI, magnetic resonance spectroscopic (MRS) data usually come in the form of a spectrum where the distinct frequencies correspond to individual neurochemicals within a discrete volume of tissue (voxel; see Fig. 2). Each peak in the spectrum represents an individual chemical environment. Proton MRS voxels in animal models are usually chosen to be small enough to fit within a single brain hemisphere, typically about 10 mm³. Although multivoxel acquisitions are possible and are commonly used in larger human brains, they are technically more challenging in animal models.

The earliest MRS experiments in animal models of TBI used ³¹P-MRS to quantify adenosine triphosphate, adenosine diphosphate, phosphocreatine (PCr), inorganic phosphate (Pi), and the free phosphomonoester and phosphodiester components of membranes. By calibrating the precise position (chemical shift) of the Pi peak within the spectrum, a noninvasive estimate of the intracellular pH can be obtained. Similarly, since ATP binds magnesium, estimates of the magnesium concentration in vivo can also be made following careful measurements of the chemical shift of the ATP resonances. Following fluid percussion injury, the PCr/Pi ratio has been shown to decline and the tissue was acidotic, although no changes in ATP, indicating impaired energy metabolism, were detected (15). Other studies have shown reduced intracellular Mg²⁺ following TBI that is attenuated by intravenous supplementation. Of potential clinical significance, Mg2⁺ concentration was correlated with postinjury performance on the rotorod (16).

Hydrogen MRS, also known as proton MRS, may be used to quantify numerous endogenous neurochemicals such as N-acetylaspartate (NAA), creatine, PCr, myo-inositol, scyllo-inositol, glutamine, glutamate, glucose, glutathione, lactate, pyruvate, taurine, and GABA. Proton MRS is rather more complicated than ³¹P-MRS, since the tissue water signal, which dominates the neurochemical signals by up to seven orders of magnitude, must be suppressed. Moreover, since skull and subcutaneous fat signals are extremely strong and can interfere with resonances of greater interest, MRS signals must be localized to discrete brain regions. Fortunately, sophisticated acquisition sequences are available on commercial MR instruments to achieve the necessary suppression rates for high-quality spectroscopy.

Since MRS, like MRI, is noninvasive, longitudinal studies are feasible. Studies by Schuhmann et al. (17) have shown that NAA, a neurochemical produced in the neuronal mitochondria, shows a postinjury time-course of changes similar to histological evidence of neuronal pathology. Similarly, choline and myo-inositol, putative markers of inflammation, followed similar time-courses to those of histological markers of inflammation. Other MRS-visible neurochemicals such as glutamate and glutamine, which are involved in the glutamatergic neurotransmission system and glutamate excitotoxicity, and glutathione, an important anti-oxidant species, are likely to be informative in pathological processes. Accordingly, proton MRS might be useful in explorations of cellular and molecular mechanisms of TBI. Proton MRS also has considerable translational value since it is widely available on commercial clinical scanners for studies in humans. Several studies have shown that MRS neurochemical biomarkers such as NAA and choline are strongly correlated with cognitive recovery following human TBI (18, 19).

Other nuclei that are of biological significance, such as ¹³C and ¹⁵N, also have MRS signals. These isotopes are low in natural abundance and their weak endogenous signals are relatively uninformative. However, it is possible to introduce metabolites that are enriched with these isotopes and carry out tracer studies to determine the metabolic fate of specific chemical moieties. Studies by Bartnik et al. demonstrated upregulated pentose phosphate metabolism after moderate to severe TBI by following the metabolism of [1,2-¹³C]-labeled glucose as the label is incorporated into glutamate, glutamine, and lactate (20). Although the spectroscopic acquisitions in these particular experiments were carried out in extracts, they are feasible in the living brain in vivo. Studies using ¹⁵N-labeled species, although to date not implemented in TBI, have the potential of following amino acid pathways noninvasively.

Magnetic resonance spectroscopy in general has certain advantages as a tool for metabolic imaging. First, acquisition sequences for proton MRS are implemented on MRI scanners using the same hardware configuration as MRI. Accordingly, MRS acquisitions can be completed in the same scanning session as an MRI study and the data sets are inherently registered in three-dimensional space allowing direct comparison between structure and metabolism. Moreover, compared to the radionuclide studies described below, labeled MRS studies do not require radioactive isotopes meaning that longitudinal studies on individual animals are feasible. Another advantage of isotopically-enriched MRS is that the metabolic fate of specific labeled nuclei can be followed as the labeled compound is metabolized over time. In contrast, the sensitivity of MRS is lower than the various forms of radionuclide imaging resulting in poorer spatial and/or temporal resolution. Finally, although radiolabeled analogs of many drugs are available for PET and SPECT tracer studies, the relatively low concentrations of drugs mean that analogous labeled studies using ¹³C, for example, are likely not feasible.

Positron emission tomography (PET) is a functional imaging modality which tracks the localization of positron-emitting radioactive tracer molecules administered to the test subject. As the radioactive tracer decays, each emitted positron collides with an electron within the tissue nearby, producing 2 gamma rays traveling in opposite directions. These gamma rays are detected by an array of scintillation detectors surrounding the subject, and signals are fed to a computer, which localizes the sources of the gamma rays and combines these points into a 3-dimensional image. The term “micro-PET” refers to a high-resolution PET system designed specifically for imaging in small animal research models. The spatial resolution of micro-PET scanners currently approaches 2 mm.