MAGNETIC RESONANCE SPECTROSCOPY (MRS)

MRS is an imaging technique used to perform molecular analysis on brain tissue. This MR modality allows to evaluate and to quantify the relative concentrations of different metabolites such as N-Acetylaspartate (NAA), choline (Cho), creatine (Cr), myoinositol (mIns), lipids, and lactate. MRS is a powerful tool for the assessment of metabolic disorders, planning of image-guided biopsy, and for the characterization, classification, and follow-up of brain lesions. Moreover, multiple studies support its use as a biomarker in the early and differential diagnosis in dementing processes. Using current protocols it is possible to obtain more robust results, however, it is crucial to take into account multiple physiological and technical factors of the MRS studies; as they may alter the spectral information being critical during interpretation of the images and final diagnosis.

LOW SPECIFICITY

A normal pattern is characterized by a positive slope between Cho and NAA peaks. A negative slope between these metabolites would suggest an abnormal process. Despite the high sensibility of MRS identifying these molecular changes, this technique has a low specificity as this abnormal pattern may be related to low or high-grade glioma, meningiomas, radionecrosis, infarcts, multiple sclerosis, metastasis, even normal in newborns.

In order to perform an accurate diagnosis, it is highly important to perform an acquisition of MRS curves in healthy brain tissue. Knowledge of the clinical condition and treatment are also important for an accurate diagnosis. Conventional MR, in conjunction with advanced MR such as DWI and PWI, should be considered in the assessment of pathologies associated with the CNS as they improve the predictive value of the MRS.

PARTIAL VOLUME 
Spectral curves of small brain lesions may appear normal because of partial volume effect. Partial volume is an artifact produced by an inadequate relationship between the brain lesion size and the voxel size. This artifact is one of the main causes of false-negative results, and it may be avoided with an adequate acquisition. MRS is not adequate for the assessment of all brain lesions, the minimum diameter should be 1.0 - 1.5 cm in order to avoid partial volume artifact. Voxel size and position of the voxel of interest is also important. If the selected voxel is too large compared to the lesion size or the position is not adequate, the spectral curves will not show an accurate metabolic measurement of the lesion, as non-lesional tissue will be mixed with lesional tissue within the same voxel. 

Fig. 1: Fifty-years-old female patient with a convulsive syndrome. MRS with single-voxel technique is performed in order to characterize and to classify the brain lesion observed in the structural images. A. Spectrum from MRS demonstrates a normal Cho/NA ratio of 0.98. If the voxel is correctly positionated covering the whole lesion, and avoiding partial volume effect (B), the spectrum from MRS shows a Cho/NAA ratio of 2.06, suggesting a low-grade tumor.

The spectral curve of the figure 1.a shows a spectral curve with Cho and NAA peaks, with a ratio Cho/NAA equal to 0.98, suggesting no alteration in the relationship of the metabolites in the selected tissue. However, the amount of healthy tissue within the voxel is greater than the brain lesion creating a partial volume artifact. When the voxel is repositioned (figure 1.b), a negative slope is obtained, with a ratio Cho/NAA of 2.06, suggesting low-grade brain lesion.

SCALE FACTOR AND SPECTRAL CURVES OF HEALTHY BRAIN TISSUE

When one single metabolite is highly altered, it may induce misinterpretation about the concentration of the other metabolites as the spectral curve appears to be flattened.

Fig. 2: MRS qualitative analysis in the tumor tissue (B) showed the Cho increased and NAA decreased compared with the normal tissue (A); suggesting a high-grade tumor. If the MRS spectrum of the tumor tissue is rescaled (C), the analysis shows a Cho conserved, and a NAA decreased, findings related to a low-grade tumor. The scale of the MRS curves, and a reference in normal tissue are highly important to perform an accurate interpretation of the MRS studies.

Figure 2 shows MRS curves of healthy brain tissue (figure 2.a), and brain lesion tissue (figure 2.b). The reduction of the Cr and NAA concentrations suggests a tumor; these metabolites are associated with energy demand and neuronal loss. Also, Cho appears to be increased, suggesting a high-grade tumor with cell proliferation. When the scale is adjusted (figure 2.c), NAA and Cr remain decreased, however, the concentration of the Cho is not increased. These latter findings are more suggestive of low-grade tumor. The histologic result, in this case, was oligodendroglioma II-grade.

NORMAL VARIATION

Multiple studies have reported that the concentration of metabolites vary in relationship to the patient age. Before the age of 4 months, the levels of Cho and mIns are higher than NAA levels due to active myelination. This tends to normalize after 2 years-old, where the spectral curves are identical to that of adults. In the elderly, a decrease in the NAA levels and NAA/Cr, and an increase in the Cho level and Cho/Cr ratio may be observed, suggesting that there is a reduction in the number of neurons related to a degradation of the membranes, or an increase in the number of the glial cells. 

Fig. 3: Normal variation according to the age. In newborns, the levels of Cho and mIns are higher than NAA levels due to active myelination (a). This tends to normalize after 2 years-old, where the spectral curves are identical to that of adults (b).

The concentration of the metabolites may also vary in different brain regions. The NAA level is similar in the white matter and gray matter; however, its level is lower in the hippocampus and cerebellum. Moreover, the Cho and Cr levels are higher in the thalamus and cerebellum than in the white matter. The highest level of Cho in the brain is situated in the pons. In basal ganglia, the concentration of the Cho is greater with the inclusion of the thalamus compared with the voxels including the caudate, putamen, and globus pallidus.

ARTIFACTS

Metal objects such as orthodontic material, dental implants, or aneurysm clips, as well as CSF, air, bone, blood, calcium, and fat tissue should not be included within the acquisition area as they may produce artifacts in the spectral curves.

An MRS study of a patient with (figure 4.a) and without (figure 4.b) orthodontic material is observed in figure 4. The metallic material alters the MR signal making the MRS curves uninterpretable which may lead to an inaccurate diagnosis. 

Fig. 4: An MRS study of a patient with (a) and without (b) orthodontic material, where the alteration of the MR signal make the spectral curves uninterpretable which may lead to an inaccurate diagnosis.

FUNCTIONAL MAGNETIC RESONANCE IMAGING (fMRI)

fMRI is a dynamic imaging modality based on physiological changes secondary to neural electrical activity. This technique is used to evaluate the functionality of different cognitive process. According to the American College of Radiology (ACR), the clinical indications to perform an fMRI are the evaluation of patients with a brain tumor, epilepsy, and arteriovenous malformations. fMRI is used to assess eloquent areas and predict potential surgical deficits. However, nowadays the use of fMRI has spread beyond medical use as it allows a better understanding of brain functioning. 

EXPERIMENTAL DESIGN

The experimental design of each task must be carefully performed and this is probably the most critical aspect of fMRI studies. There are two main types of fMRI experimental design: block design and event-related design. Although the estimation of the shape and timing of the hemodynamic response is better for event-related design, the block design is usually preferred as it generally has superior detection power, the experimental analysis is extraordinarily simple, and the tasks are easier to explain to patients.

The most basic block design consists of alternating of activation and resting blocks (ArArAr). In the activation block (A), the subject is requested to perform a specific task according to the function of interest, while in the resting block (r) the patient is asked to remain at rest. The main advantage of this design is the minimum amount of instructions that the patient must remember and execute, while the main disadvantage is the presence of confounding factors in the resulting maps as all the brain regions involved in the execution of the task will be observed, being more critical in the mapping of complex cognitive functions. This effect may be avoided, using a block design that alternates activation and control blocks (ABABAB). In this design, the control condition is used as a baseline and consists of tasks that include all the neural processes involved in the activation phase, except for the process of interest. Thus, the brain activation map will show just the brain regions related to this neural process (figure 5). An inadequate experimental design may lead to false-negative results in cases where the brain region of interest is involved in both activation and control blocks, and false-positive results if the resulting map contains confounding factors. Both scenarios are critical when fMRI is used to establish resection limits during the surgical planning, and may be deletereous for the patient.

Fig. 5: Example of an fMRI study using a language paradigm with a block design configuration (ABABAB). During the activation block, both the visual cortex and the language areas are active, while in the control block just the visual cortex remain active. After the GLM processing, the resulting brain activation map shows the language areas without confounding factors.

CEREBRAL BLOOD FLOW

As the fMRI is a blood-oxygen-level-dependent technique, an alteration of the cerebral blood flow as observed in figure 6 may lead to a decrease or increase in the activation degree of a brain region, being susceptible of false-negative results. Clinical records and previous imaging studies should be considered in cases of abnormal activations.

Fig. 6: Functional map of two patients with occlusion of the internal carotid artery. A bilateral activation of the auditory cortex is observed in patient 1 (a), while patient 2 (b) shows an asymmetry in the activation associated with an acute ischemic lesion.
References: Bilecen D, Radü E, Schulte A, Hennig J, Scheffler K, Seifritz E. fMRI of the auditory cortex in patients with unilateral carotid artery steno-occlusive disease. Journal of Magnetic Resonance Imaging. 2002;15(6):621-627.

BRAIN ACTIVATION DEGREE

Some authors have investigated how the hemodynamic response (HDR) is altered due to pharmacological and psychoactive drugs, caffeine, cigarettes, and alcohol. Caffeine has shown an increase of the HDR up to 50 %, while the cigarettes and alcohol have been associated with a decrease of the HDR up to 20 %. It has been also reported that the changes in the visual cortex are higher than those changes observed in the motor cortex (figure 7). Prior to the MRI appointment, the patients should be requested to avoid all these substances, or at less to report them during the scanning session interview.

Fig. 7: Some authors have investigated how the hemodynamic response (HDR) is altered due to pharmacological and psychoactive drugs, caffeine, cigarettes, and alcohol. Caffeine has shown an increase of the HDR up to 50 % (left), while the cigarettes and alcohol have been associated with a decrease of the HDR up to 20 % (right).
References: Huettel SA, Song AW, McCarthy G. Functional Magnetic Resonance Imaging. Second edition. Sinauer Associates.

It is important to take into account that higher activations are also related to effort, difficulty, and even failure to perform the task instead of laterality condition. Coactivations may be useful to interpret when the patient uses some brain regions as a compensatory system to perform properly a specific task. 

The activation degree of the default mode network while performing tasks of low, medium, and high complexity, have been studied in the past years. Findings have shown a deactivation of this network related to the task complexity in healthy subjects, while this network in patients diagnosed with some neurological disorders has been reported to remain active under same tasks conditions. In a previous work, we demonstrated that the activation degree of this network is related to the age and academic degree of the subjects. Subjects with a high academic degree were observed to remain the degree activation of the default mode network for all conditions, while elderly subjects showed a decrease in the network activation compared to the young subjects with the same task condition.

Clinical condition, sociodemographic information, habits and lifestyle of patients should be considered during the interpretation of fMRI results in both clinical practice and research.

THRESHOLDING

During the image processing of the fMRI studies, the threshold value must be set carefully, especially in cases where a basic design is used and the activation condition is contrasted to a resting condition. If the threshold value is too low, noise would appear in the brain map, and it can lead to false-positive results as this noise may mimic brain activations. In the other hand, if this value is too high, brain activations would tend to disappear and false-negative results may be reported. When interpreting the fMRI results, the radiologist has to take into account the tasks used in the study, their experimental design, and the patient performance during the execution of these tasks.

An fMRI study where the subject was requested to perform a verbal episodic memory task involving visual, motor, language and memory functions is shown in figure 8 with their hemodynamic response curves. As the threshold value was set z = 1.8, all brain regions involved during the task are observed. However, if the threshold were set higher than 4, the brain regions associated with language and memory functions would disappear and the results could be misinterpreted.  

Fig. 8: In this fMRI study, the subject had to perform a verbal memory task where visual, motor, language and memory functions are involved. The major changes of the hemodynamic response relative to baseline occur in visual and motor areas, while in language and memory areas the percent signal changes are lower. If the threshold is set t = 1.8 (green dotted line), the brain areas related to memory functions are not going to be displayed and false negative results may be reported. In the other hand, if the threshold is set t = 4.2 (red dotted line), brain regions related to memory functions will be displayed; however, if the threshold is set too low, noise would be displayed in the images and false positive results may be reported. Experimental design is probably the most critical aspect of fMRI studies, and it has to be carefully done in order to obtain the brain activation of interest avoiding the activation of confounding factors.

DYNAMIC CONTRAST-ENHANCED PERFUSION MR

Perfusion MR is an imaging technique that provides information about the hemodynamics of the brain, and it is useful for the assessment of patients with suspicion or diagnosis of stroke, brain tumors, and neurodegenerative diseases. However, several limitations are associated with this technique.

SUSCEPTIBILITY ARTIFACTS

PWI is highly susceptible to metal objects such as orthodontic material, dental implants as well as CSF, blood, calcium, and the information may be significantly altered as shown in figure 9. When the artifact is related to the interface tissue-bone-air instead of metallic material, this may be avoided by reducing the slice thickness, or by establishing the axial plane parallel to a reference line crossing the anterior and posterior commissures during the image acquisition planning. 

Fig. 9: PWI is highly susceptible to metal objects such as orthodontic material, where the brain information may be significantly altered.

Even if the image is correctly acquired, this artifact may appear in the PWI data. In these cases, a field mapping sequence should be considered in order to perform an unwarping process to correct the artifact. However, it is important to take into account that this process is based on the extrapolation instead of the recovery of MR signal.

In addition, as this artifact affects all the echo planar images, such as fMRI and DWI, this process can also be applied to improve the results of the alignment and registration steps during the image processing of these studies. 

BLOOD-BRAIN BARRIER

The assessment using the relative cerebral blood volume (rCBV) should be carefully performed in patients with glioblastomas multiforme or meningiomas as the rCBV these values may be altered due to the breakdown of the blood-brain barrier. It has to be considered that the rCBV measurement is not an absolute quantification of the cerebral blood volume.

VOXEL-BASED MORPHOMETRY (VBM)

Quantitative studies allow performing an objective analysis of cortical and subcortical brain atrophy. Volumetry of hippocampus is one of the main measurements used both in epilepsy and other psychiatric and neurological disorders. In epilepsy, this technique allows to identify the presence of hippocampal atrophy which may be related to mesial temporal sclerosis; it has been described as a good predictor and diagnostic tool. Compared to invasive monitoring, the quantitative analysis has shown high predictive values, mainly where the hippocampal atrophy is mild or bilateral. Another example is the measurement of the cortical thickness; this variable thickness plays an important role in differentiating dementing processes where clinical symptoms are similar; for example in normal pressure hydrocephalus and Alzheimer’s disease.

The main pitfalls of the VBM technique are related to the errors of the automatic processing software in studies with head motion, or patients with several morphological alterations such as the ventricular system dilatation in normal pressure hydrocephalus disease. In these cases, the output volumes should be checked after each processing step to determine whether or not they have to be edited prior to performing the next image processing steps.

On the other hand, the brain volume assessment should be performed using both the volumetric and asymmetry indexes. The volumetric index (VI) is defined as the ratio between the volume of the brain region of interest and the total intracranial volume, and the asymmetry index (AI) is defined as the difference between the left and the right volumetric indexes. The AI is useful to identify unilateral alterations, while the VI is useful to identify bilateral alterations. 

DIFFUSION TENSOR IMAGING (DTI)

DTI is based on the motion and diffusivity of water molecules. If water molecules move freely into any direction molecules are said to be isotropic; an example of isotropic movement is the ventricular system. In the brain parenchyma, neurons and other cells restrict the movement of extracellular water making molecules anisotropic. 

Factors contributing to changes in diffusivity of water are cell membrane integrity, myelination, edema, gliosis, and inflammation; this explains the potential usefulness of DTI in the evaluation of different pathologic process including brain tumors, congenital diseases, demyelinating diseases, or diffuse axonal injury. However, interpretation of the findings should be made with precaution. 

Pitfalls in the interpretation of DTI studies include:

QUANTITATIVE INTERPRETATION

Fractional anisotropy (FA) is a quantitative measurement for the diffusivity of water molecules. It is an index ranging from 0 (isotropic) to 1 (maximally anisotropic). When FA values are below 0.2, water molecules are assumed to be moving freely without any restriction. However, there is no single quantitative measurement to characterize each tract individually, as measurements can change along the same tract, or between patients of different ages and even between contralateral tracts.

a. White-matter pathways integrity.

During the evaluation of the integrity of the white-matter tracts, the FA values may play an important role. The analysis should be performed around the lesion with three possible situations. First, when the FA value adjacent to the lesion is increased compared to the contralateral tract, displacement and compression of the fibers may be assumed (figure 10.b). This situation has a better prognosis for the patient. Second, if the FA value is similar o decreased compared to the contralateral tract, infiltration of the fibers within the lesion may be concluded if there is not edema (Figure 10.a). This would be deleterious for the patient as postsurgical deficits may occur. Finally, if the FA value is decreased but there is peritumoral edema, infiltration can not be concluded as the decrease in the FA values could only be a transitional effect (figure 11). In these cases, follow-up studies are necessary, and postsurgical deficits can not be predicted.

Fig. 10: When the FA value is similar o decreased compared to the contralateral tract, infiltration of the fibers within the lesion may be concluded if there is not edema (a). If the FA value adjacent to the lesion is increased compared to the contralateral tract, displacement and compression of the fibers may be assumed (b). This situation has a better prognosis for the patient.

Fig. 11: If the FA value is decreased but there is peritumoral edema, infiltration can not be concluded as the decrease in the FA values could only be a transitional effec. In these cases, follow-up studies are necessary, and postsurgical deficits can not be predicted.

QUALITATIVE INTERPRETATION

Qualitative evaluation of white-matter tracts should be made with caution; well knowledge of the anatomy and trajectory of each tract is needed. DTI cannot correspond directly to an actual axonal fiber tract but only an estimate of it. 

a. Fiber direction.

DTI data is processed in each voxel and an orientation of movement is assumed depending on the direction with the highest diffusivity; this allows to delimitate white matter tracts in the central nervous system. However, DTI cannot differentiate between afferent and efferent fibers as it is a water-diffusion-based measurement instead of an electric measurement of the neuronal cells (figure 12). 

Fig. 12: In the figure, despite the spinothalamic (orange arrow) is an afferent pathway and the corticospinal tract is an efferent pathway (red arrow) there is not a difference between them. DTI cannot differentiate between afferent and efferent fibers as it is a water-diffusion-based measurement instead of an electric measurement of the neuronal cells.

b. Crossing pathways

Crossing pathways with axons oriented in multiple directions are problematic as reconstruction algorithms do not know how to interpret this information. In these regions, diffusion of water is very similar in all directions, without any predominant gradient direction. This is especially challenging in the brainstem where multiple white-matter tracts cross and arise from a small area of interest (figure 13).

Fig. 13: Crossing fiber is one of the big challenges of the DTI technique, especially in the brainstem where multiple white-matter tracts cross and arise from a small area of interest (orange arrow).