Domain Frekuensi ​Selain relaksasi, tiga aspek lain dari sinyal MR menarik: ​amplitudo,

frekuensi, dan fase. Amplitudo sebanding dengan M​0 tepat sebelum pulsa RF.sinyal ​Frekuensi
sebanding dengan medan magnet aktual yang memengaruhi proton. Karena ada perbedaan
medan magnet di seluruh jaringan, ada banyak sinyal MR pada banyak frekuensi setelah
pulsa RF. These signals are superimposed on each other, which result in a time-varied FID
that consists of many frequencies. It is much easier to examine a multifrequency signal in the
frequency domain than in the time domain.

The conversion of the digitized time domain FID to a frequency domain presentation is
accomplished by using a mathematical operation called the discrete ​Fourier transformation.
(Fig. 1.15-5) shows the relationship between a time domain FID and the corresponding
frequency domain data obtained using the ​Fourier transformation. In the frequency domain,
the NMR signal is mapped according to its frequency relative to the carrier frequency.

The specific frequency that a proton emits depends on magnetic fields that affect it. The bulk
of the magnetic field is the external magnetic field B​0 generated by the magnet. This field can
be changed slightly by the sample as well as by theuseof external field gradients, a concept
that is examined more fully later in this chapter. For each type of magnetic nucleus, the static
magnetic field it experiences is slightly modified by the circulating currents of electrons in
close proximity to it. As electrons move, they create their own small, regional, magnetic
fields that, to a slight degree, alter the final magnetic field experienced by the nucleus. This
can shift the magnetic field value B​0 by a few or tens of parts per million (ppm). Thus, the
Larmor frequency of the nucleus is changed. This change is termed ​chemical shift and is a
phenomenon of molecular origin associated with the chemical environment of the nucleus.
Because the spatial configuration of electrons is unique for a given molecule, precessing
nuclei take on altered frequencies that correlate with their chemical environment. Because
different chemical environment shield the external magnetic field, B​0 acts on an element in a
molecule, or oa different molecules, in different ways. Thus, in MRS, ​chemical shift enable
the different locations of elements within the same molecule to be probed and different
molecules to be distinguished.

In most other types of diagnostic imaging, signal localization is achieved by geometric

means. In NMR, the precession frequency of the magnetic nuclei depends on the strength of
the external magnetic field. This field dependence can also be used to assign frequencies to
different regions of space and therefore to create an image. Unlike the chemical shift in which
variations in the magnetic field are intrinsically caused by electrons circulating around the
nuclei, in the case of imaging (and localized spectroscopy and spectroscopic imaging [S1]) as
well, the magnetic field is made spatially dependent by superimposing the so-called ​magnetic
field gradients on the static magnetic field B​0​. ​Magnetic field gradient can be regarded as
small disturbances or perturbations in the static magnetic field B​0​. These perturbations are
linearly dependen on their position inside the magnet. A typical field gradient produces a
maximum field distortion of less than 1 percent, which is many orders if magnitude greater
than the chemical shift differences between protons. They are active for short periods of time
and are reffered to as ​gradient pulses (as opposed to the RF pulses). Ordinarily, three
gradients are used to produce the orthogonal field distribution required for imaging; these
gradients correspond to the x-axis, the y-axis, and the z-axis, respectively. They are each
generated by the flow of electrical current through separate loops of wire mounted into a
single hardware component known as the ​gradient coil. Variations in gradient amplitude are
produced by changes in the amount or direction of the electrical current flowing through the
gradient coil. In the presence of a magnetic field, protons at the same point along the
gradient, corresponding to a plane perpendicular to the direction of the gradient, share the
same resonant frequency, whereas those in the neighboring planes precess at different
frequencies. Gradients are easily produced in multiple directions. Quite simply, the MR
image is a map of nuclear frequencies emitted through unique magnetic fields that correspond
to each pixel (ie, point element) throughout the image. The pixel intensity is proportional to
the number of nuclei contained within each voxel (ie, volume element) weighted by the T1
and T2 or T2* relaxation times for the tissue within the voxel. The field gradients are also
used in localized spectroscopy and S1 to differentiate spectroscopic signals from various
locations in the tissue.

FIGURE 1.15-5 Time domain free induction decay (FID) versus frequency domain
presentation: The Fourier transformation converts a time domain signal (FID) into the
corresponding frequency domain signal (spectrum). (Courtesy of Patric O'Donnell).

STRUCTURAL MRI

MRI Analysis Earlier imaging studies in psycniatric disorders used qualitative and
subjective methods (such as visual inspection) to detect differences in the size of particular
brain structures between patients and controls. The quantitave methods gave rise to more
objective quantitative methods in which investigators determined the size of structures by
outlining the gray matter, white matter, or cerebrospinal fluid (CSF) borders. Recently,
voxel-based and deformation-based morphometric techniques provide automated measures of
shape and volume and are less labor-intensive compared to manual methods. Many of the
automated or semiautomated methods of analysis include steps to standardize the spatial
orientation of the brain, which reduces variability due to differences in head positioning
during scan acquisition. MRI scans are usually aligned vertically to the interhemispheric
plane and horizontally to the line connecting the anterior and posterior commissures, or, in
the case of a longitudinal structure, such as the hippocampus, scans are resliced in the coronal
plane perpendicular to the longest axis of the structure. The anatomical boundaries of the
region are defined using specific criteria and are manually traced in each slice (Fig. 1.15-6).
The areas of the manually segmented region are then calculated by a computer algorithm and
multiplied by the tnickness of each slice to give the volume. The volumes of the individual
slices are then summed to give a total volume of the ​region of interest (ROI). Although
results based on manually based ROI performed by trained raters are accurate, this method is
time consuming, labor intensive, and amenable to rater bias and depends on varying a priori
anatomical criteria amongst groups.

Although the majority of published MRI studies continue to rely on manually traced ROI
methods, semiautomated and automated morphometric techniques and analysis have become
increasingly popular. These newer techniques make it possible to quantify the volume and to
identify the shape of structures, such as hippocampi, and to provide information on sulcal
patterns and the thickness of the cortex. These semiautomated techniques also help eliminate
rater bias, and they allow for the identification of structural abnormalities that cannot be
picked up with visual inspection.

One of the earliest semiautomated methods is ​voxel-based morphometry​. This technique is

often used for analysis of functional imaging data and is based on statistical parametric
mapping (SPM) methods. It provides voxel-by-voxel comparisons of gray matter
conceatration between groups. MRI scans are preprocessed to reduce interindividual
variations in brain size and shape. Structural data are spatially transformed into a standard
stereotacticspace. The standard spatial normalization process involves superimposing an
index MRI on an atlas, such that the MRI data are transformed to the space occupied by the
atlas. The atlas used for this purposed is called the ​Talairach template and is based on
postmortem section of the brain of a 60 year-old female subject. This transformation to a
Talairach template vertically aligns the interhemispheric plane and horizontally aligns the
line connecting the anterior and posterior commissures, creating a customized template. To
overcome the limitations of the Talairach template, the International Consortium for Brain
Mapping (ICBM) constructed a composite MRI data set from young normal subjects whose
scans were indivually mapped into the Talairach system. The normalized sMRI scan
therefore conforms to a standard anatomical space but also represents local MRI scanner
features. After this intermediate step of spatial normalization, scans are then segmented into
gray matter, white matter, and CSF partitions by using SPM. Differences among image
intensities in the MRI are normalized by using nonuniformity correction. The normalized and
segmented images are then smoothed to conform to the Gaussian model used for statistical
analysis to detect regional differences. Statistical maps created in the standardized brain space
confirm the exact anatomical location of the voxel clusters having significantly different gray
matter volume.

A fixed brain atlas cannot accurately reflect the complex and subtle structural differences
among subjects. Variations in cortical patterns between normal subjects and subjects with
disorders cannot easily be identified using a standard anatomical template. To overcome this
limitation, ​deformable atlases have been developed. These atlases are based on the laws of
continuum mechanics and are driven by viscous fluid-based warping transformations.
Functionally important surfaces in one brain are elastically transformed to correspond
structurally with the target brain to precisely match its individual anatomy. Deformations
applied to the atlas then can be used to calculate the mean anatomical shape and volume of
the target structure. The algorithm calculates the high-dimensional volumetric warp as a
result of deforming one three-dimensional scan into structural similarity with the target scan.
Abnormality maps specific for a particular illness can be calibrated using deviations from
standard probability distributions. This deformable template or atlas has also been used to
estimate the morphometric variability and shape differences in groups of patients versus the
template and has given rise to term ​deformation-based ​morphometry. Deformation-based
morphometry ​has been applied to evaluations of the cortical, as well as subcortical, changes
in pattern and shape and has been used to determine the rate of structural brain changes in
clinical populations. Several reports have indicated that this technique may detect changes in
the shape of a specific structure in the absence of volumetric differences. Although manually
based ROI analysis, voxel-based morphometry, and deformation-based morphometry have
been compared head to head in a single study, a preliminary report suggested that
anatomically based manual tracing is superior to semiautomated or automated methods.

Clinical Indications and Research Caveats of sMRI in Psychiatry ​MRI has been used
clinically to evaluate structural brain abnormalities to rule out organic causes of psychiatric
illness. Other clinical indications include the evaluation of an abrupt change in mental status,
new-onset dementia or psychosis, and new-onset memory loss. Research studies that have
used sMRI to evaluate brain abnormalities have reported group differences between patients
and controls. However, sMRI abnormalities are not diagnostic of any particular psychiatric
illness. With the exception of dementia and, perhaps, schizophrenia, structural abnormalities
in other psychiatric disorders are mixed. Factors contributing to inconsistencies in sMRI
findings include differences in image acquisition and image analysis, as well as differing
anatomical criteria used for defining the regions of interest. Result are further confounded by
clinical characteristics, such as duration and severity of illness, comorbid disorders, and
medications status. This heterogeneity and varians in sMRI findings could be reduced by
collaboration and communication among researchers studying well-define groups of subjects.
Finally, even when an abnormality has been reported consistently, it is not clear whether the
structural change is a consequence, a correlate, or a risk factor for developing the illness.
Longitudinal studies in high-risk populations, family studies, and studies combining
neuroimaging with genetics could help tease out “the chicken and the egg” dilemma facing
sMRI studies in psychiatric illness.

sMRI Findings in Psychiatric Disorders Dementias ​Together with a clinical history and
a neurological examination, sMRI could serve as a diagnostic tool for Alzheimer's disease
and other dementias. The majority of sMRI studies of Alzheimer's disease found atrophy of
the whole brain along with enlargement of the ventricular and sulcal CSF spaces. Alzheimer's
disease is characterized by amyloid plaques and neurofibrillary tangles that could result in the
loss of neurons in the cortex as well as the subcortical limbic system. In addition to global
atrophy and reduction in gray matter, focal atrophy has been found in the frontal, temporal,
and parietal lobes. Although most studies in Alzheimer's disease did not confirm changes in
white matter, it is possible that subjects with early-onset Alzheimer's disease have a decrease
in white matter in addition to reductions in gray matter. Patients with Alzheimer's disease
have been distinguished from healthy subjects with 92 percent accuracy based on large
temporal and parietal ventricular CSF space along with small temporal gray matter.
Increasing demential severity was shown to be related to decreased gray matter volume.

Of the various medial temporal lobe structures evaluated in Alzheimer's disease, the
hippocampus has been the main focus, given its crucial role in mediating declarative
memory. Several groups have consistently found a reduction in hippocampal volume in
subjects with Alzheimer's disease. This volumetric reduction was most prominent in the head
of the hippocampus. The range of volume reduction of the hippocampus in Alzheimer's
disease was approximately 19 to 40 percent. Hippocampal volume has been correlated with
neuronal counts in patients with Alzheimer's disease and directly correlated with poor
performance on delayed-recall memory tasks. Although several longitudinal studies have
shown that the rate of volume loss for the hippocampus is greatest in Alzheimer's disease
when compared to healthy subjects and subjects with mild cognitive impairment, a single
report suggested that accelerated temporal lobe atrophy may be more prominent in
Alzheimer's disease than atrophy of the hippocampus. Longitudinal volumetric changes in the
hippocampus have also been used to evaluate patients with mild cognitive impairment who
are at risk for developing dementia. According to various published studies, subjects with
mild cognitive impairment who are significant hippocampal atrophy are at a higher risk for
developing Alzheimer's disease. Smaller hippocampal volume in Alzheimer's disease was
also associated with apo E ​ɛ​4 allele. Patients who carried two copies of the apo E ​ɛ​4 allele had
greater hippocampal volume loss than subjects who carried one or no E ​ɛ​4 allele. However,
not all studies have been able to replicate this genetic association in Alzheimer's disease.

Other medical temporal lobe structures have also been evaluated in Alzheimer's disease. A 15
to 20 percent reduction in the volume of the parahippocampal gyrus, a 30 to 40 percent
smaller amygdale volume, a decreased entorhinal cortical volume ( 38 to 61 percent), and a
smaller subiculum have also been reported in patients with Alzheimer's disease. Although
entorhinal cortical measurements could differentiate mild Alzheimer's disease from normal
elderly and could potentially be a tool for early detection of Alzheimer's disease, the
anatomical definitions make manual tracing of this structure difficult. Reduction in the
volume of the corpus callosum as a result of white matter degeneration is a consistent finding
in Alzheimer's disease; however, the exact focal location of volumetric reduction in the
corpus callosum has been debated. Longitudinal studies in patients with Alzheimer's disease
have demonstrated a progressive loss of whole brain volume at 1.0 to 2.8 percent per year for
whole brain volume and an approximately 4 percent annual loss in volume for the
hippocampus, with an increase in temporal horn volume at approximately 14 percent.
Deformation-based morphometry has also been used to longitudinally evaluate changes in
cortical and subcortical maps in Alzheimer's disease and minimal cognitive deficits.

sMRI has also been used in the diagnosis of other kinds of dementias, such as frontotemporal
dementia and Huntington's disease. Selective atrophy of the frontal and temporal lobes is
characteristic of frontotemporal dementia, whereas atrophy of the putamen and caudate
nucleus is more characteristic of dementia related to Huntington's disease.

Psychotic Disorders: Schizophrenia ​Johnstone and cowokers' initial report of enlarged

ventricles, as seen with CT the mind 1970s, has subquently been replicated numerous times
using CT and sMRI. The expanding fluid-filled spaces in the schizophrenic brain are noe
considered characteristic of the disease observed; they are observed not only in the ventricles,
but also in the sulci (gaps between adjacent gyri in the neocortex). However, increases in
ventricular size are often relatively modest (10 to 20 percent), and most patient values overlap
those of healthy subjects. Thus, increased ventricular size is not diagnostic of the disorder.
Furthermore, increased ventricular volume is not specific to schizophrenia, because it is
found in many other disorders of diverse developmental or degenerative etiologies (eg,.
Alzheimer's disease, hydrocephalus, and bipolar disorder).

The increased fluid spaces in the presence of a normal-sized cranial vault means that there
must be less brain tissue in this disorder. Is there less brain tissue because of atrophy or were
there fewer brain cells at birth or during the brain's development? Researchers continue to
debate this question along lines that can be described as the neurodegenerative versus the
neurodevelopmental hypothesis. The ​neurodevelopmental hypothesis is supported by the
higher incidence of gross structural abnormalities in the disorder, including cavum septum
pellucidum (a fluid-filled cyst forming in the septum located in the midline between the
hemispheres) and callosal agenesis (an underdevelopment of the corpus callosum, the wide
tract of axons that connect the two hemispheres). Evidence for a ​neurodegenerative process
is controversial and difficult to confirm, because it involves the repeated scanning of subjects
over time. Nevertheless, some studies have been able to confirm the progression of
ventricular enlargement, and loss of frontal lobe volume is greater in patients with
schizophrenia than in healthy subjects. However, these results are not consistent across
patient populations, occurring most often in chronic schizophrenia and in patients with
prominent negative symptoms.

Serial MRI studies of childhood-onset schizophrenia performed at the National Institute of

Mental Health (NIMH) provide some insight into the neurodevelopmental and
neurodegenerative hypotheses. Childhood-onset schizophrenia is a severe illness with onset
of psychosis occurring before 12 years of age. Even at the time of the initial psychotic break,
patients with childhood-onset schizophrenia have smaller brain volume than age-matched
controls, exhibiting an overall 10 percent decrease in total gray matter volume. Furthermore,
childhood-onset schizophrenia patients show a progressive loss in frontal and temporal gray
matter volumes during late adolescence that is appreciably greater than the gray (and white)
matter loss typical of normal brain development in late adolescence. Thus, childhood-onset
schizophrenia patients exhibit initial development abnormality at the onset of early
symptoms, as well as a progressive component. Should this latter progression, however, be
termed ​neurogeneration or abnormal development? In fact, selection between these two
terms may be rather arbitrary and off the point. Instead, research effort should be and are
being focused on the mechanisms for these losses in the hope that therapeutics can reverse
them. Finally, the structural neurological changes seen in childhood-onset schizophrenia may
not reflect changes seen in the adult-onset disorder. It is quite possible that childhood-onset
schizophrenia is a severe and detiologically distinct disorder.

Because the studies of enlarged ventricles have been well replicated, and because these
studies imply lower brain tissue volumes, can one assume that this deficit is localized to one
region, or is it a generalized loss in all regions? A small, global (approximately 5 percent)
volume loss has been reported in some, but not all, sMRI studies. Disproportionately large
volume losses (10 to 15 percent) are commonly seen in medial temporal lobe structures
(including amygdale , hippocampus, and parahippocampal gyrus) and the superior temporal
gyrus. However, a smaller percentage of studies also report tissue deficits in frontal and
parietal cortices, as well as the corpus callosum. Most of these localized volume deficits are
found not only in chronic schizophrenics, but also in patients exhibiting first-break psychosis,
suggesting that peripheral changes could predispose subjects to psychosis. Furthermore,
because ventricular enlargement and regional brain volume losses are commonly found in
studies of first-break patients who were previously pharmacologically naïve, these
abnormalities are most likely not caused by neuroleptic medications. However, typical (but
not atypical) antipsychotic medications have been frequently reported to increase the size of
human basal ganglia (caudate, putamen, and globus pallidus).

A so-called two-hit model is sometimes used to explain the typical age of onset of
schizophrenia in late adolescence or early adulthood. Genetic or early neurodevelopmental
abnormalities are the so-called first hit that predisposes subjects to the disorder. Later, a
variety of potential stressors in adolescence, including hormonal influences on the brain and
the social stress of this age group, may act as the second hit to precipitate a psychotic
episode.

Do regional brain volume losses correlate with specific symptoms of schizophrenia?

Although not consistently replicated, negative or deficit symptoms may be associated with
enlarged lateral ventricles and decreased volume of medial temporal lobe structures. In a
similar, but not uniform, manner, positive symptoms may correclate with decreased volume
of the superior temporal gyrus.

Mood Disordes ​Affective disorders are a heterogeneous group of illnesses with no apparent
single etiology. Various functional abnormalities in the limbic-thalamic-cortical and the
limbic-cortical-striatal-pallidal-thalamic circuits or their components have been reported in
major depressive disorder and bipolar disorder. sMRI studies of mood disorders have
investigated individual components of these circuits. Unlike schizophrenia, several controlled
MRI studies have confirmed normal total brain volumes in bipolar and unipolar disorders.
The findings of earlier CAT scan findings, as well as more recent MRI studies evaluating
third and lateral ventricular volumes in major depressive disorder and bipolar patients, are
mixed. In contrast, the majority of MRI studies in late-onset depressed elderly subjects have
consistently confirmed larger third and lateral ventricles and cortical atrophy. It is possible
that contradictory sMRI findings in mood disorders could be due to varying sociodemograpic
and illness characteristics of the study samples. In fact, lateral ventricles were significantly
larger in patients with bipolar disorders with recurrent episodes as compared to first episode
patients or healthy subjects (see the 2002 article by Strakowski et al.), suggesting that brain
morphology can change over time.

Because the prefrontal cortex plays a critical role in emotion and cognition, several groups
have investigated volumetric changes in affective disorders using sMRI. Studies cinfirmed a
volumetric reduction in the frontal lobes in unipolar depression. Subsequent sMRI studies
localized atrophy in the frontal lobe to the prefrontal cortical region. When subregions of the
prefrontal cortex were examined, focal reductions in volume were observed in the left
subgenual prefrontal cortex (an area of the anterior cingulated cortex [ACC] ventral to the
genu of the corpus callosum) in familial unipolar and bipolar disorder. Postmortem studies
confirmed a reduction in glia in the subgenual prefrontal cortex. This defect may alter
synaptic transmission and may disrupt reciprocal connections with other regions of the brain,
including the limbic system, thereby resulting in affective, neuroendocrine, and autonomic
dysfunction characteristic of depression. However, more recent studies evaluating the volume
of the subgenual prefrontal cortex have been unable to replicate the volume reduction seen in
earlier studies. Postmortem studies also confirmed a reduction of neuropil in the posterior
orbital cortex in unipolar and bipolar disorders. The volume of orbital frontal cortex,
specifically the gyrus rectus, was reduced in elderly and nonelderly unipolar depressed
subjects compared to controls. Because the orbital cortex is reciprocally connected to the
amygdale and modulates behavioral and cognitive responses related to fear and reward,
structural abnormalities in this region could underline behavioral and autonomic symptoms
related to depression.

Based on the preclinical literature supporting hippocampal neuronal damage secondary to

elevated levels of steroids, several groups evaluated hippocampal volume in unipolar
depression. Reduction in the left, right, and bilateral hippocampi in the range of 6 to 20
percent were confirmed in several, but not all, studies on unipolar depression (Fig. 1.15-7A).
Using deformation-based morphometric techniques, an abnormality of the shape of the
subiculum was identified in the absence of smaller hippocampal volumes in depressed
patients compared to controls. Hippocampal volume reduction correlated with length of the
affective illness and total duration of depression in elderly women, suggesting that
cumulative hippocampal damage occurred with repeated episodes. Depressed subjects with
treatment-resistant chronic depression had reduced the gray matter density in the left
temporal cortex using voxel-based morphometry. However, the inconsistencies in the
hippocampal volume data require more explanation. Close examination of MRI studies of
patients with major depressive disorders suggest that smaller hippocampal volume is
restricted to subjects with treatment-resistent depression, older women with major depression,
women with treatment-resistant depression, depressed woman with a history of child-hood
sexual abuse, and elderly depressed patients. This conclusion has been challenged by the
recent finding that male patients with a first episode of major depression have significantly
smaller left hippocampal volumes as compared to healthy male patients. In contrast to the
mixed findings of hippocampal volume in unipolar disorders, hippocampal volumes in
bipolar patients are similar to normals; moreover, two groups reported an increase in
hippocampal volume.

Future longitudinal studies in medication-naïve subjects may shed light on the question of
hippocampal volume and its role as a predisposing factor or as a consequence of repeated
episodes of major depression. Earlier studies evaluating hippocampal and amygdalar
measures grouped these structures together, because it was not possible to define the exact
anatomical separation. A recent study of unipolar major depression demonstrated a volume
reduction in core nuclei in the amygdale in the absence of a reduction in the total volume of
the amygdale. In bipolar disorder, two studies demonstrated amygdale enlargement, whereas
others studies confirmed a decrease or no change. The anterior portion of the superior
temporal gyrus was enlarged in one study of bipolar disorder; however, volumetric
differences were not seen in other subregions, including the planum temporal or the Heschl's
gyrus. Of note, hippocampal and amygdalar abnormalities in depressed patients occurred in
the absence of volumetric changes in the temporal lobes, suggesting that anatomical
abnormalities may be restricted to the mesial temporal lobe.

Decreased putaminal and caudate volumes have been reported in unipolar depression. In
contrast, increased caudate and putaminal volumes have been reported by some, but not all,
studies of bipolar disorder. With regard to affective disorders, thalamic volume data also
mixed, with some studies reporting an increase, whereas others reported a decrease or no
change. Volumetric reductions have been reported in the cerebellum and the vermis in
affective disorders, particularly in familial bipolar disorder. One study showed unipolar
depressed subjects had larger pituitary volumes, although this finding was not reflected in
subsequent studies. On the other hand, bipolar disorder was associate 'd with decreased
pituitary volume. The finding of subcortical white matter hyperintensities on T2-weighted
sMRI in late-life depression has received increasing attention. The term ​vascular depression
has beencoined to distinguish subjects with late-onset depression from early-onset mood
changes.

Despite the various structural change observed in mood disorders, it is not clear whether a
focal brain abnormality is central to the