**3. Structural imaging**

By far, the most established use of MR is to examine the gross anatomy of the brain. With the right specifications, MR can provide a highly detailed three-dimensional image that al‐ lows for the examination of brain structures. Weighting is used to provide contrast for the tissue of interest.

of the ε4 allele were found to have a thinning of the cortex in the entorhinal region, subicu‐ lum, and other MTL structures, although the results were stronger in those with a family

Using Magnetic Resonance Imaging in the Early Detection of Alzheimer's Disease

http://dx.doi.org/10.5772/54445

231

The detectible changes are not limited to atrophy. There have been several studies that have discovered an increase in gray matter in young adult carriers of the ε4 ellele. Increases were found in bilateral cerebellar, occipital, and thalamic regions as well as in the fusiform and right lingual gyri [22,25]. Recent work has also suggested that changes in the basal choliner‐ gic forebrain may be detectible decades before cognitive impairment, although this study

One of the significant weaknesses of analyzing structural changes is that the regions of inter‐ est can vary in size even across healthy individuals. Longitudinal studies are the only way to control for this variability. Secondly, the atrophy of brain regions likely occurs secondary to functional changes. The assessment of atrophy alone gives little information as to the un‐

Unlike T1 weighted imaging, T2 imaging relies on the dephasing of the magnetization vec‐ tor in the transverse plane. T2 weighting, specifically FLuid Attenuated Inversion Recovery (FLAIR) imaging, is used to identify White Matter Hyperintensities (WMH), which are in‐ creased in AD [27]. In contrast, diffusion tensor imaging (DTI) is able to indirectly measure the integrity of myelin sheaths surrounding white matter tracts, and Susceptibility weighted imaging (SWI) is able to distinguish tissues at a high resolution based on several properties.

If simply T2 weighted imaging was used, the signal from Cerebrospinal Fluid (CSF) is strong and therefore very bright (T2 of CSF ~ 600 ms at 3T). This makes it difficult to see subtle abnormalities in the white matter regions that partial volume with CSF. FLAIR imag‐ ing nulls the signal from CSF so that the image is focused solely on the white matter. The first RF pulse inverts the magnetization by 180 degrees. Then, when the longitudinal mag‐ netization for the CSF = 0, an excitation pulse and readout is applied. Because T1 of CSF (~4000 ms at 3T) is much longer than that of tissue (T1~700-1200 ms at 3T), residual tissue

DTI measures fractional anisotropy (FA), a quantitative measure of the coordinated move‐ ment of water molecules. FA assumes that the stronger a white matter tract is, the more like‐ ly the water molecules will be to move along the tract rather than sideways within the myelin sheath. If the myelin sheath is damaged it becomes easier for water molecules to dif‐

The loss of white matter integrity, either through WMH or FA differences, may correlate with increasing cognitive impairment [28,29]. In AD populations reduced FA values have been found in frontal and temporal lobes, the posterior cingulum, the corpus callosum, the superior longitudinal fasciculus and the uncinate fasciculus [30]. Both WMH and FA have been found to distinguish normal aging from aMCI[31] and predict conversion from aMCI

*FLuid Attenuated Inversion Recovery (FLAIR) and Diffusion Tensor Imaging (DTI)*

history of AD than those that carried the ε4 allele alone [23,24].

did not take into account genetic status [26].

derlying factors that led to neuronal loss.

signal remains at the time of the CSF nulling.

fuse through it, and the FA value will decrease.

**3.2. White matter imaging**

#### **3.1. Anatomical imaging**

T1 weighted imaging is used to visualize structural changes in tissue. At a field strength of 3 Tesla, T1 weighted images can be acquired in about five minutes and have a resolution of approximately 1 mm3 .

The most significant differences reported in patients are atrophy of the structures in the medial temporal lobe (MTL) which typically follow the "Braak stages" of AD progres‐ sion [11]. Briefly, pathology starts in the transentorhinal region (stages I and II), moves to the limbic region (stages III and IV) and ends in isocortical regions (stages V and VI). Studies that have been done in AD patients show that hippocampal and entorhinal cor‐ tex volume change, as well as temporal lobe morphology changes are the best measures to predict change over time [12]. A higher level of regional brain atrophy has also been associated with decreased levels of Aβ-42 and increased levels of phosphorylated tau in the CSF of AD patients [12].

In patients who have been diagnosed with aMCI, changes to the parahippocampal region are already apparent. It is up for debate whether the investigation of the entire brain or just volumes of interest (VOIs) are better at predicting conversion from aMCI to AD, but in a re‐ cent meta-analysis of work using data from the Alzheimer's Disease Neuroimaging Initia‐ tive (ADNI) only four methods were able to distinguish those who would convert more accurately than random chance. None of the four were more statistically reliable than the others, but three examined VOIs (Voxel-STAND, 57% sensitivity and 78% specificity; Voxel-COMPARE, 62% sensitivity and 67% specificity; Hippo-Volume, 62% sensitivity and 69% specificity) while only one examined the entire brain (Thickness-Direct, 32% sensitivity and 91% specificity) [11,13-15]. A protocol devised by Chincarini et al. to sample several VOIs has demonstrated a method of separating converters from non-converters with a sensitivity of 71% and a specificity of 65% [16,17]. Another method for predicting conversion is examin‐ ing hippocampal shape, and Costafreda et. al. were able to develop a method with 77% sen‐ sitivity and 80% specificity. [18,19].

Patients that are at-risk for AD but have no cognitive deficit are much more difficult to iden‐ tify. Most studies have been done in carriers of the ApoE ε4 allele, however it is important to remember that these studies have been cross sectional, and therefore may reflect a conse‐ quence of the gene that makes carriers more susceptible to AD, but not necessarily a stage of AD itself. There have been cortical thinning signatures identified in children, adolescent, and young adult carriers of the ε4 allele. These signatures reflect reductions in dorsolateral and medial prefrontal, lateral, temporal, and parietal cortices. [20-22]. Middle-aged carriers of the ε4 allele were found to have a thinning of the cortex in the entorhinal region, subicu‐ lum, and other MTL structures, although the results were stronger in those with a family history of AD than those that carried the ε4 allele alone [23,24].

The detectible changes are not limited to atrophy. There have been several studies that have discovered an increase in gray matter in young adult carriers of the ε4 ellele. Increases were found in bilateral cerebellar, occipital, and thalamic regions as well as in the fusiform and right lingual gyri [22,25]. Recent work has also suggested that changes in the basal choliner‐ gic forebrain may be detectible decades before cognitive impairment, although this study did not take into account genetic status [26].

One of the significant weaknesses of analyzing structural changes is that the regions of inter‐ est can vary in size even across healthy individuals. Longitudinal studies are the only way to control for this variability. Secondly, the atrophy of brain regions likely occurs secondary to functional changes. The assessment of atrophy alone gives little information as to the un‐ derlying factors that led to neuronal loss.

#### **3.2. White matter imaging**

**3. Structural imaging**

230 Understanding Alzheimer's Disease

**3.1. Anatomical imaging**

approximately 1 mm3

the CSF of AD patients [12].

sitivity and 80% specificity. [18,19].

.

tissue of interest.

By far, the most established use of MR is to examine the gross anatomy of the brain. With the right specifications, MR can provide a highly detailed three-dimensional image that al‐ lows for the examination of brain structures. Weighting is used to provide contrast for the

T1 weighted imaging is used to visualize structural changes in tissue. At a field strength of 3 Tesla, T1 weighted images can be acquired in about five minutes and have a resolution of

The most significant differences reported in patients are atrophy of the structures in the medial temporal lobe (MTL) which typically follow the "Braak stages" of AD progres‐ sion [11]. Briefly, pathology starts in the transentorhinal region (stages I and II), moves to the limbic region (stages III and IV) and ends in isocortical regions (stages V and VI). Studies that have been done in AD patients show that hippocampal and entorhinal cor‐ tex volume change, as well as temporal lobe morphology changes are the best measures to predict change over time [12]. A higher level of regional brain atrophy has also been associated with decreased levels of Aβ-42 and increased levels of phosphorylated tau in

In patients who have been diagnosed with aMCI, changes to the parahippocampal region are already apparent. It is up for debate whether the investigation of the entire brain or just volumes of interest (VOIs) are better at predicting conversion from aMCI to AD, but in a re‐ cent meta-analysis of work using data from the Alzheimer's Disease Neuroimaging Initia‐ tive (ADNI) only four methods were able to distinguish those who would convert more accurately than random chance. None of the four were more statistically reliable than the others, but three examined VOIs (Voxel-STAND, 57% sensitivity and 78% specificity; Voxel-COMPARE, 62% sensitivity and 67% specificity; Hippo-Volume, 62% sensitivity and 69% specificity) while only one examined the entire brain (Thickness-Direct, 32% sensitivity and 91% specificity) [11,13-15]. A protocol devised by Chincarini et al. to sample several VOIs has demonstrated a method of separating converters from non-converters with a sensitivity of 71% and a specificity of 65% [16,17]. Another method for predicting conversion is examin‐ ing hippocampal shape, and Costafreda et. al. were able to develop a method with 77% sen‐

Patients that are at-risk for AD but have no cognitive deficit are much more difficult to iden‐ tify. Most studies have been done in carriers of the ApoE ε4 allele, however it is important to remember that these studies have been cross sectional, and therefore may reflect a conse‐ quence of the gene that makes carriers more susceptible to AD, but not necessarily a stage of AD itself. There have been cortical thinning signatures identified in children, adolescent, and young adult carriers of the ε4 allele. These signatures reflect reductions in dorsolateral and medial prefrontal, lateral, temporal, and parietal cortices. [20-22]. Middle-aged carriers Unlike T1 weighted imaging, T2 imaging relies on the dephasing of the magnetization vec‐ tor in the transverse plane. T2 weighting, specifically FLuid Attenuated Inversion Recovery (FLAIR) imaging, is used to identify White Matter Hyperintensities (WMH), which are in‐ creased in AD [27]. In contrast, diffusion tensor imaging (DTI) is able to indirectly measure the integrity of myelin sheaths surrounding white matter tracts, and Susceptibility weighted imaging (SWI) is able to distinguish tissues at a high resolution based on several properties. *FLuid Attenuated Inversion Recovery (FLAIR) and Diffusion Tensor Imaging (DTI)*

If simply T2 weighted imaging was used, the signal from Cerebrospinal Fluid (CSF) is strong and therefore very bright (T2 of CSF ~ 600 ms at 3T). This makes it difficult to see subtle abnormalities in the white matter regions that partial volume with CSF. FLAIR imag‐ ing nulls the signal from CSF so that the image is focused solely on the white matter. The first RF pulse inverts the magnetization by 180 degrees. Then, when the longitudinal mag‐ netization for the CSF = 0, an excitation pulse and readout is applied. Because T1 of CSF (~4000 ms at 3T) is much longer than that of tissue (T1~700-1200 ms at 3T), residual tissue signal remains at the time of the CSF nulling.

DTI measures fractional anisotropy (FA), a quantitative measure of the coordinated move‐ ment of water molecules. FA assumes that the stronger a white matter tract is, the more like‐ ly the water molecules will be to move along the tract rather than sideways within the myelin sheath. If the myelin sheath is damaged it becomes easier for water molecules to dif‐ fuse through it, and the FA value will decrease.

The loss of white matter integrity, either through WMH or FA differences, may correlate with increasing cognitive impairment [28,29]. In AD populations reduced FA values have been found in frontal and temporal lobes, the posterior cingulum, the corpus callosum, the superior longitudinal fasciculus and the uncinate fasciculus [30]. Both WMH and FA have been found to distinguish normal aging from aMCI[31] and predict conversion from aMCI to AD[32]. Results have differed in whether they correlate with ApoE ε4 status, with some studies saying they do not [33,34], while several others say they do [35-37]. Note that the studies that claim white matter integrity correlates with ApoE ε4 status are more recent, and their ability to detect differences are likely more sensitive. White matter integrity has also been found to correlate with a family history of AD regardless of ApoE status [38,39].

**3.3. Future of structural imaging**

across medical centers [41].

**4. Functional imaging**

based functional imaging techniques.

**4.1. BOLD fMRI**

There is still a lot of work to be done in structural imaging. Most clinical studies to date have used 1.5 Tesla (T) scanners, however many medical centers now have 3T scanners and there are approximately 50 7T scanners worldwide. These high-field scanners allow for increased resolution, and provide better spatial resolution for observing structural changes in the same scan time. Although 7T scanners are not yet FDA approved for clinical use, they are

Using Magnetic Resonance Imaging in the Early Detection of Alzheimer's Disease

http://dx.doi.org/10.5772/54445

233

Many atrophy measurements are made either through a trained radiologist's visual assess‐ ment, or by manually tracing the area of interest. As such, the measurement of atrophy can be subjective, and is not always reproducible across testing site. In fact, one study found that the ability of radiologists to diagnose subjects based on atrophy alone had a specificity of 85% and a sensitivity of only 27% [20]. The introduction of FDA-approved methods that can automatically detect atrophy will create standardization of the field, and decrease variability

While structural imaging is important to assess brain atrophy, the hope is that AD patholo‐ gy will be identified before neuronal death so that atrophy can be prevented. One current theory is that one of the major components leading to amyloid and tau pathologies could be vascular changes [42]. Two of the risk factors for AD are mutated forms of APP, and the ApoE ε4 isoform and both of these factors are involved in cholesterol processing. The inabil‐ ity of a neuron to clear amyloid plaques may be prognostic and indicate impaired blood flow as a risk factor for AD. While it is not immediately apparent how blood flow is contri‐ buting to AD, some vascular changes are being evaluated through the use of hemodynamic-

Functional magnetic resonance imaging, or fMRI is a way to gain insight into the functional processes occurring in the brain. Most fMRI modalities are based on the blood oxygenation level-dependent (BOLD) effect. This is an indirect method of tracking the activation or inac‐ tivation of brain regions relative to a baseline state, and is based on the idea that an active area will need more energy and consume more glucose and oxygen and therefore more blood will need to be directed to that area. More specifically, oxygenated and deoxygenated blood water have different intrinsic magnetic properties (oxygenated blood is diamagnetic and deoxygenated blood is paramagnetic) and therefore affect the T2 and T2\* relaxation times of surrounding water in blood and tissue in different ways. Deoxygenated blood has a strong enough magnetic affect (paramagnetic) that it will distort the local field and decrease the signal intensity (i.e. shorten T2) of surrounding water for that region. Oxygenated blood will not have the same effect, and therefore regions containing more oxygenated blood will have higher signal intensity (longer T2). Importantly, during functional activation the cere‐

already being utilized in neuroimaging research, including in patients with AD.

White matter hyperintensities are associated with vascular abnormalities and therefore highly correlated with cardiovascular disease. For this reason, many clinicians will exclude a diagnosis of AD if there are many apparent WMH and instead diagnose the patient with vascular dementia [32]. Many non-amnestic MCI patients tend to have a higher degree of cardiovascular disease than those with aMCI or AD, however aMCI and AD patients have increased WMH scores. For this reason, increased WMH scores in cognitively impaired indi‐ viduals is likely associated with neurological disease rather than vascular disease [32]. *Susceptibility Weighted Imaging (SWI)*

Susceptibility weighted imaging is a method that can discriminate tissue content with a high level of resolution based on the tissue's intrinsic magnetic properties. SWI uses T2\* weighting along with magnitude and phase information to enhance contrast, and when combined with traditional MR weighting it can be used to detect small differences in susceptibility between blood and tissue. It is particularly useful for detecting cerebral mi‐ crobleeds because it can exploit the magnetic properties of blood since the susceptibility effects from fully oxygenated (arterial) and partially de-oxygenated (venous) blood wa‐ ter, and tissue, varies greatly – especially at high field strength. It can also be used to measure the iron content of a tissue.

Microbleeds are inversely correlated with performance during cognitive testing in healthy older adults, although this finding has never reached significance in an AD population [13,14]. SWI would allow for improved visualization of microbleeds so that if there is a rela‐ tionship between microbleeds and susceptibility to AD pathology, it can be recognized. Techniques are being developed that semi-automatically detect cerebral microbleeds with little human interference. These would significantly reduce the processing time and stand‐ ardize the quantification of microbleeds across patients and imaging centers.

In addition to microbleeds, one marker of oxidative stress is an increase in a tissue's iron content. Iron levels are highly elevated in AD patients as well as those with aMCI, and it is thought that changes in iron content may be detectible decades before the onset of the disease [16]. There is a theory that Aβ deposition may occur as a cellular response to an increased level of iron, and this is one of the underlying causes of amyloid plaque formation[40]. SWI has been shown to be a promising method to non-invasively assess iron distribution, and determine if there is a link between iron accumulation and the on‐ set of AD pathology [18].

SWI has only been used as a technique since 2004, which makes it very new technology. Al‐ though it has not yet been used in an at-risk population, SWI studies will likely be important tools in assessing AD risk.

#### **3.3. Future of structural imaging**

to AD[32]. Results have differed in whether they correlate with ApoE ε4 status, with some studies saying they do not [33,34], while several others say they do [35-37]. Note that the studies that claim white matter integrity correlates with ApoE ε4 status are more recent, and their ability to detect differences are likely more sensitive. White matter integrity has also

White matter hyperintensities are associated with vascular abnormalities and therefore highly correlated with cardiovascular disease. For this reason, many clinicians will exclude a diagnosis of AD if there are many apparent WMH and instead diagnose the patient with vascular dementia [32]. Many non-amnestic MCI patients tend to have a higher degree of cardiovascular disease than those with aMCI or AD, however aMCI and AD patients have increased WMH scores. For this reason, increased WMH scores in cognitively impaired indi‐ viduals is likely associated with neurological disease rather than vascular disease [32].

Susceptibility weighted imaging is a method that can discriminate tissue content with a high level of resolution based on the tissue's intrinsic magnetic properties. SWI uses T2\* weighting along with magnitude and phase information to enhance contrast, and when combined with traditional MR weighting it can be used to detect small differences in susceptibility between blood and tissue. It is particularly useful for detecting cerebral mi‐ crobleeds because it can exploit the magnetic properties of blood since the susceptibility effects from fully oxygenated (arterial) and partially de-oxygenated (venous) blood wa‐ ter, and tissue, varies greatly – especially at high field strength. It can also be used to

Microbleeds are inversely correlated with performance during cognitive testing in healthy older adults, although this finding has never reached significance in an AD population [13,14]. SWI would allow for improved visualization of microbleeds so that if there is a rela‐ tionship between microbleeds and susceptibility to AD pathology, it can be recognized. Techniques are being developed that semi-automatically detect cerebral microbleeds with little human interference. These would significantly reduce the processing time and stand‐

In addition to microbleeds, one marker of oxidative stress is an increase in a tissue's iron content. Iron levels are highly elevated in AD patients as well as those with aMCI, and it is thought that changes in iron content may be detectible decades before the onset of the disease [16]. There is a theory that Aβ deposition may occur as a cellular response to an increased level of iron, and this is one of the underlying causes of amyloid plaque formation[40]. SWI has been shown to be a promising method to non-invasively assess iron distribution, and determine if there is a link between iron accumulation and the on‐

SWI has only been used as a technique since 2004, which makes it very new technology. Al‐ though it has not yet been used in an at-risk population, SWI studies will likely be important

ardize the quantification of microbleeds across patients and imaging centers.

been found to correlate with a family history of AD regardless of ApoE status [38,39].

*Susceptibility Weighted Imaging (SWI)*

232 Understanding Alzheimer's Disease

measure the iron content of a tissue.

set of AD pathology [18].

tools in assessing AD risk.

There is still a lot of work to be done in structural imaging. Most clinical studies to date have used 1.5 Tesla (T) scanners, however many medical centers now have 3T scanners and there are approximately 50 7T scanners worldwide. These high-field scanners allow for increased resolution, and provide better spatial resolution for observing structural changes in the same scan time. Although 7T scanners are not yet FDA approved for clinical use, they are already being utilized in neuroimaging research, including in patients with AD.

Many atrophy measurements are made either through a trained radiologist's visual assess‐ ment, or by manually tracing the area of interest. As such, the measurement of atrophy can be subjective, and is not always reproducible across testing site. In fact, one study found that the ability of radiologists to diagnose subjects based on atrophy alone had a specificity of 85% and a sensitivity of only 27% [20]. The introduction of FDA-approved methods that can automatically detect atrophy will create standardization of the field, and decrease variability across medical centers [41].
