The most established visual rating scale for cortical atrophy is the Pasquier scale (Pasquier et al. 1996) which is a four-point (0–3) scale considering the width of the sulci and the volume of the gyri (Fig. 24.2). Cortical atrophy can be either rated globally (global cortical atrophy) or regionally for different anatomic regions (e.g., separate scores for each lobe). For the visual assessment of the posterior brain regions which are of special interest in the diagnosis of AD, a specially developed four-point (0–3) rating scale has been introduced that considers certain anatomical landmarks of the posterior brain region such as the posterior cingulate and parieto-occipital sulcus as well as the precuneus and parietal cortex (Fig. 24.10). First studies using this rating scale have conclusively demonstrated its diagnostic value particularly in terms of differentiating AD patients from other types of dementia (Koedam et al. 2011; Lehmann et al. 2012).

Since the anatomical structures and landmarks in the medial temporal lobe, particularly the hippocampus, are rather small and show a complex interaction, visual rating of MTA is more complex and requires sufficient training. The most established MTA rating scale is the five-point Scheltens scale based on the width of the choroid fissure, temporal horn, and the volume/height of the hippocampus (Table 24.2, Fig. 24.1) (Scheltens et al. 1992). In more advanced atrophy stages, consideration of adjacent anatomical structures (e.g., collateral sulcus) can further help stage the MTA. This scale allows a fast and reproducible assessment of MTA correlating well with volumetric measurements and histopathology. It is currently widely used in the clinical routine setting as well as for clinical trials.

In addition to neurodegenerative changes, the detection and classification of vascular pathology is crucial in patients with cognitive decline. This is of particular interest in patients with a clinical presentation suggestive of VaD since vascular pathology detected by neuroimaging has been incorporated by the NINDS-AIREN criteria for VaD (van Straaten et al. 2004; Román et al. 1993). However, there is also increasing evidence that vascular pathology acts in a synergistic way with neurodegenerative pathology, having relevant implications on clinical outcome measures (Scheltens et al. 1995; Inzitari et al. 2009). White matter vascular pathology can be most sensitively detected on FLAIR sequences. The so-called white matter hyperintensities (WMH) located in the deep white matter and periventricular white matter are markers of small-vessel disease that can be easily classified according to the Fazekas scale (Table 24.2, Fig. 24.3) (Fazekas et al. 1987). This scale is rather rough but robust and easily applicable, showing a high reproducibility and a good correlation with more in-depth rating scales and histopathology (Scheltens et al. 1998).

Another manifestation of small-vessel disease on CT and MRI that has been included into the NINDS-AIREN criteria for VaD is the presence of lacunar infarcts (lacunes). Lacunes are small end-artery infarcts accounting for 15–25 % of all strokes. Depending on the anatomic location, they can present as a silent infarct or cause focal neurological deficits (e.g., hemiparesis). On MRI and CT, old lacunar infarcts are focal areas of tissue loss sometimes with surrounding gliosis and hemosiderin deposits. Lacunes in the deep white matter can be sensitively detected on FLAIR images. In the basal ganglia and thalamus, lacunes are better visible on T2-weighted sequences (Figs. 24.5 and 24.6).

The detection of large-vessel disease manifestations such as territorial and watershed infarcts is also of major importance particularly when strategic brain areas (e.g., dominant hemisphere) are affected, which also is one of the key criteria of the NINDS-AIREN criteria for VaD.

The assessment of hemorrhagic brain lesions should include detection of primary intracranial hemorrhage as a possible treatable cause of dementia (e.g., due to hypertension) and secondary hemorrhagic lesions like trauma-induced hemorrhage.

MBs are increasingly gaining importance in the diagnostic work-up of patients with cognitive impair­ment. They are defined as small “dot-like” hypoin­tense brain lesions on T2-weighted GE or susceptibility-weighted images. Histopatho­logically, MBs include storage of hemosiderin deposits in surrounding macrophages. Concerning the underlying pathophysiological concept, there are two main different types of MBs.

The first type of MB is based on amyloid depositions (amyloid-β 40 and amyloid-β 42) in the vessel wall, presumably inducing an inflammatory reaction, leading to increased permeability and leakage of blood products into the brain parenchyma, in addition to increased vessel wall stiffness and reduced vasoreactivity. Most of the amyloid-related MBs are located in the cortical gray matter or in the border zone between the cortical gray matter and white matter, hence the term lobar MBs (Fig. 24.11). Lobar MBs are associated with cerebral amyloid angiopathy (CAA) and strongly related to AD pathology. The second type of MB is not amyloid related and is probably based on chronic vasculopathy as seen in chronic hypertension leading to degenerative processes in the vessel wall and leakage of blood products into the brain tissue. These MBs are located in deep brain structures such as the basal ganglia, deep white matter, and the posterior fossa and are referred to as non-lobar MBs (Fig. 24.11). The coincidental occurrence of both types of MBs on MRI is frequently observed in different types of dementia such as AD and VaD (Cordonnier and van der Flier 2011).

Since the MRI detection of MBs is based on susceptibility effects of blood degradation products (hemosiderin), sensitivity in the detection of MBs and therefore the prevalence of MBs depend on several imaging parameters such as pulse sequence, magnetic field strength, and spatial resolution. Using scanning parameters optimized for MB detection, in a healthy population, the prevalence of MBs is almost 25 % (Poels et al. 2011). In a memory clinic population, the overall prevalence of MBs is about 17 % and varies from 10 % in patients with subjective memory complaints and 18–20 % in AD patients to 65 % in patients with VaD (Cordonnier et al. 2006). MBs have clinical relevance in terms of other paraclinical biomarkers and clinical outcome measures. They are associated with vascular risk factors (hypertension most consistently), cognitive performance, age, mortality, apolipoprotein E genotype, and cerebrospinal fluid biomarkers (amyloid-β 40). In addition, they are associated with imaging findings of small-vessel diseases such as WMH and lacunar infarcts (Goos et al. 2009, 2010). There is also emerging evidence that lobar MBs might have some prognostic value for predicting AD, but there is still no way to predict VaD (Kirsch et al. 2009). MBs regardless of location, however, were found to predict progression of mild cognitive impairment patients to non-AD dementia (Staekenburg et al. 2009).

Amyloid-associated leakage of blood vessels can also occur in leptomeningeal vessels leading to hemosiderin deposition in the (sub)pial space, which has been termed superficial siderosis. Basically, superficial siderosis is a manifestation of leptomeningeal MBs. However, the imaging findings are completely different from classical MBs in the brain tissue. On T2-weighted GE and susceptibility-weighted MRI, superficial siderosis presents with a curve-like hypointense signal in the (sub)pial space covering the surface (gyri) of the brain (Fig. 24.11). From a pathophysiological point of view, superficial siderosis is related to CAA and corresponds to amyloid-related ­macrohemorrhage and MBs in the brain (Linn et al. 2010). The pathophysiogical similarities are reflected by a high coincidence of superficial siderosis and MBs. However, compared to MBs, the prevalence of superficial siderosis is quite low and is about 1 % in a normal healthy population (Vernooij et al. 2009). There is limited data available regarding the clinical and prognostic role of superficial siderosis as an imaging marker in a memory clinic population.

There are different purposes of serial imaging in a memory clinic setting such as establishing the diagnosis as well as disease and treatment monitoring.

The added value of serial imaging to establish the diagnosis of a specific underlying disease entity is still a matter of debate. If the diagnosis is fairly certain based on the clinical presentation, CSF biomarkers, and imaging findings, serial imaging does not have anything to offer. If the clinical diagnosis is unclear and the imaging findings are inconclusive, serial imaging might be helpful in establishing a definite diagnosis during follow-up. Given the fact that FDG-PET and amyloid PET are more sensitive than ­structural MRI or CT particularly in primary ­neurodegenerative disorders associated with dementia, these advanced imaging methods are likely to have more added value in establishing the diagnosis during follow-up.

Monitoring of disease progression can be of substantial clinical relevance particularly for the assessment of progressive white matter changes, in which it might lead to the diagnosis of VaD during follow-up and is likely to have clinical impact in terms of clinical outcome measures (Gouw et al. 2008; Inzitari et al. 2009). Progression of neurodegeneration can be monitored by measuring cortical atrophy and MTA either by visual rating or volumetric measurements (Figs. 24.12 and 24.15).

For treatment monitoring, progression of vascular damage, and neurodegeneration, serial imaging is also important. However, for vaccination therapy in AD, the role of serial MRI includes also important safety findings including the early detection of treatment-induced inflammation, vasogenic edema, and (micro)hemorrhage. Results from serial amyloid imaging using ¹¹C-PiB during amyloid-vaccination therapy trials have been reported (Rinne et al. 2010). However, serial amyloid-imaging documenting increase or decrease of amyloid burden in AD patients is not yet performed in a clinical routine setting.

Similar to other organs, the brain goes through stages of development during lifetime, particularly at later ages. In general, aging of the brain can be subdivided into three major categories:

The global brain volume decreases slowly by about 0.1–0.2 % per year from the second decade of life until the age of 60–70, which accelerates during the following years with a loss of 0.3–0.5 % per year. Age-related global brain atrophy is mainly based on cortical thinning and white matter loss, and shows regional differences. The medial temporal lobe is relatively spared in the early normal aging process. Similar to global atrophy, MTA variably accelerates after age 70. Therefore, in people younger than 70 years, a visual MTA score of 1 is considered as normal. In geriatric populations, even an MTA score of 2 can be considered normal aging related. Small cavities within the hippocampus (e.g., cysts) are frequently observed in normal brains of elderly persons and are not associated with neurodegenerative pathology (Bastos-Leite et al. 2006). Slight enlargement of perivascular (Virchow-Robin) spaces can be observed in normal aging brains particularly at the cortical gray matter-white matter interface close to the vertex. Extensive pathologic enlargement of perivascular spaces ­frequently observed in the basal ganglia (e.g., “état criblé”) is associated with vascular pathology and neurodegeneration and is not a normal aging phenomenon (Zhu et al. 2010; Chen et al. 2011). The formation of WMH is a frequent finding in normal aging brains and represents focal areas of hypoxic demyelination due to hypoperfusion/incomplete infarction. Punctiform WMH do not have clinical relevance in terms of cognitive clinical outcome measures and are ­considered a normal aging ­phenomenon. However, rapidly progressive and confluent WMH have to be considered as signs of pathological aging presumably associated with underlying cardiovascular risk comorbidity. Although MBs are associated with vascular pathology detected on MRI, they may also be found in normal aging persons (see above).

Another feature of the normal aging brain is neuronal iron accumulation (intraneuronal ferritin). This involves most notably the basal ganglia region particularly the globus pallidus but also the striatum, subthalamic nucleus, substantia nigra, and the dentate nucleus of the cerebellum. The brain iron accumulation usually starts in the globus pallidus, substantia nigra, dentate nucleus, and to a lesser extent in the putamen, caudate nucleus, thalamus, and cortical gray matter. Brain iron can be detected as hypointensities on T2-weighted MRI. Using higher magnetic field strengths and dedicated pulse sequences (e.g., susceptibility-weighted imaging) paramagnetic brain iron can be sensitively differentiated from diamagnetic substances such as calcifications. Extensive (pathological) iron deposition which can be observed in primary neurodegenerative diseases associated with dementia, however, can induce oxidative stress and increase the concentration of free oxygen radicals and neurotoxicity.

In addition to findings on structural neuroimaging, complex metabolic, microstructural, and functional changes occur in normal aging brains which can be measured by ¹H-MR spectroscopy, magnetization transfer MRI, as well as perfusion and functional MRI (Barkhof et al. 2011). It is worth noting that 25–35 % of normal healthy persons show cortical amyloid deposition particularly located in the frontal and posterior (posterior cingulate, precuneus) cortical areas detected on ¹¹C-PiB amyloid PET. This is in accordance with histopathological postmortem studies showing amyloid-β plaques in approximately 30 % of healthy controls. Since the clinical relevance of these amyloid depositions is not completely understood, this should not be considered as a normal aging phenomenon. In healthy controls with discrete cognitive symptoms, positive ­amyloid PET has a predictive value in terms of further cognitive decline. This supports the theory that amyloid depositions precede clinical symptoms in terms of cognitive decline by many years. On the other hand, absence of amyloid depositions based on amyloid PET makes development of AD in healthy persons unlikely.

AD is the most common neurodegenerative disease associated with dementia in elderly. Histopathologically, AD is characterized by amyloid-β and hyperphosphorylated tau accumulation (McKhann et al. 1984). Despite the fact that AD is primarily a clinical diagnosis, the role of neuroimaging in supporting the clinical diagnosis of AD is becoming increasingly important. This is corroborated by the new diagnostic ­criteria for AD in which structural neuroimaging findings (MTA) as well as PET findings (temporoparietal hypometabolism on FDG-PET, amyloid deposition on amyloid PET) have been incorporated (Dubois et al. 2007).

According to the neuropathological stages of AD pathology described by Braak and Braak (1991), atrophy of the medial temporal lobe, in particular affecting the hippocampus and best seen on oblique coronal T1-weighted MRI, is one of the first signs of neurodegeneration in senile (late-onset) AD patients (Figs. 24.13 and 24.14). Visual rating of the MTA allows a robust and reproducible assessment of the degree of atrophy and can help to distinguish AD patients from those without cognitive complaints with sensitivity and specificity values of 80–85 % (Frisoni et al. 2010).