Radiological assessment of dementia: the Italian inter-society consensus for a practical and clinically oriented guide to image acquisition, evaluation, and reporting

Background Radiological evaluation of dementia is expected to increase more and more in routine practice due to both the primary role of neuroimaging in the diagnostic pathway and the increasing incidence of the disease. Despite this, radiologists often do not follow a disease-oriented approach to image interpretation, for several reasons, leading to reports of limited value to clinicians. In our work, through an intersocietal consensus on the main mandatory knowledge about dementia, we proposed a disease-oriented protocol to optimize and standardize the acquisition/evaluation/interpretation and reporting of radiological images. Our main purpose is to provide a practical guideline for the radiologist to help increase the effectiveness of interdisciplinary dialogue and diagnostic accuracy in daily practice. Results We defined key clinical and imaging features of the dementias (A), recommended MRI protocol (B), proposed a disease-oriented imaging evaluation and interpretation (C) and report (D) with a glimpse to future avenues (E). The proposed radiological practice is to systematically evaluate and score atrophy, white matter changes, microbleeds, small vessel disease, consider the use of quantitative measures using commercial software tools critically, and adopt a structured disease-oriented report. Summary statement In the expanding field of cognitive disorders, the only effective assessment approach is the standardized disease-oriented one, which includes a multidisciplinary integration of the clinical picture, MRI, CSF and blood biomarkers and nuclear medicine. Supplementary Information The online version contains supplementary material available at 10.1007/s11547-022-01534-0.

2. Brain atrophy, white matter lesions, microhemorrhages, and vascular diseases should be radiologically evaluated. 3. Radiological assessment should be based on visual assessment, scoring, and volumetry. 4. The disease-oriented structured radiology report increases its clinical value. 5. Multidisciplinary teamwork increases diagnostic accuracy.

Background
Dementias are taking up more space in everyday clinical and radiological scenarios, considering that the worldwide population is aging [78] and that brain imaging plays a key role in the assessment of cognitive impairment [8]. In addition, the scientific community is increasingly focused on the debate over the use of AI [51], FDA-and EC-approved automated segmentation software [73], and the optimal use of neuroimaging in new drug trials [68]." A recent survey, carried out in Europe among academic and non-academic institutions [74], disclosed that the current practice in dementia imaging presents some homogeneity (mainly in imaging acquisition and image interpretation) but also differences in training and reporting, in using advanced imaging techniques and volumetric measures, as well as in communication between clinicians and radiologists. This work stems from the need of different Italian scientific societies to standardize and optimize the radiological approach for the assessment and follow-up of the aging brain and cognitive disorders. We aim to fill this gap of variability and uncertainty, providing a practical approach in evaluating, interpreting, and monitoring the aging brain and main cognitive disorders.

Methods
The promoters of the initiative (FBP and SB) representatives of the Neuroradiological Section of the Italian Society of Radiology (SIRM), with the clinical support of EC and GBF, created a core panel with experts in dementia and cognitive disorders representatives of the Italian Association of Diagnostic and Interventional Neuroradiology (AINR). The purpose of their work is to submit a preliminary consensus draft to the representatives of SIGG (Italian Society of Geriatrics and Gerontology, Società Italiana di Geriatria e Gerontologia) and AIP (Italian Association of Psychogeriatrics, Associazione Italiana di Psicogeriatria) for their revision and final approval.
The main research questions were: • Define the key concepts (A) of what the radiologist needs to know: main clinical features (definition of brain aging, cognitive syndromes, and primary and secondary dementias) and imaging findings. • Frame the MRI radiological approach for the assessment and follow-up of the aging brain and cognitive disorders-MRI protocol (B), imaging evaluation and interpretation (C), and reporting (D) for clinical use • Identify the main factors that will influence clinical and radiological practice in this population in the near future (E) • The consensus between experts was reached using a similar approach to the previously published paper (Pizzini FB et. Insights Imaging) and consisted in: • A critical review of previous literature by European/ American task forces and scientific societies related to A-D • Circulation and discussion of the draft based on this review among the core panel and then between the experts in more rounds • Changes of the original draft till the group converged towards an agreement on all the points A-E

Literature review
Literature review was performed through the PubMed database and on the web through Google and Google Scholar platforms, as well as specialized websites (Radiopaedia.org and radiologyassistant.nl/neuroradiology) and textbooks. The standardized strings used to search the database for literature were structured by combining the keywords (1) disease of interest, (2) biomarker (3) guidelines/recommendations/evaluation. As shown in the flowchart (Fig. 1), only original texts (abstract and full text) published in English were considered, without filtering for article type and publication date. Articles were selected after a review of the titles and abstracts of the first fifteen "best matches" to determine relevance and affinity to the research purpose. When it was useful, consultation was extended to the bibliographic references of the selected articles. Finally, in our bibliography, articles range from 1988 to 2022, with a predominance of the last decade.

(A) Introduction to aging, cognitive impairment, and dementia
The following boxes represent consensus findings related to the key concepts (Box 1) and clinical features and imaging findings (Boxes 2, 3, 4, 5, 6, 7, 8, 9, 10 and 11) of the major primary and secondary dementias.

(B) MRI protocol (acquisition)
1. Field of the MRI suite (1.5; 3T). The choice of field strength does not affect the evaluation of atrophy or white matter load (WML), but it can affect the detection rate of microbleeds [46]. 2. Standard protocol (core sequences and parameters).
They have been found to have a major impact on image resolution [15] thus on the detection rate of atrophy, white matter changes, and microbleeds (e.g. in GRE detection of 33% of the microbleeds identified by thinsection SW [46,65]  Global cerebral Atrophy (GCA) [49], Koedam for posterior lobes-which are most affected in atypical AD [34], MTA-Scheltens for medial temporal lobe [60].
(b) WHITE MATTER CHANGES (WMC) (see Tables 5  and 6; Fig. 5) should be evaluated in FLAIR/T2. The most used scales are Fazekas [17] and age-related white matter changes (ARWMC) [76]. The WMC have variable size, but minimal diameter of the lesions at imaging > 1 mm (in any plane). (c) MICROBLEEDS should be evaluated in GRE T2*/ SWI and the minimal diameter of the lesions at imaging is < 10 mm in any plane (Fig. 6). They could be a feature of small vessel disease (hypertension or cerebral amyloid angiopathy, CAA) and could be related to antithrombotic bleeding risk. All possible microbleeds mimics should be excluded (i.e., vessels, small cavernomas, mineralization foci, artifacts at the air-bone interface and due to partial volume, small hemorrhagic areas due to infarcts or other bleeds).
An example of scale with high intrarater and interrater reliability is MARS (Microbleed Anatomical Rating Scale) which describes their number and location in lobar and/or infratentorial and/or deep regions [23].  [69], which is equivalent to that of a severe enlargement of the perivascular spaces and/or Fazekas 3, thus indicating a higher risk of clinical consequence (ischemic and hemorrhagic brain events, dementia) [47].

Volumetric measures
Visual differentiation between brain changes due to aging or to an early stage of the disease can be difficult, so the quantification of brain structures from a single patient and its comparison to age and sex-specific reference MRI data of healthy population can improve the diagnosis (Fig. 7). Several volumetric brain assessment methods and commercial Regulatory Authority approved (e.g., FDA, CE, CFDA marking) software are clinically available and implemented in radiology reporting, but without a clear strategy in the assessment. One way to improve the diagnostic accuracy of the use Box 1 Glossary and definitions See dedicated Boxes for further details about AD, FTD, CBD, sCJD, iNPH, CAA, Alcoholic dementias, DLB, VaD NORMAL AGING: certain cognitive performances on formal testing are spared over the lifetime (e.g., visual perception/recognition), others decline after the sixth decade. These late-life changes involve especially the episodic and working memory, and executive function as well. Neuroimaging (NI) discloses variable structural brain findings, with an approximative rate of total brain volume loss per decade up to 5% after the age of 70 SUCCESSFUL AGING: this term identifies older individuals who are free from chronic disease and continue to function well into old age (both physically and cognitively). NI reveals larger brain structure volumes with the absence of vascular lesions than the age-matched population MILD COGNITIVE IMPAIRMENT (MCI): intermediate clinical state between normal cognition and dementia (with increased risk for progressing). NI can detect signs of brain pathologies (atrophy, plaques, tangles, small and big vessels diseases, etc.) DEMENTIA: decline in cognition involving one or more cognitive domains. Deficits represent a decline from previous level of function and interfere with daily function and independence. NI findings vary according to the etiology of the cognitive impairment PRIMARY DEMENTIAS: neurological diseases whose manifestations are predominantly cognitive. Most primary dementias are caused by neurodegenerative proteinopathies (e.g., tauopathy, prionopathies, α-synucleinopathies, etc.) which lead to neuronal loss and neuroinflammation with glial reaction. Neurodegenerative dementias include Alzheimer's disease (AD), Lewy body dementia (LBD), frontotemporal dementias (FTD), and prion diseases (sCJD). Vascular dementia (vaD), though not a neurodegenerative condition, is grouped together with primary dementias SECONDARY DEMENTIAS: cognitive dysfunction is caused by structural lesions of the brain (e.g., normal pressure hydrocephalus [iNPH], brain tumors, etc.), nutritional deficiencies (vitamin B12 deficiency, folate deficiency, niacin, and thiamin deficiency), endocrine disease (hypothyroidism, hyperthyroidism, etc.), inflammatory disease (e.g., systemic lupus erythematosus, vasculitis, and sarcoidosis), infectious diseases (e.g., Whipple's disease, Lyme disease, and AIDS), alcoholic dementia and metabolic causes (e.g., hypoxia and dialysis) REVERSIBLE DEMENTIAS: secondary dementias are, to varying degrees, treatable and potentially reversible MIXED DEMENTIA: also known as multiple etiology dementia; it is a combination of multiple pathologies where each has a clinical impact (e.g. vascular and degenerative, traumatic and vascular, etc.)

Box 2
Late-onset Alzheimer Disease (Lo-AD) ┼ Compared to early-onset AD, patients with late-onset disease show greater medial temporal atrophy and less cortical atrophy ¶ The involvement of precuneus is a late feature of lo-ad or an early feature of eo-AD [see Box 3] Δ Usually the FDG-PET findings correspond to the MRI atrophic changes, but in some cases the molecular findings may be more severe than structural findings Generals -The most common cause of late-onset (≥ 65 yo) dementia [52] -Incidence and prevalence increase with age (growing up because of an aging population) [1] -Cerebrovascular disease frequently coexists with AD -Cerebral amyloid angiopathy (CAA) can co-occur with AD -General neurological examination is substantially normal -Neuropsychiatric symptoms are common

Clinical features and diagnosis
Typical presentation: -Early and prominent episodic memory loss (recent events) plus -Executive, language, and visuospatial impairment MRI -Disproportionate bilateral (or mild asymmetric) temporoparietal cortical atrophy with (early) prevalent involvement of entorhinal cortex/ medial temporal lobes (± hippocampus ┼) and (more later) precuneus ¶ and posterior cingulate gyrus -Relative sparing of the primary motor&somatosensory (pericentral) and occipital cortex FDG-PET: mirrors MRI findings, with gross correspondence between hypometabolic and atrophic areas Pitfalls: -Iron-sensitive sequences should be performed to assess for hemorrhages associated with CAA (see Box 9) [22] -SVD associated WMHs frequent coexist with AD findings

Box 3
Early-onset Alzheimer Disease (Eo-AD: "not just an AD at a younger age") ┼ Other forms of PPA, such as the nonfluent/agrammatic and semantic variants, are non-Alzheimer syndromes due to FTLD [see Box 6] ¶ Gerstmann syndrome: acalculia, left-right disorientation, finger agnosia, and agraphia Δ Balint syndrome: ocular motor apraxia, optic ataxia, and simultanagnosia ◊ These patients overlap with or may eventually meet current criteria for posterior cortical atrophy or corticobasal syndrome Generals -The most common cause of early-onset (< 65 yo) neurodegenerative dementia [41] -AD with clinical onset younger than 65 yo -Better memory but worse attention/language/executive functions/visuospatial skills than Lo-AD

Clinical features and diagnosis
Up to two-thirds of patients with non-amnestic phenotypic variants: - -Possible more uptake than Late-onset-AD due to increased Tau burden in posterior neocortices

Pitfall:
if suspicious of PCA is clinically high, the absence of marked parieto-occipital atrophy should not exclude the diagnosis Box 4 Vascular dementia (VaD)-Vascular cognitive impairment (VCI) "not strictly a neurodegenerative dementia ┼" Generals -The second most common type of dementia (after AD) -May occur with single or multiple infarcts, cortical or subcortical -"Silent" strokes are significant risk factors -Often multiple-etiology dementia (VaD plus AD) -"Classically" stepwise dementia • Large vessel or atherothromboembolic disease -Multiple infarcts (e.g., bilateral anterior cerebral artery infarction, and parietotemporal plus temporo-occipital infarction of the dominant hemisphere) -Single strategically placed infarct → medial thalamus; lateral thalamus-internal capsule; caudate and pallidus; posterior cerebral artery infarction (with infarction of the paramedian thalamic region and inferior medial temporal lobe of the dominant hemisphere); left angular gyrus (Gerstman's syndrome; basal forebrain infarction) • Small vessel disease (relevant role) -Multiple (> 2) lacunar infarcts in (frontal) white matter and deep gray matter nuclei ¥ -Ischemic white matter changes (WMLs) involving at least more than 25% of the whole WM -Dilatation of perivascular spaces -Cortical microinfarcts and microhemorrhages • Hemorrhage: -Intracerebral hemorrhage -Multiple cortical and subcortical microbleeds -Subarachnoid hemorrhage • Hypoperfusion -Hippocampal sclerosis -Laminar cortical sclerosis -Watershed infarcts (aka border zone infarcts) in the dominant hemisphere (superior frontal and parietal regions) ╪ FDG-PET: mirrors MRI findings, with gross correspondence between hypometabolic and atrophic areas Ioflupane-SPECT (DaTscan): normal uptake Amyloid-PET: not useful (maybe abnormal uptake) τ-PET: no uptake of the software in clinical practice is the double assessment-visual and quantitative-which combines the visual rating and the atrophy measurements [73].

Follow up
If there are any vascular findings at MRI, an annual follow-up is recommended.
In case of trials or other pathologies, the control MRI should be scheduled according to clinical indications.
┼ Vascular-induced dementia should be included in the differential diagnosis for patients with neurocognitive impairment. however, VaD should be considered a secondary dementia, not a neurodegenerative disorder [see Box 1] ¶ The term Binswanger disease is currently out of use Δ Also account for anterior cranial fossa lesions and NPH in the differential diagnosis [see Box 11] ◊ Notably, bilateral findings indicate basilar artery involvement up to the "top of the basilar" syndrome § Adapted from VASCOG work, p. Sachdev et al. 2014 ¥ Notably, multi-infarct dementia represents a subtype of VaD (not a synonymous as in the past) ╪ Watershed cerebral infarctions are due to hypoperfusion or microemboli. Two types can be distinguished, with corresponding pathological/ radiological aspects: -Cortical → cortical laminar necrosis (early cytotoxic edema [↑DWI/↓ADC], T1 curvilinear hyperintensities, ↑/ ~ T2/FLAIR) Deep → pearl thread sign (DWI and T2/FLAIR signal abnormality as a series of rounded areas adjacent to the lateral ventricle) [57] Box 4 (continued)

Generals
-The second most common type of dementia (after AD) -May occur with single or multiple infarcts, cortical or subcortical -"Silent" strokes are significant risk factors -Often multiple-etiology dementia (VaD plus AD) -"Classically" stepwise dementia Pitfall: -WMH burden, even if moderately severe, is not sufficient evidence for VaD (WMH are not specific for cerebrovascular disease) -AD MRI findings may coexist -Besides "classical" types of vascular lesions (i.e., atherosclerosis, cardiac/atherosclerotic/systemic emboli, cerebral venous thrombosis, arteriolosclerosis, etc.) consider angiopathies (with and without inflammation), arteriovenous fistulas, hereditary angiopathies (e.g., CADASIL, CARASIL, etc.) and CAA [see Box 9] -In case of acute onset of cognitive impairment (including delirium) and/or focal neurologicalsigns/symptoms (hemispheric and/or vertebrobasilar), perform urgent brain CT and CT angiography ± advanced neuroimaging (i.e. perfusion/DWI studies) to assess acute stroke/stroke mimics and guide treatment (e.g., IVT and/or EMT if ischemic stroke)

1.
A structured reporting is often not considered useful in clinical practice and could present other critical issues [74], so a guided report is preferable (Fig. 8). We recommend using the following template, modified from a previous ESNR dementia working group 2019 proposal.
Please consider that it is advisable to mention in the "Conclusions/Impressions" of the report: • Any individual differences from a control population (cross-sectional assessment) by applying the visual qualitative and rating scale or/and the volumetric assessment • Any stability or longitudinal worsening of the radiological findings (longitudinal evaluation) from previous radiological examinations, if appropriate for comparison To do this, when evaluating brain atrophy, it is useful to take into account existing reference standards for assessing differences between a subject and the control population and individual rates of change over the life course even with respect to the trajectories of volumetric brain imaging markers [7,75].
According to these large and inclusive datasets currently available (BrainChart open source and Rotterdam study), the trajectories of volumetric changes in gray matter, white matter, and third ventricles show nonlinear curves, with accelerated change with advancing age and some differences between men and women.
Regarding the "mixed pathology" reported in "Conclusions/Impressions," it should be emphasized that the diagnosis of "mixed dementia" is clinical, not neuroradiological. The neuroradiological description of, for example, hippocampal atrophy and Fazekas 3, does not mean that the  -Acquired: iatrogenic Creutzfeldt-Jakob disease (iCJD, < 1%), and variant Creutzfeldt-Jakob disease (vCJD, < 1%) ¶ Although the prevalence of cerebellar symptoms and neuropathological findings, DWI-and FLAIR-hyperintense signal in the cerebellum is not typical [12] Δ Subtypes of sCJD are classified according to the genotype of the prion protein (prnp) gene codon 129 and the molecular properties. Clinical phenotypes agree with molecular subtypes (e.g., MM1 and MV1 correlate with the "classic CJD"; VV2 is the "ataxic variant", MM2 can present as either a thalamic variant or a cortical variant, etc.)

Generals
-Rare disease (incidence 1/1.000.000; mean age 62yo) -Most frequent (90%) of sporadic human prion disease Clinical features & diagnosis -Two cardinal clinical manifestations: • Rapidly progressive mental deterioration (akinetic mutism in the end-stage) • Myoclonus (90%) -Gait disturbance and various among pyramidal (especially in end-stages), extrapyramidal, and cerebellar manifestations (presenting symptoms in up to 40% of cases) -Brain MRI is the most sensitive diagnostic test in the early stages MRI: -FLAIR/T2-hyperintensity and DWI reduced diffusivity (unilateral or bilateral; focal, multifocal, or diffuse; and asymmetrical or symmetrical in early stages, then tendency to greater symmetrical involvement) in the head of the caudate and putamen (lesser extent in the thalamus) -Cortical ribbon especially involving the superior frontal gyrus, superior parietal lobule, cingulate gyrus, and insular cortex -Perirolandic cortex is usually spared -Generalized atrophy and ventricular dilatation (prominent in later stages) -Rare isolated limbic involvement -Possible late Cerebellar atrophy ¶ -Confluent hyperintense signal in the mesial and dorsal thalami on DWI, FLAIR, and T2-weighted MRI ("double hockey stick" or "pulvinar" sign) → typical of variant CJD (vCJD or 'mad cow disease', now rare) and rare cases of sCJD MRI abnormalities may vary with the clinical syndrome and molecular subtype [40]

Δ
The early-phase 11c-PiB PET help differentiate probable CAA from probable AD. Early-phase amyloid PET images are used as a surrogate for brain perfusion, as opposed to the standard late-phase 11c-PiB PET/CT which reflects binding to aβ deposits ◊ When predominantly placed in the centrum semiovale, the dilated perivascular spaces are frequently associated with CAA [72] § Microbleeds are not specific for CAA as they can be seen in multiple conditions, including, hypertension, cerebral cavernous malformations, coagulopathy, thrombocytopenia, anticoagulant medications, CNS vasculitis, infective endocarditis, end-stage kidney failure. Moreover, beside conditions above-mentioned, microbleeds are also more prevalent among users of antiplatelet agents [24]. Notably, while microbleeds (in general) are not specific of CAA, cortical ones (i.e., those limited to the cerebral and/cerebellar cortex and vermis), suggest CAA [48] (although not pathognomonic). In contrast, deep microbleeds (those involving the basal ganglia, thalamus, or pons) are presumably of hypertensive microangiopathic origin ¥ In those with ad treated with immunotherapy, CAA may developbecause of the rapiddestruction of parenchymal aβ, and/or ARIA may be triggered if CAA is preexistent to the immunotherapy Box 9 (1) the acute Wernicke encephalopathy (WE) syndrome (2) the chronic Korsakoff syndrome (KS, usually a consequence of WE) There is no generally accepted definition of KS, nor generally accepted criteria for the diagnosis of KS ¶ Several studies showed that the ventricles and sulci become significantly smaller with short-term alcohol abstinence (approximately one month of abstinence). Notably, cognitive abnormalities improve too with alcohol abstinence [4,62,82] Δ Similar to the reduction in ventricular dilation achieved with short-term abstinence, the amount of white matter also increases in response to alcohol withdrawal, suggesting a reversible dam-

age [19] Clinical features and diagnosis
Korsakoff syndrome ("a residual syndrome") ┼: -Memory disorder with selective anterograde and retrograde amnesia (which lead to confabulation) with apathy and relative preservation of long-term memory and other cognitive functions Most frequently represents a late neuropsychiatric manifestation of WE in alcohol abusers (80% of WE episodes), but also caused by a "non-alcoholic" deficit of thiamine -Sometimes occurs without a clinical recognized WE (that may have been subclinical or

Marchiafava-Bignami disease:
-Rare disorder (probable undiagnosed) -Demyelination or necrosis of the corpus callosum and adjacent subcortical white matter -Malnourished alcoholics -Dementia, spasticity, dysarthria, and inability to walk (acute, subacute, or chronic course) -Variable prognosis (coma and death, survival in a demented condition, interhemispheric disconnection syndrome, occasional recovery)

MRI/CT:
-Hypodense lesions in the corpus callosum (CT) -Discrete or confluent areas of decreased T1 signal and increased T2 signal in the corpus callosum (MRI)

Pitfall:
Knowledge of typical and atypical imaging findings of WE, particularly when the clinical presentation is non-specific, is mandatory since timely administration of thiamine may halt brain damage, thus preventing the development of Korsakoff Syndrome

Box 11
Idiopathic Normal Pressure Hydrocephalus-iNPH ┼ ┼ NPH syndrome can be primary (idiopathic NPH), or secondary to some condition that impairs CSF absorption (e.g., following infectious, inflammatory, or hemorrhagic events involving the subarachnoid space) ¶ Evans's index is the ratio of the largest width of the frontal horns and the widest measure of the inner table of the skull at that level. When this ratio is greater than 0.3, the ventricles are considered enlarged [54] Δ The CA helps to predict the effect of shunt intervention and is useful in differentiating iNPH from AD with a sensitivity of 97%, a specificity of 88%, and a positive predictive value of 93% at a cutoff value of 90° [31]. Notably, compared to AD, atrophy of the hippocampus is mild in iNPH ◊ The term DESH (included in the Japanese guideline for the diagnosis and treatment of NPH) describes prognostic MRI features in NPH, including a "tight high convexity" and enlargement of CSF spaces in the Sylvian fissure. DESH correlates with a good response to shunting § Notably, symptoms due to stenosis of the cerebral aqueduct may manifest in adulthood, in form of syndrome long-standing overt ventriculomegaly in adults (LOVA), which has a clinical presentation like NPH. Secondary NPH should be suspected in a patient with large head size and MRI disclosure of triventriculomegaly without the involvement of the fourth ventricle, mild or no T2/FLAIR signal change around the ventricular system, and evidence of aqueductal stenosis and/or webbing identified with sagittal fast imaging employing steady-state acquisition c (fiesta-c) sequences ¥ Possible iNPH is diagnosed if the following criteria are met: 1. More than one symptom in the clinical triad: gait disturbance, cognitive impairment, and urinary incontinence 2. Above-mentioned clinical symptoms cannot be completely explained by other neurological or non-neurological diseases 3. Preceding diseases possibly causing ventricular dilation (including subarachnoid hemorrhage, meningitis, head injury, congenital/developmental hydrocephalus, and aqueductal stenosis) are not obvious. Instead, only the "shunt responder" can be meet the diagnosis of definite iNPH [45] Generals -Insidious type of communicating hydrocephalus without substantially increasing CSF pressure -Rare and treatable disease ("reversible dementia") -A correct diagnosis of NPH allows the selection of patients who will respond to shunt surgery

Sample Case Report (in Supplemental material) (E) Future avenues
In the future, the use of volumetric information in routine radiology may be increasingly widespread, and we recommend dual assessment (combining visual scoring with volumetry, see C 3). These measurements are reproducible and automatic, but are depending on scan protocol, software, and the reference population. Other critical issues include the limited access to volumetric tools in the clinical setting (data must be transferred to the workstation and results to the PACS), and the training required to properly read the results.
Differences between men and women in neuroimaging biomarkers of neurodegeneration are reported [7,75] and these should be considered in the near future when   normative reference values will be applied in a clinical setting to assess pathology in individual patients. The new Alzheimer drugs (i.e., Aducanumab) are rapidly changing the clinical scenario and the role of MRI, leading to the need for specific MRI protocols and precise reporting of the side effects of ARIA (amyloid-related imaging abnormalities, referring to cerebral edema or microhemorrhages).
One of the next frontiers is the clinical application of artificial intelligence, as it can offer solutions and interpretation of complex, multimodal medical information, such as that provided by imaging (radiology and nuclear medicine), biology, and neurocognitive testing, thus improving the diagnostic and prognostic process. But the process of identifying international medico-legal rules is still at an early stage [51]. Table 4 Visual rating scale MTA-Scheltens Score based on the visual evaluation of the choroid fissures and temporal horns and hippocampi heights on a coronal T1 weighted image. The coronal section should be: (1) perpendicular to the long axis of both hippocampi, (2) symmetric, (3) placed at the level of the ventral pons or at the ponto-medullary sulcus. Each side should be rated and, in case of asymmetry, should be reported separately. Amygdala atrophy is part of this score and is performed by placing the previously described coronal section (1)   Punctiform or early confluent white matter lesions (Fazekas score 0-1) in periventricular or subcortical distribution is generally normal in aging Fazekas score 2 can be considered normal in subjects of more than 70 years old Confluent lesions (Fazekas score 3) indicate a risk of cognitive decline and physical impairment [69]  Lesions are counted for the left and right hemispheres separately in these brain areas: frontal, parietooccipital, temporal, infratentorial/cerebellum, and basal ganglia (striatum, globus pallidus, thalamus, internal/external capsule, and insula). For each of these regions, therefore, the sum score of the left and right hemispheres is from 0 to 6

Discussion
To complement what was presented in the results, the main practical recommendation that emerged is to try to fit all the radiological steps presented-MRI protocol (B), image evaluation and interpretation (C), and reporting (D)-to the clinical diagnosis. Unfortunately, in radiological practice, there are still several general obstacles to this [74], such as: (1) reduced confidence about the most correct approach to reading images (especially in the use of scales and volumetry), (2) report variability (with no use of structured or guided reports), and (3) generic requisition forms that do not  Three-dimensional brain rendering showing an example of quantitative analysis. Yellow and pink colors indicate the brain areas which are respectively below the 5 and 25 percentiles of the reference population (measures normalized to the intracranial volume). Powered by QyScore® allow radiologists to conclude whether imaging results are in line with clinical suspicion. The latter problem could be solved by better communication among specialists (e.g., at interdisciplinary meetings), which is essential in challenging clinical settings.
The use of all proposed scores is highly recommended, possibly accompanied by a visual description, except for MARS-which in routine is best replaced by a description of the number and distribution of microbleeds, according to the major patterns (superficial distribution in cerebral amyloid angiopathy versus deep infratentorial localization in hypertension)-and for the SVD score-which can be replaced by a description of the findings of small vessel disease according to the priority of their clinical relevance.
Although MRI findings are diagnostic only for a few conditions (e.g., late-onset AD, vascular dementia, CAA, iNPH, etc.), they support the clinical diagnosis of all forms of dementia (see Boxes, above) and provide important information on differential diagnosis, overlapping/coexisting forms (e.g., AD and VaD; FTLD and VaD; DLB and VaD), and possible side effects of new drugs.
More generally, neuroimaging is crucial for the diagnosis of dementia and is recommended in every patient with cognitive decline. In older adults, especially in the oldest old or in patients with multiple comorbidities, severe disability or behavioral disorders, completion of an MRI or nuclear imaging protocol can be troublesome, due to limited collaboration. The indications for the examination should be discussed with the treating physicians, ideally in a multidisciplinary team. Limited to these cases, volumetric CT is acceptable [2], at least to rule out some secondary and potentially reversible causes of cognitive impairment, such as subdural hematoma or brain masses.

Conclusions
The diagnostic process of cognitive disorders requires a combined assessment of the clinical picture and imaging, including CT, MRI, and nuclear medicine, and can only be achieved through the dialogue between disciplines and Conclusions/Impressions -take age (and symptoms) into account and consider cross secƟonal and longitudial variability of the radiological findings Normal, vascular, neurodegeneraƟve, mixed pathology Suggest DD for neurodegeneraƟon the ongoing review of shared knowledge, information, and reports.