An MRI scan can reveal structural brain changes strongly associated with dementia, but it does not produce a result that reads “dementia: yes or no.” What it does is show patterns of brain shrinkage, damage to white matter, enlarged fluid-filled spaces, and other physical changes that, combined with clinical evaluation, help clinicians determine what type of dementia a person has or whether something else entirely is causing cognitive problems. In many cases, the MRI findings are what shift a vague concern about memory loss into a specific, actionable diagnosis.
What an MRI Actually Shows
A standard brain MRI produces detailed images of brain structure: the size and shape of different brain regions, the condition of the white matter that connects them, and the volume of fluid-filled ventricles. None of these features alone says “dementia.” Instead, radiologists and neurologists look for patterns. For Alzheimer’s disease, the hallmark pattern is shrinkage of the medial temporal lobe, particularly the hippocampus, the brain’s memory center. MRI-based measurements of hippocampal volume can distinguish people with Alzheimer’s from healthy older adults, even fairly early in the disease.1PubMed Central. Medial temporal atrophy on MRI in normal aging and very mild Alzheimer’s disease That shrinkage correlates with the severity of Alzheimer’s pathology, though it has not historically been used as a standalone diagnostic criterion.2PubMed Central. Medial temporal lobe atrophy on MRI scans and the diagnosis of Alzheimer disease
But Alzheimer’s is just one form of dementia, and each type leaves its own fingerprint on the MRI. Those differences are often what make an MRI so valuable: the scan helps distinguish between conditions that look similar in a clinic visit but are quite different in the brain.
How Different Dementias Look on the Scan
Vascular dementia, the second most common form, produces a distinct picture. Rather than the focused hippocampal shrinkage of Alzheimer’s, MRIs of people with vascular dementia show extensive bright patches in the white matter called white matter hyperintensities. A meta-analysis found that all patients in the vascular dementia group had these high signals on brain MRI, with greater volumes of affected white matter compared to people with normal cognition, mild impairment, or even Alzheimer’s. Irregular and periventricular patterns of white matter damage were particularly specific to vascular dementia.3PubMed Central. White matter hyperintensities and the risk of vascular dementia: a systematic review and meta-analysis
Frontotemporal dementia flips the script on where shrinkage occurs. Instead of the hippocampus bearing the brunt, the frontal and temporal lobes show severe atrophy. One study found very severe tissue loss in the temporal lobes and severe loss in the frontal lobes, with only mild hippocampal shrinkage by comparison.4PubMed. The MRI pattern of frontal and temporal brain atrophy in fronto-temporal dementia The specific sub-type matters too. Classic frontotemporal dementia involves the dorsolateral frontal cortex, while the semantic variant (which impairs word and concept understanding) targets the anterior temporal cortex and the amygdala region.5PubMed. Patterns of brain atrophy in frontotemporal dementia and semantic dementia
Dementia with Lewy bodies is perhaps the trickiest to identify on MRI because its most distinctive features are not structural. The key MRI clue is actually an absence: relative preservation of the hippocampus and medial temporal lobe compared to Alzheimer’s.6PubMed. Patterns of cerebral atrophy in dementia with Lewy bodies using voxel-based morphometry When both conditions cause hippocampal shrinkage, the pattern of which sub-regions are affected differs. In Alzheimer’s, nearly all hippocampal subfields are significantly atrophied, while Lewy body dementia spares certain sub-regions.7The American Journal of Geriatric Psychiatry. Differential Atrophy of Hippocampal Subfields: A Comparative Study of Dementia with Lewy Bodies and Alzheimer Disease This distinction can help tip a clinician toward one diagnosis over the other, especially in someone with mild cognitive impairment whose symptoms could go either way.8PubMed Central. Hippocampal volumes predict risk of dementia with Lewy bodies in mild cognitive impairment
Ruling Out Treatable Causes
One of the most important jobs of a brain MRI in someone with cognitive symptoms has nothing to do with confirming dementia. It is about making sure something reversible is not being missed. Clinical guidelines recommend structural brain imaging for all patients presenting with dementia or cognitive symptoms, in part because a meaningful percentage of cases turn out to have a treatable cause. The most common reversible conditions flagged this way include brain tumors, normal pressure hydrocephalus, and medication side effects, among others.9PubMed Central. Reversible dementias
Normal pressure hydrocephalus is a good example. This condition causes a triad of symptoms: walking difficulty, urinary incontinence, and cognitive decline. On MRI, it produces a distinctive appearance of disproportionately enlarged ventricles with specific features like a narrowed callosal angle and a pattern known as DESH, where the fluid spaces near the top of the brain are squeezed tight while spaces near the base are widened.10PubMed Central. Application of Evans Index in Normal Pressure Hydrocephalus Patients: A Mini Review Catching this matters because it can be treated surgically with a shunt. Without an MRI, it might be mistaken for Alzheimer’s and left to progress.
Can MRI Detect Dementia Early
This is where the research has become genuinely encouraging. MRI-based measurements of medial temporal structures can detect changes before a person meets the full criteria for dementia, during the stage known as mild cognitive impairment. Automated MRI measures of the medial temporal cortex have been shown to predict who will progress from mild impairment to Alzheimer’s, outperforming even metabolic and cellular biomarkers in head-to-head comparisons.11PubMed Central. Automated MRI measures predict progression to Alzheimer’s disease
Newer research has pushed this further using machine learning. One study built a model combining clinical scores with MRI-derived features and achieved roughly 90% accuracy in predicting which people with mild cognitive impairment would go on to develop Alzheimer’s.12Scientific Reports. A radiomics model predicts progression from mild cognitive impairment to alzheimer’s disease using structural MRI Another study found that the volume of the right entorhinal cortex, a small region near the hippocampus, could correctly classify nearly all of those who would later convert to Alzheimer’s.13Brain Communications. Evaluating conversion from mild cognitive impairment to Alzheimer’s disease with structural MRI: a machine learning study The entorhinal cortex is one of the first areas affected by Alzheimer’s pathology, so its volume acts as an early warning signal.
That said, these are research tools, not routine clinical practice yet. Your doctor is not currently handing you a machine-learning-based risk score from your MRI. But the direction of travel is clear, and it’s why MRI remains central to most dementia workups despite the availability of newer biomarkers.
Beyond Standard MRI
A conventional brain MRI looks at structure: how big is this region, how much fluid is there, are there any obvious lesions. But specialized MRI techniques can also look at how the brain’s wiring is holding up and how blood is flowing through different regions.
Diffusion tensor imaging, or DTI, measures the integrity of white matter tracts, the bundles of nerve fibers that connect distant brain regions. White matter damage is a feature of many dementias, and DTI can detect changes that a standard MRI misses entirely. Studies have found that DTI measures of how water moves along white matter tracts can distinguish Alzheimer’s patients from healthy controls and, importantly, can pick up more subtle changes in people with mild cognitive impairment where standard structural MRI sees nothing unusual.14NeuroImage: Clinical. Effectiveness of regional DTI measures in distinguishing Alzheimer’s disease, MCI, and normal aging The fornix, a key white matter tract connecting the hippocampus to the rest of the brain, seems to be especially sensitive. Reduced integrity there correlates with cognitive performance even in people who do not yet have dementia.15JAMA Neurology. White Matter Integrity Determined With Diffusion Tensor Imaging in Older Adults Without Dementia: Influence of Amyloid Load and Neurodegeneration
Arterial spin labeling, or ASL, is another MRI technique that measures blood flow through the brain without needing a contrast injection. Reduced blood flow in specific regions follows a predictable pattern in Alzheimer’s, with decreases concentrated in the hippocampus, precuneus, and posterior cingulate cortex. A systematic review found that ASL MRI at higher field strengths achieved diagnostic accuracy comparable to PET scanning while being completely non-invasive and radiation-free.16PubMed Central. Arterial Spin Labeling MRI in Alzheimer’s Disease: A Systematic Review of Cerebral Perfusion Biomarkers ASL can also help tell Alzheimer’s and Lewy body dementia apart: Lewy body patients show more blood flow deficits in the frontal and insular cortices, while Alzheimer’s patients show them in the parietal region. In one study, these perfusion patterns classified Lewy body dementia patients with sensitivity as high as 100% and specificity up to 96%, depending on the disease stage.17PubMed Central. Brain perfusion in dementia with Lewy bodies and Alzheimer’s disease: an arterial spin labeling MRI study on prodromal and mild dementia stages
Functional MRI adds yet another layer. By tracking changes in brain activity at rest, it can map the “default mode network,” a set of brain regions that are active when you are daydreaming or not focused on a task. This network is disrupted in Alzheimer’s, and longitudinal research has shown that increasing connectivity changes within it track with cognitive decline over time.18PubMed. Default-Mode Network Connectivity Changes During the Progression Toward Alzheimer’s Dementia: A Longitudinal Functional Magnetic Resonance Imaging Study
Telling Dementia Apart from Depression
One of the most clinically meaningful things an MRI can do is help distinguish dementia from late-life depression, which often mimics it. Older adults with depression frequently complain of memory problems, have difficulty concentrating, and may score poorly on cognitive tests. The overlap is significant enough that depression has historically been called “pseudodementia.” An MRI can provide a crucial piece of the puzzle.
Depressed older adults do show some brain shrinkage compared to healthy controls, including lower overall brain volume and enlarged ventricles. But the pattern is different from Alzheimer’s. In depression, the temporal lobes and the hippocampal complex tend to be relatively spared, while in Alzheimer’s those structures are severely affected.19Journal of Affective Disorders. Quantitative magnetic resonance imaging in geriatric depression and primary degenerative dementia Temporal lobe atrophy ratings on MRI showed strong ability to separate Alzheimer’s from depression, with sensitivity of 83% for detecting Alzheimer’s and specificity of 87% for correctly identifying depressed patients as not having it. The parahippocampal gyrus and entorhinal cortex were especially useful, reaching 97% specificity for depression.20Psychological Medicine. Temporal lobe magnetic resonance imaging can differentiate Alzheimer’s disease from normal ageing, depression, vascular dementia and other causes of cognitive impairment
Adding hippocampal volume to more advanced biomarkers may further sharpen the distinction. One study found that hippocampal volume alone correctly classified about half of participants with geriatric depression or probable Alzheimer’s, but adding amyloid PET data raised that to 87%.21Psychiatry Research: Neuroimaging. An optimized MRI and PET based clinical protocol for improving the differential diagnosis of geriatric depression and Alzheimer’s disease The takeaway for patients: an MRI showing a normal-looking hippocampus in someone with memory complaints should prompt serious consideration that depression, medication effects, or another reversible cause may be responsible.
What MRI Cannot Do on Its Own
For all its usefulness, MRI has real limitations in dementia diagnosis. Normal aging itself causes some degree of brain shrinkage, and the overlap between healthy older brains and early-stage dementia brains is wider than most people realize. A deep learning study found that the brain regions generating “brain age gaps” in Alzheimer’s showed patterns similar to those in normal aging, making the two harder to distinguish than you might hope from imaging alone.22Nature Aging. Deep learning-based brain age prediction in normal aging and dementia
This is why dementia diagnosis increasingly relies on combining MRI with other biomarkers. Blood-based biomarkers for Alzheimer’s, particularly plasma measurements of certain proteins, are becoming more widely available. When combined with MRI-measured hippocampal atrophy, these plasma markers achieved very high specificity (96%) for identifying Alzheimer’s, though sensitivity was low (34%), meaning they caught only a third of true cases.23The Journal of Prevention of Alzheimer’s Disease. Performance of Plasma Biomarkers Combined with Structural MRI to Identify Candidate Participants for Alzheimer’s Disease-Modifying Therapy A separate study of people with subjective cognitive complaints found that combining plasma markers with MRI measurements produced the best predictions of who would actually decline over time, outperforming either type of test used alone.24PubMed Central. Cognitive and neurodegenerative trajectories of subjective cognitive decline according to baseline biomarkers: Results of the CoSCo study
In practice, most clinicians use MRI as one element in a broader evaluation that includes cognitive testing, medical history, blood work, and sometimes PET scans or spinal fluid analysis. The MRI provides structural evidence, but interpreting that evidence requires clinical context.
Incidental Findings and What Not to Panic About
When you get a brain MRI for any reason, there is a reasonable chance the scan will turn up something unexpected and unrelated to your original symptoms. A large population-based study found asymptomatic brain infarcts in about 7% of people, cerebral aneurysms in roughly 2%, and benign tumors (mostly meningiomas) in about 2%.25PubMed. Incidental findings on brain MRI in the general population White matter changes and small areas of dilated fluid spaces around blood vessels are even more common, particularly in older adults and those with high blood pressure.26Scientific Reports. Incidental findings on brain MRI among Chinese at the age of 55–65 years: the Taizhou Imaging Study
The spectrum of incidental findings ranges from things that need immediate attention (like a growing tumor or a large aneurysm) to things that are completely benign and essentially a normal variant of brain anatomy.27PubMed Central. Incidental findings on brain magnetic resonance imaging (MRI) in adults: a review of imaging spectrum, clinical significance, and management If you are getting a brain MRI as part of a dementia workup and the report mentions small vessel disease or white matter changes, that is worth discussing with your doctor but is not, on its own, a dementia diagnosis. Many people have these findings and remain cognitively sharp.
Ultra-High-Field MRI and the Research Frontier
Most clinical MRI scanners operate at 1.5 or 3 Tesla. Research centers are increasingly using 7 Tesla scanners, which produce dramatically sharper images and can reveal structural changes previously visible only at autopsy. A systematic review found that 7T MRI can detect hippocampal subfield atrophy in people with mild cognitive impairment that lower-field scanners miss entirely. It is also more sensitive to tiny microbleeds and iron deposits in the brain, both of which are relevant to Alzheimer’s and other dementias.28Journal of Neurology, Neurosurgery, & Psychiatry. 7T MRI for neurodegenerative dementias in vivo: a systematic review of the literature The resolution at 7T approaches the scale of the pathologies that define these diseases, such as amyloid plaques and changes in specific cortical layers.29PubMed Central. Seven Tesla MRI in Alzheimer’s disease research: State of the art and future directions: A narrative review
These ultra-high-field scanners are not available in routine clinical practice; they are expensive, not widely distributed, and the scans take longer. But the findings from 7T research are shaping what clinicians look for on standard 3T scans and informing the development of more sensitive imaging protocols.
Access and Cost Barriers
Even if MRI is the gold standard for structural brain imaging in dementia workups, getting one is not straightforward everywhere. In low- and middle-income countries, which are projected to experience the greatest increases in dementia prevalence, access to MRI is severely limited.30Nature Reviews Neurology. The potential of low-field MRI for global dementia care In some settings, the few available scanners are concentrated in capital cities, and the cost of a scan is passed directly to patients who typically lack insurance coverage.31PLOS ONE. Perceptions and practices of imaging personnel and physicians regarding the use of brain MRI for dementia diagnosis in Uganda
Low-field MRI systems, which are cheaper to buy, install, and maintain, are emerging as a potential solution. These scanners produce lower-resolution images than their 1.5T or 3T counterparts, but they may still be sufficient to detect the gross structural changes relevant to dementia diagnosis: large ventricles, significant atrophy, tumors, and hydrocephalus. The research community is actively investigating how well low-field systems perform for dementia-specific use cases.32PubMed. Brain MRI for the diagnosis of dementia in low- and middle-income countries: a systematic review and meta-analysis protocol of diagnostic utility and imaging findings If the results hold up, these systems could dramatically expand access to brain imaging in regions where dementia is growing fastest but diagnostic infrastructure is thinnest.
AI-Assisted Reading of Brain Scans
The human eye can spot obvious atrophy and large white matter lesions, but many of the MRI changes relevant to early dementia are subtle. Automated tools are increasingly being developed to measure brain volumes, segment gray and white matter, and flag abnormal regions. A scoping review of automated MRI segmentation found that AI models, particularly deep learning approaches, demonstrated high accuracy in brain age prediction, classification of neurodegenerative diseases, and identification of disease-specific brain patterns.33PubMed. A scoping review of automatic and semi-automatic MRI segmentation in human brain imaging Software packages like FreeSurfer, which automatically parcellates the brain into regions and measures their volumes, are already widely used in research settings and are beginning to appear in clinical workflows.
The practical impact is that a radiologist reading your MRI may eventually have access to an automated report showing exactly how your hippocampal volume compares to age-matched norms, flagging regions that have shrunk more than expected. That kind of quantitative output could make the scan more useful in borderline cases where the naked eye sees “maybe a little shrunken, maybe not.” The gap between research-grade MRI analysis and what your local hospital can offer is narrowing, though it has not closed yet.