The scene has become familiar in memory clinics. A 68-year-old woman explains that she loses first names, repeats a question she has already asked, and catches herself counting appointments twice. She is not “lost.” Not yet. What she feels most clearly is a distinct worry: that of slowly slipping toward something irreversible.
The MRI arrives a few days later. The report mentions “moderate atrophy,” “compatible with age,” or “to be correlated with the clinical context.” These phrases are cautious, and sometimes necessary. But they have one flaw: they leave untouched the most painful part of the diagnosis, the area of uncertainty. Moderate compared with what? At what age? At what pace? And above all: is it progressing?
For a long time, brain imaging was an art of exclusion. One looked for a tumor, a hematoma, hydrocephalus, a stroke. Neurodegeneration, by contrast, does not present itself as a mass, but as an erasure. It is not an “object” that can be outlined with a single stroke. It is a slow, diffuse, sometimes uneven transformation, inscribed in time.
That is where a shift is now taking place: moving from a largely qualitative reading of the brain – shape, overall impression, obvious signs – to a quantified, compared, monitored, almost “biological” reading. Not to replace the clinician’s eye, but to provide it with a measurement, a reference point, and sometimes a shared language.
This quiet change could weigh heavily in the years to come. Because neurodegenerative diseases are no longer a marginal question in neurology. They are becoming a social fact.
A slow epidemic whose scale society is only just beginning to grasp
Dementia is not a single disease, but a syndrome: a set of symptoms (memory, language, judgment, autonomy) caused by several possible pathologies. The World Health Organization estimated that in 2021, 57 million people worldwide were living with dementia, with nearly 10 million new cases each year. Alzheimer’s accounts for the largest share, often estimated at 60-70% of cases.
In Europe, projections follow the same slope. Alzheimer Europe estimates that by 2025, more than nine million people will be living with dementia in the EU-27 countries. And in the United States, the Alzheimer’s Association, in its annual report, estimates that in 2025, 7.2 million Americans aged 65 and over are living with Alzheimer’s dementia.
Behind those figures are lives that grow smaller, families that reorganize themselves, healthcare systems absorbing duration, dependence, and human cost. There is also a harsher truth: neurodegeneration often advances for a long time before it can be “seen” in everyday life. By the time symptoms become undeniable, the pathological process may already have been in place for years.
So the challenge becomes this: detect earlier, understand better, monitor more precisely. Imaging, long confined to “taking pictures,” is now being asked to illuminate an evolving story: what is changing, where, how fast, and with what clinical meaning?
The problem with a brain that is “normal for age”: everyone is right, and no one is helped
In neuroimaging, caution is a virtue. Normal aging comes with a certain amount of brain volume loss. Sulci widen, ventricles may appear somewhat larger. To the naked eye, the boundary between normal and pathological is not always clear-cut, especially at the beginning.
The risk, then, is twofold.
On one side, the radiologist may over-interpret an ordinary variation and cause unnecessary concern. On the other, the radiologist may under-interpret the first marks of a neurodegenerative process and delay care. In both cases, a phrase such as “moderate” or “mild” acts like a half-open door: it suggests something without allowing it to be grasped.
To correct this, medicine has long equipped itself with visual scales: semi-standardized ways of describing what the eye sees, with grades. The best known in suspected Alzheimer’s disease is the medial temporal atrophy scale (MTA), popularized from Scheltens’ work, which relies on the appearance of the hippocampus and neighboring structures. These scales are clinically useful and can be robust, but they still depend on training, context, and ultimately on a reader’s eye.
Studies of the MTA scale show that it can be a valid and useful marker in practice, including in heterogeneous populations, but they also underline the issue of reproducibility and inter-rater variability. In other words: the brain has not changed, but two readers may assign it slightly different grades. And when one is trying to detect subtle changes over several years, those differences become more important than one might imagine.
Neurodegeneration is a matter of millimeters and time. That is what sometimes makes the “all-visual” approach so cruel: it is not that clinicians see nothing, but that what they see does not always suffice to answer the most human questions: “is it progressing?”, “how quickly?”, “what does this mean for me?”
Quantification: giving a number
To quantify a brain is not to slap a grade on it. It is to measure volumes, thicknesses, ratios, and then compare them with references.
In its most widespread form, quantification relies on MRI volumetry: a 3D anatomical sequence (often T1) is used, then software automatically segments different structures (hippocampus, lobes, ventricles, and so on), calculates their volume, and can compare them with a normative database adjusted for age and sometimes sex. One then obtains values that are no longer merely “small” or “moderate,” but situated: percentile, deviation from the norm, trajectory over time.
A recent review of brain volumetry in neurology insists on this central idea: the clinical interest of volumetry is to objectify volume changes associated with pathologies, rather than relying only on visual impression.
At this stage, one nuance is necessary: a number is not an absolute truth. It depends on acquisition quality, the segmentation model, alignment, technical “noise.” But it introduces a new possibility: following evolution with a sensitivity that the eye alone does not always possess.
And that is where the subject becomes concrete. Because neurodegeneration is not a mere “image” to interpret. It is a process to monitor.
Why this measurement is already changing care today
There are several ways in which quantification is already transforming the care pathway – not in theory, but in what patients actually experience.
The first has to do with diagnostic clarification.
In clinical practice, the great challenge is not merely to say “there is dementia.” It is to understand which kind. Alzheimer’s, frontotemporal dementias, Lewy body dementia, vascular damage, mixed disorders: the pictures overlap, the symptoms resemble one another, the trajectories differ. MRI provides topographical clues: more temporal, more frontal, more posterior, more subcortical. But at the early stage, those clues can be subtle.
Quantification does not replace clinical reasoning, but it helps one move beyond vagueness. A hippocampus at the 2nd percentile for age does not tell the whole story, but it gives a clearer signal than a “slight thinning.” And above all, it makes comparison over time possible.
The second has to do with prognosis: who will decline, and how quickly?
A study published in 2024 on “clinically feasible” automated volumetry suggests that regional measurements (particularly hippocampal ones) and total brain volume may help predict later cognitive decline in patients presenting early symptoms.
Clinicians are interested in this kind of result for a simple reason: in mild cognitive impairment, not everyone progresses at the same rate. Some remain stable for a long time. Others decline rapidly. Families, for their part, experience this uncertainty as a suspension: should the home be adapted? should help be anticipated? should closer follow-up be arranged? Quantification does not assign a destiny, but it can help stratify risk more effectively.
The third has to do with follow-up: measuring progression instead of guessing at it.
A 2025 study compared visual assessment of atrophy progression and automated volumetry on successive examinations in “real life.” It explores this very question: is what the eye perceives as progression reflected by quantitative measurements, and with what robustness in non-ideal cohorts? The very fact that this question is being studied under realistic conditions shows the key point: the challenge is no longer to prove that volumetry works in the laboratory, but to make it reliable and useful in actual practice.
In short, quantification does not “replace” conventional MRI. It extends its meaning: instead of being an isolated photograph, it becomes a measurement that can be compared, tracked, and discussed.
The “therapies” effect: when imaging becomes a condition of treatment
There is a very current reason why quantitative imaging is becoming more important: the gradual arrival of treatments that require, in order to be prescribed and monitored, a more rigorous characterization of the disease.
In the United States, for example, the anti-amyloid lecanemab (Leqembi) has prescribing information that recommends a baseline MRI and MRI follow-up, notably because of the risk of ARIA – radiological abnormalities related to amyloid deposition (edema, microhemorrhages) that can be asymptomatic or severe. In August 2025, the FDA even recommended earlier MRI monitoring for certain patients, in order to identify ARIA-E sooner.
The same logic applies to donanemab (Kisunla): its label emphasizes MRI surveillance and specifies an MRI schedule before certain infusions.
One can debate the clinical magnitude of these drugs, their accessibility, their benefits, and their risks. But one thing is already certain: they are changing the place of imaging in the care pathway. MRI is no longer only a tool for differential diagnosis. It also becomes a tool for safety and therapeutic monitoring.
And when medicine enters an era in which treatment depends on a precise and repeated assessment, the demand for quantification increases almost mechanically. Because one cannot base a major decision on vague adjectives. One wants thresholds, trajectories, proof of stability or change. In a way, the brain is joining what the heart has known for a long time: the transition from an impression (“it looks tight”) to a measurement (ejection fraction, gradient, and so on).
The brain is a landscape: quantification helps map it
Talking about volumetry can seem abstract. But for a patient, the most concrete effect is often a change in the conversation.
When a neurologist or geriatrician shows an examination and says, “I find this a little more marked,” that is caution speaking – but it leaves the family alone with its own interpretations. When the physician can say, “this structure is clearly smaller than what we would expect at this age, and it has shrunk by this much in two years,” the conversation changes in density. It becomes harder, sometimes. But it also becomes clearer.
And that clarity can help avoid two common pitfalls.
The first is diagnostic wandering. In some situations, patients move from consultation to consultation because no sign is “clear” enough. Quantification can speed up the identification of a profile compatible with a given pathology, and thus allow more appropriate follow-up, more targeted advice, and a more structured management plan.
The second is inequality of access to expertise. A university memory center has an experienced neuroradiologist, familiar with subtle patterns. A less well-resourced region does not always have that luxury. Quantification, when it is robust and correctly interpreted, can help reduce that disparity by giving less specialized teams an objective reference point. Some publications on automated volumetry in fact emphasize its potential usefulness for identifying at-risk patients “without advanced neuroradiological expertise,” even if it never exempts anyone from clinical interpretation.
This point is delicate: no one wants automated medicine. But many people want less unequal medicine. And measurement, when properly framed, can be a tool for equity.
What the numbers do not say, and what they risk making us believe
Quantification has persuasive power. A number gives the impression of solid truth. Yet in imaging, truth is often a compromise.
First, measurements are sensitive to technique. Two MRIs performed on different machines, with different protocols, can yield slightly different volumes. Software tries to correct for this, but the risk of “false progression” exists if non-harmonized examinations are compared.
Next, normative databases raise a simple question: who is “the norm”? If a reference database is built predominantly on certain populations, it may be less relevant for others. This issue goes beyond technique: it concerns equity and potential bias in the construction of the tools themselves.
Finally, neurodegeneration is not only a matter of volume. Some diseases play out in microstructure (white matter), others in connectivity, others in protein deposits visible on PET (amyloid, tau), still others in functional networks. Volumes are a powerful indicator, but a partial one. A practical review of imaging biomarkers in Alzheimer’s also reminds us that the role and usefulness of different biomarkers vary depending on the context, and that translation into clinical practice must remain cautious.
The question, then, is not to replace judgment with numbers. It is to avoid the false alternative between “the eye” and “the machine.”
The best quantitative reading is often an augmented reading: a radiologist who sees the pattern, a neurologist who knows the patient, and a measurement that helps objectify, compare, and monitor.
Toward a medicine of the long term: to quantify is also to accept duration
Neurodegeneration is a medicine of the long term. It forces one to think in years, sometimes decades. In this field, the greatest enemy is not only the disease: it is the drift of attention. When symptoms are mild, when life goes on, when uncertainty settles in, we postpone, minimize, get used to it.
Quantification, paradoxically, can act as a reminder. Not an anxiety-producing reminder, but a structural one: “this is where we are,” “this is what is changing,” “this is what remains stable.” It can help decide when to intensify follow-up, when to propose additional assessments, when to refer to a specialist consultation, when to discuss driving, work, or family organization.
In an age obsessed with quick solutions, this approach has something almost countercultural about it: it makes a slow transformation visible. It forces us to look directly at a trajectory instead of guessing at it.
The deeper question: what are we measuring, and why?
Every measuring tool carries an implicit philosophy. To measure atrophy is to affirm that structure matters. To measure the hippocampus is to affirm that memory has a geography. To measure progression is to affirm that time is a biomarker.
But medicine does not measure only in order to know. It measures in order to act. For a long time, early detection of Alzheimer’s was criticized for generating anxiety without offering a solution. Things are changing slowly: because prevention is improving, because risk is better understood, because treatments are evolving, because support is being organized, because interventions on risk factors can delay or reduce the probability of decline in some people. The debates remain lively, but the movement is there.
In this context, quantitative imaging occupies a particular place: it does not promise a cure, it promises lucidity. It makes it possible to speak earlier about what is happening, and therefore to organize earlier what can be organized: follow-up, support, adaptations, prevention, research, clinical trials, care pathways.
It becomes one piece of a larger puzzle: that of a medicine of the brain that stops functioning on impression alone and enters into measurement.
A quiet revolution, but not an automatic one
It would be tempting to conclude that quantification will quickly impose itself everywhere. Reality is rougher.
For it to be useful, standardized acquisitions are needed, as are quality-control practices, training, the ability to explain results to patients, and correct integration into care pathways. It also requires accepting that measurement will never be perfect, and that it must always be discussed and contextualized.
But despite these limits, the direction is clear. In a field in which the eye stood alone for so long before the brain’s slow erasure, the arrival of comparable, monitorable measurements marks a change in nature. A passage from description to trajectory.
This is not spectacular progress. It is not the kind of AI demonstration that makes the rounds on social media. It is a rarer kind of progress: the kind that is not immediately visible, but that can save time, bring clarity, and sometimes offer a little peace.
In memory clinics, one often hears the sentence: “I just want to know.” Behind it lies fear, of course. But there is also a demand for landmarks. Quantification will not answer everything. It will not abolish anxiety. It will not replace words, listening, or the dignity of care. What it does is something simpler, and perhaps more precious: it turns an impression into information. And in the medicine of neurodegenerative diseases, that is already a tremendous change.
