MRI can estimate the age of a stroke with reasonable accuracy, and it does so by reading a sequence of signal changes that unfold in brain tissue over hours, days, and weeks. No single MRI finding gives an exact timestamp, but specific patterns on different MRI sequences allow radiologists to place a stroke into broad time windows. In the first several hours, the technique hinges on comparing two types of MRI images to detect a mismatch that fades as time passes. Over days and weeks, other signal changes take over as markers. The whole system is imperfect and influenced by individual anatomy, but it is clinically useful enough to guide treatment decisions for patients who wake up with stroke symptoms and have no idea when the stroke actually began.
The DWI-FLAIR Mismatch in the First Hours
The most clinically important MRI technique for timing an early stroke relies on comparing two types of brain images taken during the same scan session. One is called diffusion-weighted imaging (DWI), which detects water movement in tissue and lights up within minutes of a stroke. The other is called FLAIR, which is slower to show changes. In a fresh stroke, you see a bright spot on DWI but nothing abnormal on FLAIR. That gap between the two is the “DWI-FLAIR mismatch,” and it has become a widely used marker suggesting the stroke is still in its early hours.
A large observational study across multiple centers found that DWI-FLAIR mismatch identified patients who were still within four and a half hours of symptom onset with about 62% sensitivity and 78% specificity, along with a positive predictive value of 83%.1The Lancet Neurology. MRI-based selection for thrombolysis in stroke with unknown onset time (STIR and VISTA Imaging): a multicentre observational study In other words, when the mismatch was present, the stroke was genuinely recent about four out of five times. The mismatch is not perfect, though. It catches only about six in ten early strokes, meaning some recent strokes already show FLAIR changes and get miscategorized as older.
Another study found that FLAIR changes appeared faster in patients with larger areas of damaged brain tissue, which makes sense because bigger strokes cause more swelling and fluid shifts that FLAIR picks up sooner.2PubMed. Fluid-attenuated inversion recovery evolution within 12 hours from stroke onset: a reliable tissue clock? Researchers also found that the brightness of the FLAIR signal, measured as a ratio, correlated positively with time from symptom onset and could distinguish strokes imaged within three hours with roughly 90% sensitivity and 93% specificity when a specific cutoff was used.3PubMed. MR imaging helps predict time from symptom onset in patients with acute stroke: implications for patients with unknown onset time So rather than a simple yes-or-no mismatch, the degree of FLAIR change carries timing information too.
How the Mismatch Disappears Over Time
The DWI-FLAIR mismatch is not a light switch. It fades gradually, and the rate of fading helps refine the estimate. In a study comparing people who woke up with stroke symptoms (and therefore had unknown onset times) to people whose stroke onset was directly witnessed, the mismatch was present at similar rates when MRI was done within two hours of detection. But by two to three hours, the rates diverged sharply. Among wake-up stroke patients scanned at the two-to-three-hour mark, only about 15% still showed the mismatch, versus 60% of witnessed-onset patients scanned at the same interval after symptom start.4PubMed Central. Use of DWI-FLAIR Mismatch to Estimate the Onset Time in Wake-Up Strokes The implication is that many wake-up strokes have actually been going on for hours before the patient realizes it, and the mismatch has already closed. For patients scanned very soon after waking, though, a persistent mismatch suggests the stroke may have occurred close to the time of waking, making them potential candidates for clot-dissolving treatment.
Beyond the First Day and Into the First Weeks
Once you move past the first several hours, different MRI measurements become more useful. One of the most studied is the apparent diffusion coefficient (ADC), a number derived from diffusion-weighted images that reflects how freely water molecules move through brain tissue. In the first days after a stroke, the ADC drops because cells swell and restrict water movement. Over the first week, it gradually climbs back up. During the second week it reaches roughly normal values, and after that it actually rises above normal as dead tissue breaks down and fluid fills the space.5PubMed Central. Evolution of apparent diffusion coefficient, diffusion-weighted, and T2-weighted signal intensity of acute stroke A low ADC value could identify a stroke as less than ten days old with about 88% sensitivity and 90% specificity.
The brightness on DWI images themselves stays elevated throughout the first two weeks and beyond, so DWI alone is not great at distinguishing a two-day-old stroke from a ten-day-old one. The ADC value underneath is what carries the timing information in that window. After two weeks, the DWI brightness starts to fade while the ADC keeps climbing, creating a new kind of divergence that signals a later phase of injury.
When Contrast Enhancement Tells the Story
Giving a contrast agent (gadolinium) during an MRI adds another time-dependent marker. In the first few days after an ischemic stroke, most strokes do not enhance with a standard contrast injection. Enhancement becomes increasingly common over the first week, and by the end of that week, nearly all strokes show some enhancement.6PubMed Central. Evolution of MR contrast enhancement patterns during the first week after acute ischemic stroke An older study found that during the first week, most infarcts did not enhance after a standard bolus injection of contrast, but they did enhance when contrast was given as a continuous infusion, suggesting the blood-brain barrier was already leaking but not enough for a quick bolus to show it.7PubMed. MRI of acute cerebral infarcts, increased contrast enhancement with continuous infusion of gadolinium In the weeks following, enhancement became obvious with standard dosing and typically followed a gyriform (ribbon-like) pattern along the brain’s surface folds.8PubMed. Magnetic resonance imaging of cerebral infarction: time course of Gd-DTPA enhancement and CT comparison
Enhancement that is already present when you first image the patient suggests the stroke is at least a few days old. No enhancement at all, combined with a low ADC, points toward a very recent event. Enhancement that is starting to fade hints at a stroke that is several weeks to months old. Together, these patterns create overlapping time markers that radiologists combine to triangulate a rough age.
Cortical Laminar Necrosis as a Late Marker
For strokes that are weeks to months old, a distinctive pattern called cortical laminar necrosis appears on MRI. This shows up as a bright line along the surface of the brain on certain MRI sequences, representing a specific layer of dead brain cells. Serial MRI studies have shown that these bright cortical lesions begin to appear about two weeks after the stroke, become most prominent at one to two months, and then gradually fade.9PubMed. Cortical laminar necrosis in brain infarcts: chronological changes on MRI In some cases, the signal can persist for over a year.10PubMed. Cortical laminar necrosis in brain infarcts: serial MRI When a radiologist sees this bright cortical ribbon on a scan, it immediately places the stroke in the subacute-to-chronic range rather than the first few days.
Dating a Hemorrhagic Stroke
Everything discussed so far applies mainly to ischemic strokes, which are caused by blocked blood vessels. Hemorrhagic strokes, caused by bleeding into the brain, have their own MRI timeline, and it is arguably more visually dramatic. As a blood clot in the brain ages, the hemoglobin inside the red blood cells goes through a predictable chemical transformation. Each stage produces different magnetic properties that change the MRI signal in characteristic ways.
Five distinct stages of hemorrhage have been described based on these chemical changes:11PubMed. MR appearance of hemorrhage in the brain
- Hyperacute: fresh bleeding with oxyhemoglobin inside intact red cells, appearing bright on most sequences.
- Acute: deoxyhemoglobin forms inside intact red cells within hours, causing the blood to turn very dark on certain sequences.
- Early subacute: methemoglobin forms inside intact red cells over the next few days, producing a characteristic bright signal on one sequence and dark on another.
- Late subacute: the red cells break open and release methemoglobin, which now appears bright on most sequences.
- Chronic: hemoglobin breaks down into iron storage products (ferritin and hemosiderin), leaving a permanent dark rim around the old bleed.
A specialized MRI sequence called susceptibility-weighted imaging (SWI) is particularly sensitive to these iron-containing breakdown products. In the hyperacute phase, a hemorrhage can appear bright on SWI because there is not yet enough paramagnetic material to darken it. In the acute and early subacute phases, the signal goes very dark. In the chronic phase, the center of the old bleed may have mixed signal, but the rim stays dark because iron deposits remain permanently.12PubMed Central. Susceptibility-weighted imaging in intracranial hemorrhage: not all bleeds are black SWI in ICH: appearances, pitfalls and mimickers SWI is also more sensitive than older techniques at detecting tiny old bleeds called microbleeds, which are essentially chronic-stage remnants of past small hemorrhages.13American Journal of Neuroradiology. MR Imaging Detection of Cerebral Microbleeds: Effect of Susceptibility-Weighted Imaging, Section Thickness, and Field Strength
Why Getting the Timing Right Changes Treatment
The practical reason all of this matters comes down to clot-dissolving drugs. Standard guidelines allow intravenous thrombolysis (the clot-busting drug alteplase) within four and a half hours of stroke onset. But roughly one in five ischemic stroke patients wake up with symptoms or cannot say when they started, and those patients used to be automatically excluded from treatment because nobody could confirm the time window.
MRI changed that. The landmark WAKE-UP trial enrolled patients who had stroke symptoms of unknown onset and used DWI-FLAIR mismatch to select those whose strokes appeared recent enough for treatment. Patients with the mismatch who received alteplase had better outcomes at 90 days compared to those given placebo: about 53% achieved a favorable outcome versus roughly 42% in the placebo group.14PubMed. MRI-Guided Thrombolysis for Stroke with Unknown Time of Onset Rates of symptomatic brain bleeding were low in both groups. A later meta-analysis combining individual patient data from multiple trials confirmed that patients treated up to nine hours after onset, or after waking with symptoms, had better functional outcomes when selected using advanced imaging, including MRI-based perfusion methods.15The Lancet. Thrombolysis guided by perfusion-diffusion MRI or CT perfusion up to 9 hours after onset of ischaemic stroke: a systematic review and meta-analysis of individual patient data MRI-based timing has effectively expanded the treatment window for a group of patients who previously had no options.
What Throws Off the MRI Clock
Several factors can make a stroke look older or younger than it actually is on MRI, and radiologists have to account for these when interpreting scans.
Collateral blood flow is one of the biggest variables. When a major brain artery is blocked, nearby smaller vessels sometimes compensate by rerouting blood around the blockage. Patients with good collateral circulation can maintain partial blood flow to the affected area, which slows the development of FLAIR signal changes and makes the stroke appear more recent than it is. One study found a clear association between time from symptom onset and FLAIR brightness in patients with poor collaterals, but virtually no association in patients with good collateral flow.16PubMed Central. Association between time from stroke onset and FLAIR lesion intensity is modified by status of collateral circulation In effect, good collaterals slow down the tissue clock.
Treatment itself also alters the MRI timeline. After clot retrieval (thrombectomy), patients who achieve good blood flow restoration show different signal evolution than those who do not. Patients with incomplete reperfusion after clot retrieval had larger lesion growth and more swelling by 24 hours compared to those with complete reperfusion.17PubMed Central. Early Post-Thrombectomy MRI Markers: Temporal Evolution and Association With Reperfusion and Clinical Outcome In patients with near-complete reperfusion, the FLAIR signal in the initial stroke area stayed essentially stable between the post-procedure scan and day five. In patients with poor reperfusion, the signal continued to increase, meaning the stroke “looked” like it was progressing.18PubMed. Evolution of Volume and Signal Intensity on Fluid-attenuated Inversion Recovery MR Images after Endovascular Stroke Therapy So when reading a follow-up MRI after treatment, a radiologist needs to know whether the artery was reopened and how completely, because that context changes what the signals mean about timing and prognosis.
Stroke size also matters. Larger strokes tend to show FLAIR changes faster and have lower ADC values at any given time point, which can make them look older. Small strokes in deep brain structures may barely show FLAIR changes at all for several hours, making them look deceptively fresh.
MRI Perfusion and the Concept of Salvageable Tissue
Beyond estimating when a stroke happened, MRI can also estimate how much brain tissue is still at risk but not yet dead. Perfusion MRI measures blood flow through the brain, and when the area of reduced blood flow is much larger than the area that has already died (seen on DWI), the difference represents tissue that could still be saved if blood flow is restored quickly. This “diffusion-perfusion mismatch” is conceptually different from the DWI-FLAIR mismatch discussed earlier. It does not directly tell you when the stroke started, but it tells you whether there is still a treatment opportunity, which is sometimes more useful than knowing the exact onset time.
A newer perfusion technique called arterial spin labeling uses magnetically tagged blood as a natural tracer instead of injected contrast dye. Research has found that the penumbra volumes measured by this technique show excellent agreement with those from CT-based perfusion imaging, suggesting it could serve as a contrast-free alternative.19PubMed. Multidelay Arterial Spin Labeling Versus Computed Tomography Perfusion in Penumbra Volume of Acute Ischemic Stroke The advantage is that MRI perfusion can be done at the same time as the DWI and FLAIR sequences, giving a comprehensive picture of both stroke age and treatment potential in a single scan session.20Egyptian Journal of Radiology and Nuclear Medicine. Role of arterial spin labeling magnetic resonance perfusion in acute ischemic stroke
Where AI Fits In
Humans can read MRI patterns and estimate stroke age, but the process is subjective and varies between radiologists. Researchers have been training deep learning models to do the same thing, feeding them raw DWI and FLAIR images and asking the algorithm to predict whether a stroke is recent enough for treatment. Early work has focused on the same binary question that the DWI-FLAIR mismatch answers: is this stroke within the treatment window or not?21NeuroImage: Clinical. A deep learning analysis of stroke onset time prediction and comparison to DWI-FLAIR mismatch The promise is that an algorithm might pick up on subtle signal differences that the human eye misses, potentially improving both sensitivity and specificity beyond what the mismatch alone achieves. This is still an area of active research rather than routine clinical practice, but it represents the direction the field is moving.
Why MRI Still Loses to CT in Many Emergency Rooms
Given how much timing information MRI provides, you might wonder why it is not the default scan for every stroke patient. The answer is logistics. A CT scan takes a few minutes and is available around the clock in virtually every emergency department. MRI takes longer, requires a patient to lie still in a confined space, and is not always available on short notice. For the initial decision about whether a stroke is hemorrhagic (which changes the treatment plan entirely), CT is fast and definitive. MRI’s timing advantages become most relevant for the subset of patients whose stroke onset is unknown, and for those patients, many stroke centers do have protocols that include urgent MRI.
There are also contraindications to MRI. Patients with certain implanted devices, severe claustrophobia, or hemodynamic instability may not be able to get into the scanner. And the signal changes that MRI reads as a tissue clock can be mimicked by other conditions. Certain infections, tumors, and inflammatory diseases can cause restricted diffusion on DWI that resembles a stroke, potentially confusing the picture. In practice, experienced neuroradiologists combine the MRI findings with the patient’s clinical presentation, lab work, and vascular imaging to avoid these pitfalls. The MRI is a powerful piece of the puzzle, but it does not provide the answer by itself.
Microbleeds and the Archaeology of Past Strokes
One of the more striking things MRI can do is detect evidence of strokes that happened long ago, sometimes ones the patient never knew about. Cerebral microbleeds, tiny deposits of iron from old leaks in small blood vessels, appear as small dark dots on SWI sequences. These are essentially permanent scars. Using high-resolution SWI, researchers have been able to track individual microbleeds over time and found that they often increase in size slowly over the years.22PubMed Central. Imaging cerebral microbleeds using susceptibility weighted imaging: one step toward detecting vascular dementia The number and location of microbleeds carry information about the underlying cause: when they cluster in deep brain structures, they point toward damage from high blood pressure; when they are scattered across the outer brain, they suggest a condition called cerebral amyloid angiopathy, where abnormal proteins weaken vessel walls. Neither pattern gives a precise date, but both tell a story about the brain’s vascular history that can inform treatment choices and risk assessment going forward.