A CT scan does not always show a stroke, and this gap is most pronounced in the first several hours after symptoms begin. Standard non-contrast CT is excellent at detecting bleeding in the brain, which makes it indispensable for ruling out hemorrhagic stroke before treatment. But for the more common ischemic type, where a clot blocks blood flow, a plain CT scan can look completely normal even while brain tissue is actively dying. One large prospective comparison found that CT detected acute ischemic stroke in only about 10% of patients, compared with 46% for MRI, when both were performed in an emergency setting. Understanding when and why CT misses strokes matters because millions of people rely on it as the first scan performed in the emergency department.
Why CT Misses Early Ischemic Strokes
The fundamental problem is timing. An ischemic stroke happens when a blood clot cuts off supply to part of the brain. In the first few hours, the affected tissue swells and softens, but these changes are subtle on CT because the density difference between healthy and damaged brain tissue is initially tiny. A study of 232 patients with suspected acute ischemic stroke found that among those who arrived within four hours of symptom onset, CT picked up the stroke in only about 5% of cases, while diffusion-weighted MRI detected it in roughly 86%.1Journal of Health, Wellness and Community Research. Diagnostic Accuracy of Non-Contrast CT in Detection of Acute Ischemic Stroke Taking Diffusion-Weighted Magnetic Resonance Imaging as Gold Standard After four hours, that gap narrowed substantially, with CT sensitivity climbing above 90%. So the scan is not fundamentally flawed; it just takes time for ischemic damage to become visible on CT.
In the ultra-early window, specifically within the first 90 minutes, radiologists themselves often disagree about whether the scan shows anything abnormal. Research has documented that reader agreement on early ischemic changes is lowest in this ultra-early phase.2PubMed. Time dependence of reliability of noncontrast computed tomography in comparison to computed tomography angiography source image in acute ischemic stroke The signs are there in some cases, but they are so faint that even experienced neuroradiologists read them differently.
The Signs Radiologists Look For
When a plain CT does catch an early ischemic stroke, it is usually through a handful of well-known subtle findings. One is the hyperdense artery sign, where the clot itself appears as a bright white vessel, most commonly in the middle cerebral artery. Another is a loss of the normal contrast between gray and white matter in specific regions, such as the insular cortex losing its distinct ribbon-like appearance, or the lentiform nucleus fading into the surrounding tissue.3PubMed. Brain CT scan for acute cerebral infarction: early signs of ischemia Swelling that flattens the brain’s normal grooves is another clue. None of these signs appear in every patient, and all require a trained eye to spot. A brain CT that looks “normal” to you or me might carry one of these whispers for a neuroradiologist, but equally, a genuinely normal scan cannot reliably rule out an ischemic stroke in progress.
Hemorrhagic Stroke Is a Different Story
CT’s reputation in stroke care rests largely on its ability to detect bleeding. Fresh blood in the brain is dramatically denser than surrounding tissue, lighting up as a bright white mass on the scan. This is why CT remains the universal first-line imaging test: the single most urgent question in the emergency department is whether the patient is bleeding, because the treatment for a bleed is the opposite of the treatment for a clot. Giving a clot-busting drug to someone who is actually hemorrhaging could be fatal.
Where CT falls short, even with hemorrhage, is in detecting very small bleeds within areas of ischemic damage. A study comparing CT to advanced MRI sequences found that CT identified hemorrhagic transformation in only five of 38 acute infarct cases, while a specialized MRI technique picked it up in 16.4PubMed. Reliability in detection of hemorrhage in acute stroke by a new three-dimensional gradient recalled echo susceptibility-weighted imaging technique compared to computed tomography: a retrospective study So CT is superb at spotting a large, frank hemorrhage but less reliable at catching micro-hemorrhages or hemorrhagic conversion within an existing infarct.
Lacunar and Small Strokes Often Escape Detection
Not all strokes involve large vessels. Lacunar strokes are caused by disease in tiny arteries deep within the brain, and the resulting areas of damage are often no larger than a centimeter or two. Because of their small size and deep location, these strokes are particularly hard for CT to detect. An analysis of patients from the IST-3 trial found a negative CT scan in 45% of people who actually had a lacunar stroke, consistent with prior estimates that ranged from 35% to 50%.5PubMed Central. Improving Clinical Detection of Acute Lacunar Stroke Analysis From the IST-3 Detection improves with careful clinical correlation, meaning that knowing the patient’s specific symptoms helps the radiologist know exactly where to look, but even then the lesion may simply be too small for CT to resolve.6PubMed. Imaging Review and Clinical Manifestations of Lacunar Stroke Syndromes
Similarly, a multimodal CT study found that small infarctions averaging about 1.5 centimeters were missed by all CT-based methods, including perfusion CT and CT angiography, in a subset of patients.7Radiology. Acute stroke assessment with CT: do we need multimodal evaluation? Tiny strokes can be genuinely invisible on CT regardless of technique.
Posterior Fossa Strokes Present Extra Challenges
Strokes in the brainstem and cerebellum, collectively the posterior fossa, are notoriously difficult to see on CT. The bony skull base surrounding these structures creates artifacts that degrade image quality. Even on MRI, which is far more sensitive, small brainstem infarcts can be blurred or hidden by the same proximity to bone.8PubMed Central. Improved detectability of acute and subacute brainstem infarctions by combining standard axial and thin-sliced sagittal DWI For CT, the problem is worse. A patient with symptoms like vertigo, double vision, or difficulty swallowing may have a posterior circulation stroke that a standard CT simply cannot visualize.
How CT Compares to MRI
The gap between CT and MRI for stroke detection is substantial and well documented. In a prospective head-to-head comparison, MRI had an overall sensitivity of 83% for any acute stroke, while CT managed only 26%.9PubMed Central. Magnetic resonance imaging and computed tomography in emergency assessment of patients with suspected acute stroke: a prospective comparison Even when the scan was performed within three hours of symptom onset, MRI detected acute ischemic stroke in about 46% of patients versus 7% for CT. Other studies have confirmed that diffusion-weighted MRI correctly identifies the acute lesion in virtually all cases, while CT catches it in roughly 42% to 63%, depending on the reader.10PubMed. Comparison of diffusion-weighted MRI and CT in acute stroke
If MRI is so much better, why isn’t it used first? Practical constraints explain most of it. CT scans take a few minutes; MRI takes considerably longer. CT machines are available in virtually every emergency department, often around the clock. MRI requires specialized equipment, takes longer to set up, and many hospitals do not have an MRI scanner available for emergency use at 3 a.m. Patients who are confused, agitated, or on a ventilator may not be able to lie still long enough for MRI. For these reasons, CT remains the gatekeeper, and MRI is used as a follow-up when CT is negative but clinical suspicion is high.11PubMed Central. Comparison of CT and MR imaging in ischemic stroke
Advanced CT Techniques That Improve Detection
A standard non-contrast CT is just one tool in the CT toolkit. When stroke is suspected, many hospitals now perform a multimodal CT protocol that adds two components: CT angiography and CT perfusion. CT angiography maps the blood vessels after injecting contrast dye, revealing where a clot is blocking flow. A meta-analysis of CT angiography’s ability to detect large vessel blockages in the front of the brain found pooled sensitivity around 88% and specificity near 99%.12European Journal of Radiology. Diagnostic performance of single-phase CT angiography in detecting intracranial large vessel occlusion: A systematic review and meta-analysis
CT perfusion takes things a step further by measuring how blood actually flows through brain tissue in real time. It can distinguish brain that is already dead from brain that is at risk but still salvageable, a distinction that directly guides treatment decisions.13PubMed. Imaging of stroke: Part 1, Perfusion CT–overview of imaging technique, interpretation pearls, and common pitfalls Research has shown that optimized CT perfusion thresholds can predict final infarct size with strong agreement compared to MRI reference values.14PubMed. Whole-Brain CT Perfusion to Quantify Acute Ischemic Penumbra and Core So while plain CT alone may miss the stroke, the full multimodal protocol catches substantially more cases, though even the combined approach can miss very small lesions.
When the CT Is Normal but It Really Is a Stroke
Emergency physicians make treatment decisions based on the whole clinical picture, not just the scan. If a patient arrives with one-sided weakness, speech difficulties, and a known time of onset, a normal CT does not mean they are not having a stroke. It often means the stroke has not yet become visible on CT. In fact, treatment guidelines allow clot-busting medication (IV alteplase) to be given when the CT is normal, because the primary role of the initial scan is to exclude hemorrhage rather than confirm ischemia. Research on “wake-up strokes,” where the patient wakes with symptoms and the exact onset time is unknown, has affirmed that treating patients who have a normal brain CT is safe when hemorrhage has been excluded.15PubMed Central. CT-guided thrombolytic treatment of patients with wake-up strokes
This is a point that often confuses patients and families. A “normal” CT in the emergency department is not reassurance that nothing is wrong. It is a green light to treat, because it tells the clinical team that the brain is not bleeding and clot-busting therapy can proceed.
Stroke Mimics Add Another Layer of Complexity
Complicating matters further, not every patient who arrives at the emergency department looking like they are having a stroke is actually having one. A range of other conditions can produce sudden neurological symptoms that closely resemble stroke. One study found that about 22% of patients evaluated with a multimodal CT stroke protocol turned out to have a stroke mimic. The most common mimics were seizures, migraine with aura, and conversion disorder.16American Journal of Neuroradiology. Stroke Mimics in the Acute Setting: Role of Multimodal CT Protocol Other research has reported even higher mimic rates, with one emergency department study identifying mimics in about 40% of patients initially suspected of stroke.17PubMed Central. The Incidence of Stroke Mimics in the Emergency Department of a Tertiary-care Center in Lebanon
When a CT scan is normal in the emergency department, the differential therefore includes both “it’s a real stroke that’s too early to see” and “it’s not a stroke at all.” Diffusion-weighted MRI is the most sensitive tool for sorting this out early, but it is not always available urgently.18PubMed Central. Stroke mimics: incidence, aetiology, clinical features and treatment Clinicians rely on the full multimodal CT protocol, clinical examination, and sometimes follow-up imaging to distinguish true strokes from mimics.
Stroke Detection in Children
Pediatric stroke is rare but often diagnosed late, partly because the symptoms overlap heavily with much more common childhood conditions like migraine and seizures. CT faces the same sensitivity limitations in children as in adults, compounded by the fact that doctors may not be thinking about stroke in a young patient. Neuroimaging plays an important role in pediatric stroke, but clinical suspicion has to come first.19PubMed. Imaging of Suspected Stroke in Children, From the AJR Special Series on Emergency Radiology When a child presents with sudden focal neurological deficits, the same principle applies: a normal CT does not rule out stroke, and MRI is the preferred follow-up when the clinical picture warrants it.
Motion Artifacts and Technical Quality
A CT scan is only as good as its image quality, and stroke patients are not ideal imaging subjects. Many are confused, restless, or unable to follow instructions to hold still. Patient movement during the scan creates artifacts that degrade the images and can obscure or mimic pathology. This is a particular problem for CT perfusion, which requires the patient to remain still for a longer acquisition than a standard CT. One study found that roughly 19% of acute stroke patients moved enough during CT perfusion to compromise diagnostic and therapeutic accuracy, even after correction algorithms were applied.20PubMed. Head movement during cerebral CT perfusion imaging of acute ischaemic stroke: Characterisation and correlation with patient baseline features Head movement during perfusion imaging can degrade the analysis enough to potentially affect treatment decisions.21PubMed. Automatic detection of CT perfusion datasets unsuitable for analysis due to head movement of acute ischemic stroke patients
Novel motion-correction algorithms are being developed to address this issue. These computational fixes can recover useful diagnostic information from scans that would otherwise be unreadable, though they cannot fully compensate for severe motion.22Journal of NeuroInterventional Surgery. Motion artifact correction for cone beam CT stroke imaging: a prospective series
How Artificial Intelligence Is Changing CT Stroke Reading
One of the more promising developments is the use of AI to help radiologists spot early ischemic changes on plain CT. These algorithms are trained on thousands of scans and can flag subtle density changes that a human reader might overlook, especially under the time pressure of an emergency. A crossover reader study found that when radiologists used AI assistance, their overall diagnostic accuracy for acute ischemic stroke on non-contrast CT improved by about 3.6 percentage points, and sensitivity rose by roughly 4 points. The gains were largest for big strokes but were still present for small lesions under 5 milliliters.23Scientific Reports. Deep learning algorithm for automatic detection of acute ischemic stroke on noncontrast brain CT
AI-driven scoring systems, including automated versions of a widely used clinical scoring tool for early CT changes, have shown performance comparable to or better than physicians in recognizing early ischemic signs.24Journal of NeuroInterventional Surgery. Artificial intelligence-driven ASPECTS for the detection of early stroke changes in non-contrast CT: a systematic review and meta-analysis These tools do not replace the radiologist but function as a second set of eyes, particularly valuable during overnight shifts or at hospitals without on-site neuroradiology expertise.
Distinguishing Hemorrhage From Contrast Staining After Treatment
A separate challenge arises after stroke treatment rather than before it. When a patient undergoes a clot-retrieval procedure known as mechanical thrombectomy, follow-up CT may show bright areas in the brain. The critical question is whether that brightness represents new bleeding, a serious complication, or harmless leakage of the contrast dye used during the procedure. Standard CT cannot reliably tell the two apart. A newer technique called dual-energy CT addresses this by acquiring images at two different energy levels, which allows the software to distinguish blood from contrast material based on their different absorption characteristics. Research has shown that this approach provides high accuracy and excellent specificity for making that distinction, and it improves agreement between different radiologists reading the same scan.25PubMed Central. A Novel Dual-Energy CT Method for Detection and Differentiation of Intracerebral Hemorrhage From Contrast Extravasation in Stroke Patients After Endovascular Thrombectomy Getting this call right matters because misidentifying contrast staining as hemorrhage can lead to unnecessarily withholding blood thinners, while missing true hemorrhage can be dangerous.
Transient Ischemic Attacks and the Normal Scan
A transient ischemic attack, commonly called a TIA or “mini-stroke,” causes stroke-like symptoms that resolve on their own, usually within minutes to an hour. By definition, the symptoms are temporary, but that does not mean the brain escaped unscathed. A prospective study of TIA patients found a relevant cerebral infarction on CT in a meaningful subset of cases, yet these patients could not be clinically distinguished from TIA patients whose scans were clean.26JAMA Neurology. Transient Ischemic Attacks With and Without a Relevant Infarct on Computed Tomographic Scans Cannot Be Distinguished Clinically In other words, a normal CT after a TIA does not guarantee that no damage occurred. Many TIA patients with normal CT scans will show small infarctions on subsequent MRI. A TIA is now treated as a warning event regardless of what the CT shows, because the risk of a full stroke in the following days and weeks is substantial.