CTA Carotid Scan: Procedure, Accuracy, and Stroke Risk

CT angiography of the carotid arteries is one of the most accurate noninvasive ways to visualize the blood vessels that supply the brain, reaching diagnostic accuracy of about 97% for grading carotid stenosis in comparative studies.¹ The scan works by injecting iodine-based contrast dye into a vein while a fast CT scanner captures detailed cross-sectional images from the aortic arch up through the skull. It takes only seconds, and the result is a three-dimensional map of the carotid arteries that shows not just how narrow the vessels have become but what the walls themselves look like.

How the Scan Is Performed

A carotid CTA is fast, usually completed in well under a minute of actual scanning. You lie on a CT table while a small IV catheter delivers contrast dye into an arm vein. Typical protocols use around 80 mL of contrast injected at 4 to 5 mL per second, followed by a saline flush to push the contrast through.² The scanner uses “bolus tracking,” which means it watches for the contrast to reach a target area (often the aorta or the carotid arteries themselves) and triggers the scan at exactly the right moment so the vessels are filled with bright contrast when the images are captured.

Some newer protocols have reduced the contrast volume dramatically. One study compared a standard 80 mL injection with a low-volume protocol using just 30 mL, finding that the smaller dose still produced diagnostic images when the timing was optimized.³ Less contrast is better for the kidneys, which is relevant for patients who get scanned repeatedly or who already have reduced kidney function. The coverage typically runs from the aortic arch to the top of the skull, so the scan captures the entire path blood takes from the heart to the brain.

Diagnostic Accuracy Compared With Other Imaging

The gold standard for measuring carotid narrowing has traditionally been digital subtraction angiography, or DSA, which threads a catheter directly into the artery and injects dye under fluoroscopy. It is invasive and carries a small stroke risk, so there has been a decades-long push to replace it with noninvasive alternatives. CTA has largely succeeded. In a head-to-head comparison involving 170 patients, CTA matched DSA’s grading with 97% accuracy, outperforming both MR angiography (92 to 95%) and Doppler ultrasound (76%).⁴

For detecting severe stenosis specifically, CTA performs even more impressively. One prospective study found 100% sensitivity and 100% specificity for severe internal carotid artery stenosis on CTA, while gadolinium-enhanced MR angiography reached 93% sensitivity in the same patients.⁵ That said, CTA is less sensitive for moderate stenosis in the 50 to 69% range, where one study found sensitivity dropping to about 65%.⁶ This matters because the moderate-stenosis zone is exactly where clinical decisions get tricky: some of those patients benefit from surgery and some do not, so a missed moderate narrowing could change the treatment conversation.

The trade-offs between CTA and MRA often come down to the patient. CTA involves radiation and iodine contrast, while MRA avoids radiation but is less accurate without its own contrast agent (gadolinium). MRA is the go-to when kidney function makes iodine risky, or when calcification in the artery walls is expected to obscure the CTA images. Duplex ultrasound remains the first-line screening tool in most settings because it is cheap, fast, and carries no radiation, but it is operator-dependent and misses details that CTA captures.

How Stenosis Is Measured on CTA

Measuring how narrow a carotid artery has become sounds straightforward, but there are actually multiple competing grading systems, and they produce different numbers from the same artery. The two most widely used are the NASCET method (named after the North American Symptomatic Carotid Endarterectomy Trial) and the ECST method (from the European Carotid Surgery Trial). Both compare the narrowest point in the internal carotid artery to a reference diameter, but they choose different reference points, so ECST grades always come out higher than NASCET grades for the same stenosis. A study comparing all three major grading methods on 64-slice CTA found that radiologists could reproduce all of them reliably, with high agreement between readers.⁷

One advantage CTA has over catheter angiography is that it lets you measure the actual millimeter diameter of the residual lumen directly, rather than relying solely on percentage ratios. Research has shown a tight linear relationship between these direct millimeter measurements and the traditional NASCET-style percentages: a residual lumen of about 1.3 mm corresponds to 70% stenosis by NASCET, and 2.2 mm corresponds to 50%.⁸ Using the millimeter cutoff for severe stenosis (70% or greater) yielded roughly 88% sensitivity and 92% specificity in that study. Direct measurement can be a useful cross-check when the percentage calculation is ambiguous, as sometimes happens with unusual vessel anatomy.

Plaque Composition and Stroke Risk

Knowing how narrow the artery is tells you only part of the story. Two patients with the same degree of stenosis can have wildly different stroke risks depending on what the plaque is made of. A heavily calcified, stable plaque is much less likely to rupture than a “soft” plaque filled with lipid and prone to internal bleeding. CTA can distinguish between these types based on how dense the plaque components appear on the scan.

A systematic review and meta-analysis of CTA plaque features found that soft plaque roughly tripled the odds of stroke on the same side, with an odds ratio of about 2.9. Plaque ulceration, where the surface has broken down to form a crater, doubled the odds. Thickening of the common carotid artery wall was the strongest predictor, with an odds ratio over 6. Calcified plaque, by contrast, was actually associated with fewer downstream strokes.⁹ In other words, calcification is protective rather than dangerous, which is counterintuitive if you think of any “hardening” as bad.

More advanced techniques can push plaque characterization further. Dynamic enhancement patterns on CTA, which track how much contrast the plaque absorbs over time, correlate with histology: fibrous tissue takes up more contrast, while lipid-rich necrotic cores with hemorrhage take up less.¹⁰ Dual-energy CT, which scans at two different X-ray energy levels simultaneously, has shown strong performance in separating lipid-rich necrotic cores from other plaque components. In one imaging-histopathology comparison, low-energy CT images achieved about 90% accuracy in identifying necrotic cores, with nearly 89% sensitivity and 91% specificity.¹¹ Distinguishing intraplaque hemorrhage from fibrous tissue was harder, with accuracy hovering around 65%, so there is still room for improvement.

When Calcification Gets in the Way

Ironically, the same calcification that signals lower stroke risk creates the biggest headache for reading the scan. Dense calcium in the artery wall blooms outward on CT images, making the residual lumen look smaller than it really is. This artifact is one of the most common reasons CTA overestimates stenosis severity, sometimes grading a moderate narrowing as severe or calling a tight stenosis a complete occlusion.

Dual-energy CT offers a partial solution. By acquiring images at two energy levels, software can generate “virtual non-calcium” reconstructions that digitally subtract the calcium, revealing the underlying lumen. A modified dual-energy algorithm showed excellent agreement with DSA for stenosis grading, with a correlation of 0.991 between the virtual non-calcium images and catheter angiography, while conventional CTA and a simpler subtraction method both tended to overestimate the narrowing.¹² Not every scanner has dual-energy capability, though, so at many hospitals this remains a known limitation rather than a solved problem.

Pseudo-Occlusion and Other Diagnostic Traps

Another pitfall is pseudo-occlusion. When a large clot blocks a major artery inside the brain, blood flow through the carotid on that side can slow to a trickle. On CTA, the sluggish flow means contrast barely fills the carotid, and the vessel can look completely blocked even though it is technically still open. Distinguishing true occlusion from pseudo-occlusion matters enormously: a truly blocked carotid cannot be reopened the same way, while a pseudo-occluded one might respond to treatment aimed at the intracranial clot.¹³ In acute stroke settings where minutes count, this distinction can shape whether and how interventional teams proceed.

Carotid artery dissection is another condition that can be tricky to catch on CTA. Dissection occurs when the inner lining of the artery tears, allowing blood to track into the vessel wall. It is a major cause of stroke in younger adults. Cross-sectional imaging including CTA can pick up the classic signs — a crescent of blood in the wall, a narrowed or irregular lumen, or a flap of tissue inside the vessel — but the diagnosis can be difficult, and misinterpretation is well-documented.¹⁴ MRI with fat-saturated sequences is sometimes better at confirming intramural blood, so CTA and MRI complement each other rather than compete in dissection workups.

Carotid Webs

A less familiar finding that CTA has helped bring to attention is the carotid web. This is a thin shelf of tissue that protrudes into the lumen of the carotid artery, usually at the bulb just above the bifurcation. On CTA, it appears as a thin intraluminal filling defect along the posterior wall on oblique views and as a septum on axial slices. Histopathology in cases that have been surgically removed suggests these webs are a form of fibromuscular dysplasia.¹⁵

Carotid webs are thought to cause strokes by disrupting blood flow and promoting clot formation in the pocket behind the shelf. Computational fluid dynamics studies, built from CTA-based 3D models of the artery, have confirmed that the area behind a carotid web develops recirculation zones, low wall shear stress, and high oscillatory shear — all conditions that encourage platelets to stick together.¹⁶ Before CTA became routine, many of these webs were probably missed, and the resulting strokes were filed under “cryptogenic,” meaning no cause was found. CTA has given clinicians a way to identify them and potentially prevent recurrence with anticoagulation or surgery.

Radiation and Contrast Safety

Radiation dose is an unavoidable part of any CT scan. For carotid CTA, effective doses have historically been in a range comparable to a few chest X-rays, but modern dose-reduction techniques have pushed them substantially lower. Automated kilovolt selection, which lets the scanner choose a lower tube voltage for thinner patients, has achieved dose reductions of more than 30% while maintaining image quality.¹⁷ Going further, a protocol combining low tube voltage (80 kVp), reduced iodine concentration, and iterative image reconstruction cut the effective dose in half compared with standard settings, and the images were still diagnostic.¹⁸ The trade-off with lower voltage is grainier images and sometimes more streak artifacts, but for most patients these do not affect the clinical read.

Iodine contrast carries its own risks. Allergic reactions range from mild hives to rare anaphylaxis. The more common concern, especially in older patients and those with diabetes or existing kidney disease, is contrast-induced kidney injury, where kidney function dips temporarily after iodine exposure. For a one-time diagnostic CTA, the risk is low in people with normal kidneys, but it becomes more relevant for patients who undergo repeated contrast-enhanced studies, such as those being monitored after stent placement.

Pediatric Considerations

Children and young adults present a different risk-benefit equation. Radiation carries a greater long-term cancer risk in younger bodies because their cells are dividing faster and they have more years ahead in which a radiation-triggered cancer could develop. A literature review on pediatric carotid imaging after blunt trauma noted that the radiation dose from a CTA is substantially higher than a standard head CT — a median of about 138 mGy versus 33 mGy — and that cumulative CT doses around 50 mGy roughly triple the risk of leukemia in children.¹⁹ Meanwhile, only a tiny fraction of pediatric CTAs performed after trauma actually reveal an injury: between 0.17% and 1.2% in the studies reviewed. That low yield makes the decision to scan a child much more fraught than in an adult, and many centers now use clinical screening criteria to limit CTA to children at highest risk.

Guiding Surgery and Stenting Decisions

CTA’s real clinical power goes beyond just establishing a diagnosis. Surgeons and interventionalists rely on the scan to plan the procedure. For carotid endarterectomy, where the plaque is physically removed through an incision in the neck, the surgeon needs to see the extent of disease, the anatomy of the bifurcation, and whether there are features like a high carotid bifurcation that would make neck access difficult. For carotid stenting, the anatomy of the aortic arch matters just as much: a tortuous or heavily calcified arch can make it impossible to navigate a catheter safely into position. CTA and MRA have taken on particular importance for stenting decisions because they can map arch anatomy in three dimensions before the patient ever reaches the catheterization lab.²⁰

CTA also plays a role after surgery. A study of post-endarterectomy CTA findings showed that most patients had minimal residual narrowing, with about 78% showing 0 to 20% residual stenosis. But the scan also revealed common post-surgical quirks: a step-off at the lower end of the surgical site was seen in 60% of patients, and intimal flaps — small tags of tissue protruding into the lumen — appeared in the external carotid artery in 18% of cases. When flaps appeared in the internal carotid artery (6%), all of those patients went on to develop a clot at that site.²¹ Post-operative CTA gives the surgical team an early warning about complications that might otherwise surface as a new stroke days later.

Collateral Circulation and the Bigger Vascular Picture

When one carotid artery is severely narrowed or blocked, the brain does not simply lose all blood supply on that side. Collateral pathways — alternate routes through connecting arteries at the base of the brain and from the external carotid system — can partially compensate. CTA can map these collaterals by capturing the full vascular tree from the aortic arch to the top of the skull in a single acquisition.²² The quality of collateral circulation turns out to be one of the strongest predictors of how a patient with carotid occlusion will do over time. A patient with robust collaterals may tolerate a completely blocked carotid with few symptoms, while someone with poor collaterals can suffer a major stroke from the same degree of disease.

Researchers have also begun using CTA-derived 3D models as the starting geometry for computational fluid dynamics simulations, which calculate the forces blood flow exerts on the artery walls. These hemodynamic models can identify regions of disturbed flow and predict where plaque is likely to grow or rupture, adding a layer of functional information on top of the anatomical picture CTA provides.²³ This remains largely a research tool, but it represents one direction the field is moving.

Artificial Intelligence in CTA Interpretation

Reading a carotid CTA is not trivial. The datasets are large, the reconstructions take time, and subtle findings like small ulcerations or thin webs can be missed. AI tools, particularly deep learning models, are being developed to automate parts of the process. A study published in Radiology described a convolutional neural network trained on head-and-neck CTA images that could detect vessel stenosis and classify plaque with accuracy equivalent to experienced radiologists.²⁴ Other AI approaches focus on segmenting the carotid arteries in three dimensions, which is a necessary first step before any automated measurement can happen.²⁵

The promise of AI here is less about replacing radiologists and more about triage and consistency. An automated system could flag scans with high-risk plaque features or significant stenosis for priority reading, reduce variability between different readers, and speed up reporting in acute stroke scenarios where every minute of delay costs brain tissue. These tools are still being validated in clinical settings, but some are already working their way into commercial software packages integrated with CT scanners.