A coronary CT angiography (CTA) is highly accurate for ruling out significant coronary artery disease in suitable patients, with sensitivity and specificity commonly above 90 percent in modern multicenter studies. Its accuracy is lower for estimating the exact severity of a borderline stenosis and for predicting which individual plaque will rupture. In practice, a negative scan is powerful, a positive scan often needs corroboration, and the technology is best understood as a plaque-imaging tool rather than a simple blockage detector.

Coronary CTA pairs a contrast injection with a rapidly rotating X-ray scanner and ECG gating. The scanner freezes cardiac motion, and the software reconstructs the coronary tree in three dimensions. Accuracy depends less on the physics of the tube and more on patient selection, heart rate control, calcium burden, and reader experience. The sections below unpack each of those dependencies with numbers.

How the Scanner Turns Motion Into Images

A coronary artery moves several millimeters during each heartbeat, which is far more than the diameter of the vessel. To get a sharp image, the scanner must either outrun that motion or synchronize with it.

Multidetector scanners now acquire a full heart volume in a single gantry rotation. Current 64-detector and wider systems complete a cardiac scan in roughly 5 to 15 seconds. The system records the electrical signal from the ECG and reconstructs images only during a quiet phase, typically diastole at 60 to 80 percent of the R-R interval. A slow, regular heart rate gives a wider quiet window, which is why patients often receive rate control before scanning.

Temporal resolution matters more than raw slice count. A scanner with 100 to 175 milliseconds of temporal resolution can freeze a heart beating 60 times per minute but struggles above 75 beats per minute. Dual-source designs and newer single-heartbeat scanners reach much shorter effective windows.

Spatial resolution sets the smallest detectable plaque. Modern systems resolve structures near 0.3 to 0.5 millimeters. A noncalcified plaque can be seen when it narrows a vessel enough to produce a visible rim of soft tissue.

Contrast makes the lumen visible. The dense lumen appears bright, and plaque appears as a darker filling defect or as calcified bright spots. The interpretable question is not simply “is there a narrowing” but “is the lumen visibly narrowed relative to the reference vessel.”

What the Accuracy Numbers Actually Say

Accuracy splits into distinct claims, and each has a different number attached.

Diagnostic sensitivity and specificity come from comparing CTA against invasive coronary angiography (ICA) as the reference. Large multicenter trials and meta-analyses generally place sensitivity for detecting a stenosis of 50 percent or greater in the low-to-mid 90s and specificity in the mid-80s to low-90s. Per-vessel specificity is consistently lower than per-patient specificity.

The high sensitivity is what makes a negative scan clinically strong. A normal coronary CTA in a patient with low-to-intermediate pretest probability makes obstructive disease very unlikely, and guidelines from major cardiology societies endorse CTA in that setting. That endorsement rests on the same trials that produced the accuracy figures.

Specificity is where the conversation gets honest. A positive CTA in a patient with heavy calcification can reflect “blooming” of calcium rather than a true tight stenosis. This is the counterintuitive part: more calcium does not automatically mean a more informative scan. Very dense calcified plaque scatters X-rays and can overestimate narrowing. That is why the coronary calcium score is reported separately and why a high Agatston score (commonly above 400, and more so above 1000) reduces confidence in stenosis grading.

A useful way to frame it: CTA is excellent at confirming that disease is absent and good at confirming that disease is present, but only moderate at grading how tight a specific lesion is when calcium is heavy. The plain version is that a clean scan is reassuring, while a muddy scan with dense calcium may under- or over-call a blockage.

Comparison With Alternative Tests

Test What it measures best Typical accuracy profile Main limitation
Coronary CTA Plaque presence and anatomy Sensitivity >90%, specificity mid-80s to low-90s for 50% stenosis Overcalls stenosis with heavy calcium
Invasive angiography (ICA) Lumen narrowing, with pressure wire for physiology Reference standard for anatomy Invasive, small procedural risk, radiation
Stress echocardiography Ischemia under exercise or drug stress Good sensitivity, lower specificity in some groups Cannot see plaque, operator dependent
SPECT or PET perfusion Regional blood flow and ischemia High specificity, moderate sensitivity Radiation, cannot show plaque anatomy
Cardiac MRI stress perfusion Perfusion and function without ionizing radiation Good accuracy in experienced centers Longer protocols, limited availability
CT-derived FFR (CT-FFR) Functional significance of a lesion Adds specificity to anatomic CTA Needs suitable images and software
Coronary calcium score Total calcified plaque burden Strong risk prediction, not a stenosis test Ignores noncalcified plaque

The table makes the trade-off visible. CTA answers an anatomic question. Stress tests answer a functional one. CT-FFR sits between them by estimating flow significance from CTA data, which can rescue specificity when a lesion looks borderline.

The Timeline From Scan to Answer

Stage What happens Typical timing
Preparation Heart rate check, rate control if needed, IV line 15 to 60 minutes before scan
Scout and timing Test bolus or bolus tracking to sync contrast arrival 1 to 3 minutes
Acquisition Contrast injection plus scan 5 to 15 seconds of scanning
Reconstruction ECG-gated image building, motion correction 5 to 30 minutes
Interpretation Radiologist or cardiologist reads, calcium scored Same day to several days
Report delivery Structured report to referring clinician Typically 1 to 3 days
Follow-up decision Further testing or medical management Days to weeks

The acquisition itself is fast. Most of the elapsed time is preparation and reading. A patient should not expect the answer in the scanner room.

Edge Cases That Change Accuracy

Several patient factors move accuracy in predictable directions.

Heart rate and rhythm. Atrial fibrillation and frequent ectopy blur gating and degrade image quality. Many centers decline CTA in uncontrolled atrial fibrillation or use specific arrhythmia protocols.

Calcium burden. As noted, dense calcium reduces stenosis grading confidence. The calcium score itself, however, is highly reproducible and adds risk information independent of stenosis.

Body habitus. Large body size increases X-ray scatter and noise. Higher tube voltage and iterative reconstruction help, but image quality still trends downward.

Stent and bypass history. Stents cause metal artifact, and in-stent restenosis is hard to exclude. Bypass grafts are easier to assess than the native vessels they replace.

Age and pretest probability. In very low-probability patients, a positive result has a higher chance of being false, simply because of how prevalence interacts with specificity. In high-probability patients, a negative result carries more weight than in a screening population.

Kidney function and contrast allergy. These are safety considerations rather than accuracy ones, but they determine whether the test can be done at all. Renal function is checked before contrast in many protocols.

Misconceptions Worth Correcting

CTA does not require a catheter in the groin or wrist. It is an intravenous study. The word “angiography” overlaps with the invasive procedure, which confuses patients.

A high calcium score is not a diagnosis of a blockage. It is a measure of calcified plaque burden and a risk marker. Patients with high scores often have nonobstructive disease.

A normal CTA does not guarantee no future event. It makes obstructive disease unlikely at the time of the scan. Noncalcified plaque can exist below resolution, and plaque can progress.

Radiation is not zero. A modern coronary CTA typically delivers a few millisieverts, and protocols vary widely. The dose has fallen substantially with newer scanners and iterative techniques, but it is still a real consideration, particularly in younger patients.

The test is not a screening tool for everyone. Guidelines position it for symptomatic patients with low-to-intermediate pretest probability or for selected patients with equivocal prior testing, not for mass population screening.

Practical Steps and Thresholds

Discuss pretest probability with the referring clinician before scheduling. CTA is strongest when pretest probability is low to intermediate. Expect heart rate control. Many protocols target a resting rate below roughly 60 to 65 beats per minute for older scanners, though targets vary by hardware. Ask how the calcium score will be reported alongside the stenosis assessment. If the CTA shows a borderline lesion (roughly 50 to 70 percent) in a patient with symptoms, ask whether CT-FFR or a stress test would clarify functional significance. Red flag: a report that grades a stenosis as “severe” in a heavily calcified vessel without functional corroboration deserves a second look before any invasive step. Red flag: a negative CTA in a patient with ongoing typical angina should prompt reassessment rather than reassurance.

These steps reflect how accuracy translates into decisions. The scan is one input, and the pretest probability is the multiplier that determines what a result means.

Limits of What Accuracy Can Promise

The strongest evidence supports CTA as a rule-out test for obstructive disease in appropriately selected patients. The evidence is weaker for precise stenosis grading in calcified vessels, for predicting plaque rupture, and for outcomes in asymptomatic populations.

Accuracy figures also come from controlled trial settings with experienced readers. Community performance can differ. A scan read at a high-volume center with motion-correction software is not identical to one read elsewhere.

One more limit deserves emphasis. Accuracy is a property of a test applied to a population, not a guarantee for an individual. A 95 percent sensitive test still misses some disease, and a 90 percent specific test still produces false positives. The right question is not “is CTA accurate” in the abstract but “what does this result change for this patient.”

For a patient with atypical symptoms and a low pretest probability, a negative CTA meaningfully lowers the chance of obstructive disease. For a patient with heavy calcium and typical angina, the same test may generate more questions than answers. Matching the test to the question is the real accuracy strategy.