Alicia sees a three-dimensional coronary CT image and assumes the scanner took a photograph of the arteries. Tricia points out that the heart was moving, the x-ray source was rotating and the computer never saw one complete three-dimensional heart at a single instant. Kai Kai asks the right question: how did many x-ray measurements become a stable image of a structure that never stopped moving?
Cardiac CT works by sending x-rays through the chest from many angles, measuring how strongly tissues attenuate those x-rays, synchronising acquisition or reconstruction with the cardiac cycle and using mathematical reconstruction to create cross-sectional images and three-dimensional representations. Intravenous iodine can make blood highly attenuating so vessel lumens are visible. ECG timing helps select phases when coronary motion is smaller or lets images be reconstructed across the whole heartbeat.
This article supports How the Heart Works and complements How Cardiac MRI Works and How Echocardiography Works. Here the reader job is x-ray CT measurement physics, not diagnosis or personal scan interpretation.
1. CT begins with x-ray attenuation
X-rays are electromagnetic radiation energetic enough to pass through much of the body. Different materials attenuate the beam by different amounts.
Dense or high-atomic-number materials tend to attenuate more strongly than air or many soft tissues.
The NIBIB overview Computed Tomography (CT) explains how a rotating x-ray source and opposing detectors collect many transmission measurements that computers reconstruct into slices.
2. A CT detector measures transmitted x-rays, not anatomy directly
For each projection angle, detectors measure how much x-ray energy reaches them after passing through the body.
The scanner therefore begins with a large set of attenuation measurements along many paths rather than with a finished image.
Anatomy emerges only after those measurements are combined mathematically.
3. Reconstruction solves an inverse problem
The scanner knows the x-ray paths and detector measurements and must infer the distribution of attenuation inside the body that could have produced them.
Filtered back projection and iterative or deep-learning-assisted methods are different approaches to this inverse problem.
The final CT image is therefore a reconstruction constrained by measured data and the mathematical method used to recover spatial information.
4. CT values describe relative attenuation
Reconstructed voxel values are commonly expressed in Hounsfield units, with water near 0 and air near −1000 under the standard convention.
Bone, calcification and iodine-enhanced blood can have substantially positive values.
A CT value is therefore a calibrated measure of x-ray attenuation, not a direct chemical concentration unless additional assumptions and calibration are used.
5. Cardiac CT has a motion problem that ordinary static CT does not
Coronary arteries are only a few millimetres wide and move continuously with the beating heart.
If data from substantially different cardiac positions are combined into one reconstruction, vessel edges blur or appear discontinuous.
Cardiac CT therefore adds a timing problem to the ordinary spatial-reconstruction problem.
6. ECG gating provides the cardiac clock
An ECG recorded during scanning identifies repeating cardiac phases.
Prospective ECG triggering activates x-ray acquisition during selected portions of the cycle. Retrospective ECG gating acquires data over a broader interval and sorts or reconstructs it according to recorded cardiac phase.
The 2025 review Acquisition and Reconstruction Techniques for Coronary CT Angiography describes how modern gating, wide-detector coverage and motion-correction methods reduce coronary-motion artifacts.
7. Diastole is often useful because coronary motion can be lower
At many ordinary heart rates, portions of diastole provide relatively low coronary motion compared with rapid systolic contraction.
Prospective acquisitions often target an expected low-motion phase to reduce radiation exposure while preserving coronary detail.
The optimal phase is not universal; heart rate, rhythm and scanner technology influence which interval is most stable.
8. Systolic phases can sometimes be more stable at higher heart rates
As heart rate rises, the diastolic rest interval shortens. Some scanners therefore reconstruct end-systolic or other phases when motion is lower.
Automatic phase-selection and motion-correction algorithms can compare multiple phases to identify the least-motion reconstruction.
The phrase “scan in diastole” is therefore a useful starting rule, not a universal physical law.
9. Temporal resolution determines how much motion is frozen
Temporal resolution describes the time interval represented by the data contributing to one image.
Faster gantry rotation, dual-source systems and specialised reconstruction can reduce the effective acquisition window.
Shorter temporal windows reduce motion blur but do not eliminate the need for sufficient photon signal or spatial sampling.
10. Spatial resolution determines how finely anatomy can be separated
Small coronary branches, vessel walls and calcifications require high spatial resolution.
Voxel size depends on detector geometry, field of view, reconstruction matrix and slice thickness.
Higher spatial resolution generally increases image noise unless acquisition or reconstruction compensates with more photons or improved algorithms.
11. Iodine contrast makes the blood pool conspicuous
Iodine has high x-ray attenuation. Intravenous iodinated contrast therefore increases attenuation of blood during the appropriate circulation phase.
When coronary lumens fill with contrast, their attenuation differs strongly from surrounding fat, myocardium and many plaque components.
Contrast does not dye the artery permanently; it briefly changes the x-ray attenuation of flowing blood while the scan is acquired.
12. Contrast timing is a transport problem
The injected bolus travels through peripheral veins, right heart, lungs, left heart and aorta before reaching the coronary arteries.
Scan timing must align with a period when arterial enhancement is sufficient for the intended measurement.
Bolus tracking or test-bolus methods estimate this circulation time rather than assuming every person has identical transit timing.
13. Coronary CT angiography is different from non-contrast calcium imaging
Coronary CT angiography uses iodinated contrast to display the coronary lumen and vessel wall relationships.
Coronary calcium scoring is usually performed without intravenous contrast and quantifies high-attenuation calcified material using a specific acquisition and scoring method.
The two scans can use the same underlying CT physics while answering different measurement questions.
14. Calcium produces strong attenuation and can create blooming
Calcification can appear larger than its true physical dimensions because of finite spatial resolution and partial-volume effects.
This blooming can make the remaining contrast-filled lumen appear narrower than it is.
High-resolution acquisition and modern reconstruction methods aim to reduce this artifact, but the measurement remains limited by physics.
15. Partial-volume averaging mixes tissues inside one voxel
If one voxel contains calcium, contrast-filled blood and soft tissue, its CT value reflects a mixture rather than one pure material.
Small structures therefore become harder to classify when they occupy only part of a voxel.
Spatial resolution is not simply how sharp the image looks; it determines which physical structures can be separated quantitatively.
16. Wide-detector CT can cover the heart in fewer rotations
Detector arrays with greater z-axis coverage can image more of the heart during one rotation.
Whole-heart coverage can reduce stair-step artifacts caused when adjacent slabs are acquired during slightly different cardiac positions.
This illustrates how scanner geometry changes the motion-reconstruction problem even before software correction is applied.
17. Dual-source CT improves temporal resolution
Dual-source systems use two x-ray source-detector pairs positioned around the gantry.
Because each source needs to cover a smaller angular range to reconstruct a cardiac image, effective temporal resolution can improve.
The design therefore attacks motion blur through hardware geometry rather than only through faster rotation or software.
18. Photon-counting CT changes detector physics
Conventional energy-integrating detectors sum deposited x-ray energy over an interval. Photon-counting detectors register individual photons and their energy ranges.
NIBIB notes that photon-counting CT can improve spatial resolution and spectral information while supporting dose or contrast reductions in selected applications.
The fundamental CT task remains the same—reconstruct attenuation from transmitted x-rays—but the detector provides richer raw information.
19. Iterative and deep-learning reconstruction trade computation for noise control
Filtered back projection is fast and mathematically direct but can become noisy at low photon counts.
Iterative methods repeatedly compare candidate images with the measured data and system model. Deep-learning reconstruction uses trained models to suppress noise or artifacts while preserving structure.
The 2025 Radiographics review describes deep-learning reconstruction as one modern route to lowering noise and potentially radiation dose while maintaining image quality.
20. Motion correction is an inference about where the artery moved
Motion-correction algorithms can use information from neighbouring phases to estimate coronary displacement during acquisition.
The corrected image is therefore not a raw frame captured in one instant; it is a reconstruction informed by a motion model.
This can improve clarity, but it also means understanding the algorithm matters when interpreting small structures near the resolution limit.
21. Retrospective gating can reconstruct function across the whole cardiac cycle
If data are available across many cardiac phases, CT can reconstruct cine sequences showing chamber motion.
End-diastolic and end-systolic volumes can be measured and used to calculate stroke volume and ejection fraction.
A 2026 review of CT functional assessment stresses that full-cycle cine measurements and limited-phase approaches are informative but not interchangeable.
22. Radiation dose is part of the acquisition design
CT uses ionising radiation, so acquisition protocols aim to obtain sufficient diagnostic signal while avoiding unnecessary exposure.
Tube current, tube voltage, scan length, gating mode, pitch, heart rate and reconstruction method all influence dose.
Prospective ECG triggering often reduces exposure compared with continuously irradiating through the entire cardiac cycle, though the appropriate protocol depends on the measurement job.
23. Lower tube voltage changes both dose and iodine contrast
Reducing x-ray tube voltage can reduce radiation output and can increase the relative attenuation produced by iodine at suitable energies.
However, lower voltage also changes photon penetration and noise, especially in larger bodies.
Protocol selection therefore balances body size, detector performance, desired contrast and acceptable noise rather than simply choosing the lowest voltage.
24. Arrhythmia challenges the assumption of repeatable cycles
Many gated CT methods combine data from a selected cardiac phase across one or more heartbeats.
If cycle length varies substantially, the same percentage of two RR intervals may not represent an identical mechanical state.
Irregular rhythm can therefore produce misregistration or phase inconsistency, although newer scanners and reconstruction techniques can reduce some of these effects.
25. Breath-holding reduces a second source of motion
The heart also moves with the diaphragm and chest during respiration.
A short breath-hold reduces respiratory displacement during the coronary acquisition window.
Cardiac CT therefore synchronises two moving systems by controlling breathing while electronically tracking the heartbeat.
26. Worked problem: same artery, different cardiac phase
A coronary segment moves 4 mm during one reconstruction window but only 1 mm during another.
If spatial resolution is near 0.5–1 mm, the first window can produce substantial blur while the second preserves vessel edges much better.
The anatomy did not change; the motion during sampling changed the image.
27. Worked problem: partial-volume error in a tiny vessel
Imagine a vessel lumen only two voxels across. One boundary voxel contains half contrast-filled blood and half calcification.
Its reconstructed attenuation becomes a mixture and the apparent lumen boundary shifts.
A small one-voxel error can therefore represent a large percentage of the apparent diameter when the structure itself is only a few voxels wide.
28. Worked problem: prospective versus retrospective acquisition
A prospective scan acquires only a selected low-motion phase. It cannot reconstruct every cardiac phase because those other phases were never irradiated and sampled.
A retrospective acquisition samples across more of the cycle and can support cine reconstruction, but usually at a higher radiation cost unless dose modulation and modern hardware reduce exposure.
The acquisition must therefore match the reader job before scanning begins.
29. Worked problem: a sharper image does not automatically mean a more accurate measurement
Suppose a reconstruction algorithm strongly suppresses noise and makes vessel edges look smoother.
If it also changes the representation of tiny high-contrast structures, visual sharpness and quantitative fidelity may not improve by the same amount.
Image appearance and measurement accuracy therefore must be validated separately.
30. The cardiac-CT mechanism in one causal chain
An x-ray tube emits photons while rotating around the chest. Tissue attenuates the beam according to composition and path length. Detectors record transmitted intensity from many angles. Intravenous iodine temporarily increases blood attenuation so coronary lumens become conspicuous. ECG timing assigns projections to cardiac phases. Prospective triggering selects a limited low-motion window, while retrospective gating can preserve broader cycle information. Reconstruction algorithms solve for the spatial attenuation map, and motion-correction methods can estimate displacement during acquisition. Spatial resolution, temporal resolution, photon statistics, calcium blooming, partial-volume effects and respiratory motion determine the fidelity of the final coronary and chamber images.
Alicia stops calling the 3D display a photograph. Tricia marks the heart phase and contrast timing beside every reconstruction. Kai Kai asks which information was actually acquired and which was inferred by reconstruction before trusting a smooth-looking vessel.
The deeper lesson is that cardiac CT freezes motion computationally. Its remarkable image is the endpoint of x-ray physics, cardiac timing and reconstruction—not a direct optical view inside the chest.
Evidence trail and connected reading
For general CT physics, see NIBIB: Computed Tomography. For contemporary coronary CT acquisition and reconstruction, see Acquisition and Reconstruction Techniques for Coronary CT Angiography. For current functional CT strategies, see CT functional assessment: Why and how?.
Return to the parent: How the Heart Works. Continue to How Cardiac MRI Works, How Echocardiography Works and How the ECG Works.