Resolution of a CT scan is not one dial. It is three pillars: spatial (how small a thing you can separate), contrast (how close two densities can be and still look different), temporal (how short a time you need to freeze motion). This page is those three. It is not a slice-thickness protocol.
If you meant slice thickness in millimetres → CT slice thickness. If you meant 4D / gating → 4D CT scan. If you meant motion as an artifact → artifacts in computed tomography.
The three pillars
| Pillar | What it measures | Why a reading changes |
|---|---|---|
| Spatial | Two small objects stay two objects (lp/cm, µm) | Hairline fracture, small nodule, inner-ear ossicles |
| Contrast | Two similar densities stay two greys | Liver lesion vs parenchyma; grey vs white matter |
| Temporal | Time to a complete image — freeze motion | Coronary CTA; a beating heart is otherwise a smear |
A “high-res” scan that is sharp but slow is useless on the heart. A fast scan that cannot tell two soft tissues apart is useless on the liver. The protocol picks a balance for one question.
Spatial
Think pixel density. Modern scanners quote spatial resolution under 0.5 mm. Units you will see: line pairs per centimetre, or micrometres. Hardware that sets it: detector-element pitch, focal-spot size (smaller spot, less geometric unsharpness), and — along z — slice thickness. Slice thickness is one input to spatial resolution along the table axis. The mm table and acquired-vs-reconstructed split are CT slice thickness, not a second copy here.
Contrast
Telling two close HU values apart. Soft tissue lives here. IV iodinated contrast raises the density of vessels and enhancing tissue on purpose so a lesion is not the same grey as the organ. The HU scale itself is Hounsfield units. This pillar is the ability to see the difference, not the scale.
Temporal
Time to one image. A high-speed camera vs a smear. Cardiac CTA is the proof case: dual-source systems take two tubes and two detector arrays at once and get a picture in about 65 ms — short enough to sit between beats. That is why coronary CTA exists as a non-invasive look at the arteries. 4D / respiratory gating (phase bins, RPM, beam-on only in a window) is a different job: 4D CT scan.
Hardware underneath
The tube makes the beam. The detector array counts what gets through. Tighter, smaller detector elements raise spatial resolution the way more pixels do on a sensor. A smaller focal spot sharpens edges. Sensitivity of those detectors also feeds contrast resolution — a detector that cannot tell two close fluences apart will not tell two close tissues apart.
Single-slice scanners took one cut per rotation. MDCT stacked detector rows: many slices per turn, scan times of a few seconds for a chest or abdomen, and isotropic voxels (cubes) that reformat without a squash. Rotation times under a second were already a 1970s third-generation goal; MDCT is the volume version of that race.
The human half
A still patient is temporal resolution you do not have to buy. Breath-hold on chest and abdomen is the cheap fix for voluntary motion. Residual blur — heartbeat, bowel, a fidget — is the motion artifact on the catalog. Body habitus eats photons and raises noise; the technologist changes mA and reconstruction, not “resolution” as a slogan.
Contrast agents are the other human-side lever: they do not sharpen an edge (spatial); they open a density gap (contrast).
Higher is not automatically better
Match the pillar to the question. A hairline fracture wants spatial. A faint liver lesion wants contrast (often with IV contrast). A coronary wants temporal. Pushing spatial on a routine abdomen costs dose and noise for a finding you could have seen on 3 mm. ALARA still wins.
Iterative and deep-learning reconstruction take some of the old dose-for-sharpness tax back. They reconstruct a usable picture from noisier data. They do not invent a fourth pillar.
PYCAD builds the software side of products that have to expose these three as separate facts. Case studies.