What Are Image Artifacts? CT, MRI, Radiography & Ultrasound

Imaging systems do not copy anatomy exactly. The x-ray tube, detector, magnetic field, ultrasound probe and reconstruction algorithms can add structures that aren't there, or show real structures distorted. That is why a radiologist, technologist or physicist looking at an image faces a constant question: "Is this really there?" Not every mark in the image belongs to the patient; some arise directly from the imaging process. These false appearances are called artifacts. Artifacts can lead to false-positive diagnoses, missed real pathology, and unnecessary repeat exposures — so recognizing them is as important as interpreting the image.

What is an artifact?

An artifact is a feature seen in the image that does not correspond to the imaged anatomy — or shows anatomy distorted. Its source is not the patient: it is the tube, detector, magnetic field, ultrasound physics, reconstruction or patient motion. An important distinction: some artifacts are wholly false (e.g. reverberation lines or a mirror image); others distort real anatomy (e.g. partial volume or beam hardening). So an artifact is not always "something that isn't there"; sometimes it is "the wrong version of what is."

Why do artifacts form?

There are three main origins:

• Physics-based: beam hardening, partial volume, chemical shift, reverberation — these arise inevitably from imaging physics; they cannot be wholly removed, only reduced.
• Hardware-based: a dead/drifting detector element (ring artifact), grid error, residual signal (lag), calibration drift.
• Patient-based: motion, metal implants, lack of cooperation.

One artifact can have several origins at once; metal, for instance, is both patient (implant) and physics (excessive attenuation) based, and worsens if the patient moves.¹

How does it affect diagnosis?

The risk runs two ways. False positive: mistaking an artifact for a real finding (reading a streak as a bleed). False negative: an artifact hiding a real finding (a metal artifact obscuring an adjacent lesion). The third, often silent cost is the repeat exposure — extra dose and workload. This is exactly where DoseSave's focus intersects: recognizing an artifact is not only a diagnostic skill but also a dose-optimization matter.

Radiography and DR

• Motion blur: if the patient or organ moves during exposure, edges blur and spatial resolution drops. Remedy: short time (high mA) and patient cooperation.
• Grid lines / grid cutoff: if the anti-scatter grid is misaligned or moves out of sync, a fine line pattern or one-sided signal loss is superimposed.
• Ghosting / residual signal (lag) and dead pixels — DR-specific: residual signal from a prior exposure can bleed into the new image as a faint "ghost"; dead/stuck pixels produce fixed-pattern dots.⁵ These are the artifacts quality control monitors regularly.
• Quantum mottle: low dose → few photons at the detector → a grainy, mottled appearance.⁴ This is the direct visual consequence of the dose–noise relationship (√N); not an "artifact" in the classic sense but a sign of insufficient signal — yet it degrades image quality just the same. It is the limit of acceptable noise when lowering dose — detail in Dose and Noise (√N).

Computed tomography (CT)

CT has the greatest variety of artifacts; reconstruction, the polychromatic beam and fast scan geometry produce many of them together. The most common:

• Beam hardening: standard backprojection does not fully account for the polychromatic spectrum. In a beam crossing dense tissue (e.g. petrous bone or metal), low-energy photons are absorbed first, so the beam's average energy rises (it hardens).¹ The result is cupping (a uniform object darkening toward the center) or dark wedge/webbing bands between two dense structures (e.g. bilateral hip prostheses). Remedy: pre-hardening filtration (e.g. ~10 mm Al) and software correction.
• Streak and metal artifacts: streaking occurs when a region's attenuation exceeds the detector's dynamic range. Classic sources are metallic dental fillings, implants like pacemakers/neurostimulators, and bullet fragments; motion of the high-density object (jaw movement, swallowing) markedly amplifies the streaks.¹
• Partial volume effect: if a voxel is large enough to span tissues of different density, the voxel value becomes their average; small structures deviate from their true value. Multi-slice CT (MDCT) with thin sections (e.g. ~1.25 mm) has largely reduced this.¹
• View aliasing: if too few projections are acquired to render sharp, high-frequency objects, aliasing appears. Rare clinically since modern scanners collect many views (950–3,000).¹
• Cone beam artifacts: undersampling along z at large cone angles; false connections appear between dense structures. Remedy: a more complete data set.¹

MRI

• Motion artifacts: breathing, cardiac motion or patient movement produce repeating "ghost" copies along the phase-encoding direction.
• Susceptibility: at metal or tissue–air boundaries the local magnetic field is distorted, causing signal loss and geometric distortion — the voids around dental fillings are the classic example.²
• Chemical shift: because protons in fat and water precess at slightly different frequencies, anatomy is spatially shifted at fat–water boundaries; a dark/bright band appears at the interface.²
• Wraparound (aliasing): anatomy outside the field of view (FOV) but within the slice volume is mapped to the opposite side of the image. Typically seen in the phase-encode direction; reduced by enlarging the FOV or applying an "anti-aliasing" saturation pulse just outside it.² One of the artifacts technologists meet most often in practice.

Ultrasound

• Reverberation: sound echoes back and forth between two closely spaced strong reflectors (or the probe and a surface); equally spaced, progressively fading false lines appear.³
• Mirror image: near a strong reflector (classic example: the liver–diaphragm interface), multipath reflections place a mirror copy of the real mass beyond the reflector.³
• Acoustic shadowing / posterior enhancement: behind a strongly attenuating structure (stone, calcification, bone) a dark shadow forms; behind a weakly attenuating one (e.g. a cyst), enhancement appears.
• Side / grating lobes: echoes from energy spreading outside the main beam are mismapped onto the main line, creating an artifact.³

Artifact or pathology?

The modalities differ, but the practical clues for telling them apart are shared:

• Check the geometry: artifacts often ignore anatomic boundaries; they run along a physical axis (streaks are radial, reverberation is equally spaced, wraparound sits at the FOV edge).
• Check repeatability: in a different plane/slice, or on a repeat with changed parameters, the artifact disappears or moves; a real finding stays.
• Look for the source: an artifact is usually tied to a known cause (metal, motion, FOV edge, strong reflector).

In short, an image is not reality itself but the result of a physical process — and that process sometimes adds what isn't there and sometimes hides what is. For a definitive call, physicist–radiologist collaboration and, when needed, re-imaging with changed parameters is the safest route. So artifact knowledge is a diagnostic skill for the radiologist, a daily reflex for the technologist, and a check on the system for the physicist.

← All articles