Artefacts

For Medical Students

What It Shows

An artefact is a feature of the image that does not correspond to anything in the patient. 5.2.12 covers the basic principles.

Each one has a mechanism and a characteristic appearance, so they are recognisable rather than random. 5.2.18 covers gradient echo imaging.

It matters because an artefact that looks like pathology is reported as pathology, and this chapter exists largely to prevent that. 5.1.6 covers imaging pitfalls.

How It Is Acquired

Motion during the acquisition blurs and ghosts along the phase direction, which is why a restless knee produces repeated edges. 5.5.1 covers the standard protocol.

Metal distorts the local field, producing signal loss and geometric distortion around it. 5.5.3 covers the post-operative protocol.

Chemical shift displaces the fat signal relative to water at a fat and water boundary, producing a bright and a dark line. 5.2.16 covers fat suppression.

Truncation produces alternating bright and dark lines at a high contrast edge, which is the mechanism behind a line inside a meniscus. 5.6.2 covers meniscal signal grading.

Pulsation from a vessel propagates across the image, which can cross a ligament. 5.7.1 covers the normal cruciate.

Where It Misleads

A truncation line inside a meniscus is read as a tear, and it is the one artefact in this list that imitates the commonest knee diagnosis. 5.6.2 covers meniscal signal grading.

A pulsation artefact crossing the cruciate is read as a tear. 5.7.3 covers complete cruciate tears.

Metal signal loss is read as absent tissue. 5.5.3 covers the post-operative protocol.

Motion ghosting is read as a second structure. 5.5.1 covers the standard protocol.

Chemical shift at a cartilage surface is read as a chondral lesion. 5.11.2 covers chondral defects.

MRCS Level

How It Is Acquired

In a viva the expected answer names the mechanism and the appearance together, because the mechanism is what makes the appearance predictable.

The expected points are motion, metal, chemical shift, truncation, and pulsation.

Say that a finding on one sequence or one plane only should raise artefact. 5.5.1 covers the standard protocol.

Say that an artefact's position relates to the acquisition rather than to anatomy. 5.2.12 covers the basic principles.

Say that changing the phase direction moves an artefact and not a lesion. 5.2.18 covers gradient echo imaging.

How Good It Is

Artefact recognition is the difference between a reliable reader and an unreliable one, more than any other single skill in this section.

Its best contribution is negative: it prevents a report rather than producing one. 5.20.8 covers common reporting errors.

It is limited by depending on the reader knowing the mechanism, since no gate or checklist detects an artefact. 5.1.6 covers imaging pitfalls.

No figures appear for accuracy on this page, under decision 028, and no displacement or band width values appear either: those depend on field strength and bandwidth and this environment cannot check them.

Where It Misleads

The commonest failure is not considering artefact at all when a finding sits on one sequence. 5.5.1 covers the standard protocol.

An artefact reported once is then confirmed by the next reader, because the report anchors them. 5.20.8 covers common reporting errors.

A repeat sequence with a changed parameter would settle it and is rarely requested. 5.2.18 covers gradient echo imaging.

A technically poor study is reported rather than repeated. 5.5.1 covers the standard protocol.

An artefact in a post-operative knee is attributed to the surgery. 5.5.3 covers the post-operative protocol.

FRCS Level

How Good It Is

At decision level artefact recognition decides whether a finding should change an operation.

A meniscal line present on one sequence and absent on the others is not a tear, and operating on it is the error this page prevents. 5.6.2 covers meniscal signal grading.

A cruciate crossed by a pulsation artefact in a stable knee is intact, and the examination settles it. 5.7.1 covers the normal cruciate.

Signal loss around fixation is not tissue loss, which matters in revision planning. 5.7.12 covers tunnel assessment.

Any finding that will change an operation is worth reviewing on the images rather than from the report. 5.20.1 covers structured reporting.

Where It Misleads

The commonest error is accepting a single sequence finding as real. 5.5.1 covers the standard protocol.

Truncation is operated on as a meniscal tear. 5.6.2 covers meniscal signal grading.

Metal distortion is read as bone loss before a revision. 5.7.14 covers revision cruciate imaging.

Motion artefact in a painful knee is read as pathology rather than as a reason to repeat. 5.5.1 covers the standard protocol.

An artefact is recognised and not stated, so the next reader repeats the error. 5.20.8 covers common reporting errors.

Fellowship Level

Where It Misleads

At fellowship level artefact is the part of imaging knowledge that depreciates fastest and is taught least.

Each artefact belongs to a sequence and a field strength, so the set a reader learned on one generation of scanner is not the set they meet on the next. 5.2.12 covers the basic principles.

Three dimensional isotropic acquisition and metal artefact reduction both change which artefacts appear, and the literature describing artefacts predates both. 5.2.18 covers gradient echo imaging.

An artefact is never recorded as such in a structured report, so there is no record of how often one is mistaken for pathology. 5.20.1 covers structured reporting.

Automated reading has no artefact concept unless it was trained on labelled artefacts, and the labelled sets are small, so a model confidently reports what a reader would dismiss. 5.18.8 covers artificial intelligence in imaging.

The conclusion the whole chapter has been building towards is that an imaging appearance is a product of the tissue and the instrument together, and this is the page where that is plainly true. 5.2.10 covers anisotropy.