Discover the 18 most common MRI artifacts that compromise diagnostic accuracy, from motion ghosting and chemical shift to susceptibility blooming and dielectric effects, with proven remedies and real clinical examples for radiology teams.

18 Types of MRI Artifacts: Causes, Remedies & Clinical Examples Every Radiologist Must Know

⏱️ 24 min read 📁 Diagnostic Imaging ✓ Medically Reviewed

At a glance

• MRI artifacts arise from patient motion, magnetic field inhomogeneity, RF interference, gradient non-linearity, and sampling limitations
• Artifacts are classified as motion-related (ghosting), physics-related (chemical shift, susceptibility), hardware-related (RF inhomogeneity, zipper), and sampling-related (Gibbs, aliasing)
• Susceptibility-weighted imaging (SWI) deliberately exploits susceptibility artifacts to detect microhaemorrhages and cavernomas
• Modern parallel imaging, motion correction, and advanced shimming have significantly reduced artifact burden
• This guide covers 18 distinct artifact types with visual examples and evidence-based remedies for each

Introduction to MRI artifacts

Magnetic resonance imaging generates images through a complex interplay of static magnetic fields, radiofrequency pulses, and magnetic field gradients. Unlike CT, where artifacts primarily arise from X-ray physics and detector limitations, MRI artifacts originate from a broader spectrum of sources: patient physiology, magnetic field imperfections, RF coil design, gradient performance, and Fourier transform sampling theory. This diversity makes MRI artifact recognition both more challenging and more clinically consequential.

The diagnostic impact of MRI artifacts cannot be overstated. A motion ghost in a brain FLAIR sequence can mimic vasogenic oedema. Chemical shift misregistration at the kidney-adrenal interface can create a pseudo-lesion. Susceptibility blooming from a cavernoma can exaggerate its true size by 300%. Conversely, some artifacts are deliberately exploited: susceptibility-weighted imaging (SWI) uses the magnetic field distortion around deoxyhaemoglobin to detect microbleeds invisible on conventional sequences.

This comprehensive guide examines 18 distinct MRI artifact categories, organised by their underlying physical mechanism. For each artifact type, we present: the physics of origin, characteristic imaging appearance, clinical scenarios where it commonly occurs, evidence-based remedies, and illustrative examples. The goal is to equip radiographers, radiologists, and MRI physicists with practical knowledge to optimise sequences and maintain diagnostic confidence.

1. Motion artifacts (ghosting)

Physics and mechanism

Motion artifacts are the most common and clinically significant MRI artifacts. They arise when the patient moves during data acquisition, causing phase inconsistencies between different lines of k-space. Because k-space is filled line by line over several minutes, even microscopic motion—cardiac pulsation, CSF flow, respiration, peristalsis, or voluntary movement—creates characteristic ghost images that replicate the moving anatomy at displaced positions.

The ghost appears along the phase-encoding direction because each line of k-space corresponds to a different phase-encoding gradient strength. Motion during the frequency-encoding (readout) direction has minimal effect because each line is acquired in milliseconds. The ghost intensity depends on motion amplitude and the point in k-space where it occurs: motion during central k-space lines (low spatial frequencies) creates strong, diffuse ghosts, while motion during peripheral lines (high spatial frequencies) creates weaker, sharp ghosts.

Clinical examples

Motion in brain MRI creates ghost images of the scalp and skull that overlay the cerebral parenchyma, potentially obscuring infarcts or tumours. In spine MRI, CSF pulsation causes flow-related ghosts that mimic intramedullary lesions. In abdominal MRI, respiratory motion degrades liver lesion characterisation. In cardiac MRI, cardiac motion without gating produces uninterpretable images.

Remedies and solutions

• Patient instruction and immobilisation: The single most effective intervention. Earplugs, headphones, and padding reduce voluntary motion.
• Navigator sequences: Prospective Acquisition Correction (PACE) tracks diaphragm or head position and adjusts slice position in real-time.
• Fast sequences: Single-shot TSE/HASTE, EPI, and gradient-echo sequences reduce acquisition time to sub-second intervals.
• Cardiac gating: ECG-triggering synchronises acquisition to the cardiac cycle for cardiac and thoracic imaging.
• Respiratory gating/navigators: For abdominal MRI, respiratory bellows or navigator echoes trigger acquisition during end-expiration.
• Saturation bands: Presaturation pulses null signal from moving tissues (e.g., subcutaneous fat).
• Phase-encoding direction: Swap phase and frequency directions to move ghosts away from region of interest.
• Motion correction algorithms: Post-processing tools like MOtion correction on INterpolation (MOTION) and data-driven motion correction.

2. Chemical shift artifacts

Physics and mechanism

Chemical shift arises because hydrogen protons in fat and water resonate at slightly different frequencies due to differences in their local magnetic environment (electron shielding). At 1.5 T, the frequency difference is approximately 3.5 ppm, or 224 Hz. The MRI system assumes all protons resonate at the water frequency, so fat protons are misregistered by a few pixels along the frequency-encoding direction.

This misregistration creates two effects: a bright band (signal overlap) on one side of the fat-water interface and a dark band (signal void) on the opposite side. The displacement in pixels equals the chemical shift frequency divided by the bandwidth per pixel. Thus, low bandwidth sequences (conventional spin echo) show more severe chemical shift than high bandwidth sequences (gradient echo with wide bandwidth).

Clinical examples

Chemical shift is most visible at the kidney-adrenal interface, where perinephric fat creates a dark band that can mimic a subcapsular haematoma. In orbital MRI, the fat-water interface around the globe creates artifacts that can obscure optic nerve pathology. In ovarian imaging, dermoid cysts with fat content show pronounced chemical shift.

Remedies and solutions

• Increase receiver bandwidth: Higher bandwidth reduces chemical shift displacement (at the cost of SNR).
• Fat suppression: STIR, SPIR, SPAIR, and Dixon techniques eliminate fat signal, removing the source of the artifact.
• Dixon method: Separates water and fat into distinct images, completely eliminating chemical shift.
• Swap phase/frequency direction: Move the artifact away from the region of interest.
• Gradient-echo with opposed-phase: Deliberately exploits chemical shift to detect microscopic fat (in-phase/out-of-phase imaging).

3. Magnetic susceptibility artifacts

Physics and mechanism

Magnetic susceptibility describes how a material becomes magnetised in an external magnetic field. Tissues with different susceptibilities (air, bone, haemorrhage, metal) create local magnetic field gradients that distort the main field homogeneity. This causes rapid signal dephasing, particularly in gradient-echo (GRE) sequences where refocusing pulses are absent.

The artifact manifests as a signal void (black area) surrounded by a bright rim (blooming). The severity increases with: higher field strength (3T > 1.5T), longer TE, and smaller voxel size. While usually problematic, susceptibility effects are the basis of SWI (susceptibility-weighted imaging), which exploits deoxyhaemoglobin and hemosiderin to detect microbleeds, cavernomas, and venous structures.

Clinical examples

In brain MRI, susceptibility artifacts from the skull base, paranasal sinuses, and mastoid air cells obscure the temporal lobes and brainstem. In spine MRI, metallic fixation hardware creates massive artifacts. In abdominal MRI, gas in the bowel obscures adjacent structures. Dental amalgam and tattoos with ferromagnetic pigments cause severe facial artifacts.

Remedies and solutions

• Spin-echo sequences: 180° refocusing pulses compensate for local field inhomogeneity, dramatically reducing susceptibility artifacts compared to GRE.
• Fast spin-echo (TSE): Multiple 180° pulses provide even better compensation than conventional SE.
• Lower field strength: 1.5T produces less susceptibility distortion than 3T.
• Shorter TE: Reduces time for dephasing.
• Larger voxel size: Partial volume averaging reduces apparent artifact severity.
• Shimming: Active and passive shimming improve field homogeneity around challenging anatomy.
• View Angle Tilting (VAT): Specialised technique for metal artifact reduction in MRI.
• SEMAC/MAVRIC: Advanced techniques for imaging near orthopaedic implants.

4. Gibbs (truncation) artifacts

Physics and mechanism

Gibbs artifacts, also called truncation artifacts, arise from the Fourier transform’s inability to perfectly represent sharp edges with a finite number of frequency components. When k-space is truncated (sampled with finite matrix size), the reconstruction produces ringing oscillations near high-contrast boundaries. The artifact appears as alternating bright and dark lines parallel to sharp edges.

The severity is inversely proportional to the matrix size: a 128×128 matrix produces more severe Gibbs artifacts than 256×256 or 512×512. The artifact is most problematic where there are sharp, high-contrast boundaries: spinal cord-CSF, brain-skull, and tendon-muscle interfaces.

Clinical examples

In spine MRI, Gibbs artifacts at the spinal cord-CSF interface can mimic a syringomyelia (pseudo-syrinx). In brain MRI, ringing near the cortex can simulate cortical dysplasia. In musculoskeletal MRI, Gibbs artifacts near tendons can be mistaken for partial tears.

Remedies and solutions

• Increase matrix size: 256×256 or 512×512 matrices significantly reduce Gibbs artifacts.
• Zero-filling interpolation: Increases apparent matrix size without additional scan time.
• Apodisation filters: Window functions (Hamming, Hanning) smooth k-space edges to reduce ringing.
• Phase-encoding direction: Orient the direction of sharp edges along frequency encoding.
• 3D acquisitions: Higher through-plane resolution reduces truncation in the slice direction.

5. Aliasing and wrap-around artifacts

Physics and mechanism

Aliasing (wrap-around) occurs when anatomy outside the field-of-view (FOV) is misregistered inside the image. In the phase-encoding direction, the system assumes a periodic signal. If anatomy extends beyond the FOV, its signal is incorrectly assigned to the opposite side of the image, creating a “wrapped” appearance. In the frequency-encoding direction, aliasing can occur if the receiver bandwidth is insufficient to separate the true signal from folded frequencies.

Clinical examples

In brain MRI, the nose and scalp may wrap into the posterior fossa. In spine MRI, the abdomen wraps into the spinal canal. In cardiac MRI, the chest wall wraps across the myocardium.

Remedies and solutions

• Phase oversampling: Acquire extra phase-encoding lines beyond the FOV (typically 50-100% oversampling).
• No Phase Wrap (NPW): Vendor-specific option that automatically applies oversampling.
• Frequency oversampling: Increase receiver bandwidth and sampling rate.
• Spatial presaturation: Null signal from anatomy outside the FOV.
• Surface coils: Limited field-of-view coils naturally suppress signal from distant anatomy.

6. Partial volume effects

Physics and mechanism

Like CT, MRI suffers from partial volume averaging when a voxel contains multiple tissue types. However, MRI partial volume is more complex because the averaged signal depends on T1, T2, and proton density weighting. A voxel containing CSF and cord parenchyma will have different signal characteristics depending on sequence weighting.

Clinical examples

In orbital MRI, thin extraocular muscles may be partially volume-averaged with orbital fat, reducing apparent signal. In vessel imaging, small vessels may appear to have laminar flow artifacts due to partial volume.

Remedies and solutions

• Thin slices: Reduce slice thickness to minimise averaging.
• 3D acquisitions: Isotropic voxels allow reformatting in any plane without partial volume.
• Oblique imaging: Align slices perpendicular to structures of interest.

7. Flow artifacts

Physics and mechanism

Flow artifacts occur because moving protons experience different magnetic field environments as they travel through the slice. In time-of-flight (TOF) imaging, inflow enhancement creates bright signal. In conventional spin echo, flowing blood appears dark (flow void) because excited protons move out of the slice before the echo is read. In gradient echo, flowing blood typically appears bright due to inflow enhancement.

Flow-related artifacts include: flow voids (signal loss), flow enhancement (unexpected bright signal), phase dispersion (signal loss in turbulent flow), and CSF pulsation artifacts (ghosting from pulsatile CSF motion).

Clinical examples

In brain MRI, flow voids in arteries are normal but can be confused with thrombosis if asymmetrical. In spine MRI, CSF pulsation creates flow artifacts in the spinal canal that mimic intramedullary cysts or syrinx. In MR angiography, turbulent flow causes signal loss that mimics stenosis.

Remedies and solutions

• Flow compensation (gradient moment nulling): Additional gradients correct for velocity-induced phase shifts.
• Presaturation bands: Null inflowing blood signal upstream of the slice.
• Cardiac gating: Synchronise acquisition to specific cardiac phases for vascular imaging.
• Spatial presaturation: Saturate signal from vessels outside the region of interest.
• 3D TOF MRA: Minimises flow-related artifacts by using thin slab excitation.

8. Cross-talk artifacts

Physics and mechanism

Cross-talk (slice interference) occurs when adjacent slices are excited simultaneously or in rapid succession, causing partial saturation of protons at slice edges. The RF pulse profile is not perfectly rectangular—there is a transition zone where adjacent slices overlap. Protons in this zone experience multiple excitations, reducing their available magnetisation and creating signal loss at slice interfaces.

Clinical examples

Cross-talk is most visible in spine MRI where multiple contiguous slices are acquired. Signal loss at vertebral endplates can mimic disc desiccation or Schmorl’s nodes.

Remedies and solutions

• Interleaved acquisition: Acquire odd-numbered slices first, then even-numbered slices, allowing time for magnetisation recovery.
• Gap between slices: A small gap (10-20% of slice thickness) reduces cross-talk but risks missing pathology.
• 3D acquisitions: Eliminate cross-talk by using volume excitation.
• Optimized RF pulse design: Sharper slice profiles reduce transition zones.

9. RF inhomogeneity artifacts

Physics and mechanism

RF inhomogeneity occurs when the transmit or receive RF field is not uniform across the imaging volume. At 3 T and higher field strengths, the RF wavelength in tissue becomes comparable to body dimensions (26 cm at 3T, 11 cm at 7T), creating standing wave patterns and dielectric resonance. This causes signal drop-out in the image centre and bright signal at the periphery.

Receive coil inhomogeneity also contributes: surface coils produce higher signal near the coil and lower signal at depth. Modern scanners use parallel imaging with multiple coil elements and intensity correction algorithms to compensate.

Clinical examples

In 3T abdominal MRI, RF inhomogeneity causes dark signal in the liver centre and bright subcutaneous fat. In 3T brain MRI, temporal lobes may show reduced signal. In body MRI at 7T, the effect is severe enough to limit clinical utility.

Remedies and solutions

• Dielectric pads: High-permittivity pads placed on the patient improve RF field uniformity at 3T.
• Multi-transmit coils: Independent RF channels with phase adjustment optimise field distribution.
• Intensity correction: Post-processing algorithms (N3, N4) normalise signal across the image.
• Body coils: Volume coils provide more uniform excitation than surface coils.
• Parallel imaging calibration: Proper coil sensitivity calibration improves reconstruction uniformity.

10. Zipper artifacts

Physics and mechanism

Zipper artifacts appear as horizontal or diagonal lines of noise across the image. They are caused by external RF interference from sources such as: unshielded electrical equipment, fluorescent lights, elevator motors, mobile phones, and nearby radio transmitters. The interference frequency matches the Larmor frequency (63.9 MHz at 1.5T, 127.8 MHz at 3T), creating a coherent noise spike in k-space that reconstructs as a line.

Clinical examples

A zipper across a brain MRI can obscure cortical lesions. Multiple zippers create a “bar code” appearance that renders the image non-diagnostic.

Remedies and solutions

• RF shielding: Ensure Faraday cage integrity (door seals, waveguides, penetrations).
• Identify interference source: Systematically turn off equipment to isolate the culprit.
• Filter power lines: Install RF filters on all lines entering the scan room.
• Scan at different times: Avoid periods when interfering equipment is active.
• Vendor service: Persistent zippers may require RF system inspection.

11. Magic angle artifacts

Physics and mechanism

The magic angle phenomenon occurs when collagen fibres are oriented at approximately 54.7° (the magic angle) relative to the static magnetic field B₀. At this angle, dipolar interactions between water protons and collagen are minimised, causing increased T2 signal that can mimic pathology. The effect is most pronounced with short TE sequences (<20 ms) and diminishes with longer TE.

Clinical examples

In shoulder MRI, the supraspinatus tendon normally appears dark on all sequences. When the arm is positioned such that tendon fibres approach the magic angle, the tendon shows increased signal that mimics tendinosis or partial tear. In ankle MRI, the Achilles tendon and plantar fascia can show similar pseudo-pathology.

Remedies and solutions

• Longer TE sequences: T2-weighted or PD-weighted with TE >30 ms reduces magic angle effect.
• Alter patient position: Change arm/tendon orientation relative to B₀.
• Clinical correlation: Recognise that increased signal without morphologic change likely represents magic angle.
• STIR sequences: Magic angle effect is less pronounced on fat-suppressed sequences.

12. Dielectric effect artifacts

Physics and mechanism

At 3 T and above, the RF wavelength in tissue becomes short enough to create standing wave patterns within the body. The dielectric properties of tissue (permittivity and conductivity) cause the RF field to resonate, creating regions of constructive interference (bright signal) and destructive interference (dark signal). This is particularly problematic in large body regions like the abdomen and pelvis.

Clinical examples

In 3T abdominal MRI, the liver centre often appears dark while the periphery is bright, potentially obscuring central lesions. In 3T cardiac MRI, the left ventricular cavity may show non-uniform signal.

Remedies and solutions

• Dielectric pads: High-permittivity pads (e.g., water bags, ceramic pads) placed on the patient alter the RF field distribution.
• Multi-transmit technology: Independent RF channel control optimises field uniformity.
• RF shimming: Adjust transmit phases and amplitudes to minimise standing waves.
• Lower field strength: 1.5T is less affected by dielectric effects.

13. Truncation artifacts

Physics and mechanism

Distinct from Gibbs artifacts, truncation artifacts in MRI occur when the acquisition matrix is too small to represent the object, causing signal from outside the FOV to fold into the image. This is particularly problematic with rectangular FOV acquisitions where the phase-encoding direction is reduced to save time.

Clinical examples

In spine MRI with rectangular FOV, the abdomen may wrap into the spinal canal. In extremity MRI, the contralateral limb may alias into the image.

Remedies and solutions

• Oversampling: Acquire extra phase-encoding lines.
• Saturation bands: Null signal from outside the FOV.
• Surface coils: Limited sensitivity reduces aliasing.

14. Moire fringes

Physics and mechanism

Moire fringes appear as wavy, zebra-like patterns superimposed on the image. They result from interference between the sampling grid and fine periodic structures in the object, such as trabecular bone, mesh implants, or woven fabrics on clothing. The artifact is analogous to the moire patterns seen in digital photography.

Clinical examples

Moire patterns appear in wrist MRI over carpal bones and in spine MRI over trabecular vertebral bodies. Surgical mesh can create pronounced moire artifacts.

Remedies and solutions

• Increase matrix size: Finer sampling reduces moire.
• Change slice orientation: Avoid aligning slices with periodic structures.
• Remove metallic fabrics: Ensure patients are not wearing clothing with metallic threads.

15. Random noise

Physics and mechanism

Random noise in MRI arises from thermal noise in the patient (Brownian motion of ions) and electronic noise in the receiver chain. The signal-to-noise ratio (SNR) depends on: field strength, coil quality, voxel volume, number of averages, and receiver bandwidth. SNR increases with voxel volume and √NEX (number of excitations).

Clinical examples

Noise limits the detection of small lesions, subtle enhancement, and fine anatomical detail. In high-resolution imaging (inner ear, cranial nerves), noise can obscure critical structures.

Remedies and solutions

• Larger voxels: Increase slice thickness or reduce matrix size (at the cost of resolution).
• More averages (NEX): Doubling NEX improves SNR by √2 but doubles scan time.
• Lower bandwidth: Narrower bandwidth improves SNR but increases chemical shift.
• Better coils: Phased-array coils with more elements improve SNR.
• Higher field strength: 3T provides approximately 2× SNR of 1.5T.
• Parallel imaging optimisation: Use the minimum acceleration factor necessary.

16. N/2 ghost artifacts

Physics and mechanism

N/2 ghosts are specific to echo-planar imaging (EPI) sequences. In EPI, k-space is filled with a single RF excitation using rapidly switched gradients. Alternate lines of k-space are acquired with opposite gradient polarity (positive and negative readout). If there are eddy currents, gradient timing errors, or phase inconsistencies between even and odd lines, a ghost image appears displaced by N/2 pixels (half the matrix size) from the true image.

Clinical examples

N/2 ghosts in diffusion-weighted imaging (DWI) can obscure acute infarcts. In fMRI, ghosts reduce activation detection sensitivity. In cardiac EPI, ghosts from the chest wall overlay the myocardium.

Remedies and solutions

• Gradient calibration: Regular service ensures gradient linearity and timing accuracy.
• Phase correction: Software algorithms correct for even-odd line phase differences.
• Navigator echoes: Measure and correct phase errors before image acquisition.
• Parallel imaging: Reduced EPI train length decreases ghost severity.

17. Chemical misregistration

Physics and mechanism

Chemical misregistration is closely related to chemical shift but specifically refers to the spatial misalignment of fat and water images in sequences that acquire them separately, such as Dixon methods and fat-suppressed sequences. If the field-of-view or matrix differs between water and fat acquisitions, the images do not align perfectly, creating edge artifacts at fat-water boundaries.

Clinical examples

In Dixon-based fat suppression, misregistration at the orbital fat-globe interface can create pseudo-lesions. In breast MRI, chemical misregistration at the chest wall can simulate enhancing lesions.

Remedies and solutions

• Consistent acquisition parameters: Ensure identical FOV, matrix, and slice position for all echoes.
• Advanced Dixon algorithms: Modern multi-point Dixon methods use phase correction to minimise misregistration.
• Field map correction: Acquire B₀ field maps to correct for spatial distortion.

Advanced artifact reduction techniques

Deep learning and artificial intelligence approaches

The application of deep learning to MRI artifact reduction represents one of the most rapidly evolving areas in radiology physics. Convolutional neural networks (CNNs) trained on pairs of artifact-corrupted and artifact-free images can learn to predict and remove artifacts without explicit knowledge of the underlying physics. Generative adversarial networks (GANs) have shown particular promise for motion artifact correction, with some studies reporting subjective quality improvements equivalent to a 50% reduction in scan time.

In motion correction, deep learning approaches analyse k-space data to detect and compensate for patient movement in real-time. Deep Resolve (Siemens) and Compressed SENSE AI (Philips) are commercial implementations that use neural networks to reconstruct high-quality images from undersampled data, effectively reducing scan time while maintaining diagnostic quality. These systems are particularly valuable in paediatric MRI, where motion is the primary cause of non-diagnostic studies.

For metal artifact reduction in MRI, deep learning models can predict the signal that would have been present in the absence of metal, effectively “filling in” the signal void. This approach complements traditional methods like SEMAC (Slice Encoding for Metal Artifact Correction) and MAVRIC (Multi-Acquisition Variable-Resonance Image Combination) by providing more accurate signal estimation in regions where traditional methods struggle.

Quantitative susceptibility mapping (QSM)

Quantitative susceptibility mapping represents a paradigm shift in how radiologists approach susceptibility artifacts. Rather than treating susceptibility as a problem to be eliminated, QSM uses sophisticated post-processing to calculate the underlying magnetic susceptibility distribution from phase images. This enables quantitative measurement of tissue iron content, calcification, and blood products.

Clinical applications of QSM include: differentiation of haemorrhage from calcification (both appear dark on SWI but have opposite susceptibility values), quantification of brain iron deposition in Parkinson’s disease and multiple sclerosis, and assessment of venous oxygen saturation in stroke and tumour imaging. The technique requires careful handling of phase unwrapping and background field removal, but when properly implemented, it transforms susceptibility from an artifact into a quantitative biomarker.

Simultaneous multi-slice (SMS) imaging

Simultaneous multi-slice imaging, also known as multiband imaging, excites multiple slices simultaneously using tailored RF pulses and separates them using coil sensitivity profiles. This technique can reduce scan time by factors of 2-8, dramatically reducing motion sensitivity. However, SMS introduces its own artifacts: slice leakage (signal from one slice appearing in another) and g-factor penalty (increased noise amplification).

Modern SMS implementations use blipped CAIPIRINHA (Controlled Aliasing in Parallel Imaging Results in Higher Acceleration) to control the aliasing pattern, placing leaked signal at the image periphery where it is less clinically relevant. For fMRI, SMS has become the standard approach, enabling whole-brain coverage in under one second. For clinical anatomical imaging, SMS is increasingly used in T2-weighted and diffusion-weighted sequences.

Artifacts that mimic pathology: a differential diagnosis

Perhaps the greatest clinical danger of MRI artifacts is their ability to mimic genuine pathology, leading to misdiagnosis, unnecessary procedures, and patient harm. The following table summarises the most common artifact-mimic pairs and how to distinguish them:

Artifact	Mimics	Differentiating Features
Motion ghost	Vasogenic oedema, infarct	Ghost repeats anatomy; changes with phase direction swap; disappears on fast sequences
Chemical shift	Subcapsular haematoma, adrenal mass	Located at fat-water interface; bright and dark bands; disappears with fat suppression
Gibbs pseudo-syrinx	Syringomyelia, intramedullary cyst	Parallel to cord edge; disappears with higher matrix; no true fluid signal on T2
Magic angle	Supraspinatus tendinosis, partial tear	No morphologic abnormality; disappears with TE >30 ms; changes with arm position
Flow void	Venous sinus thrombosis	Normal flow voids are symmetrical; thrombosis shows expanded sinus with abnormal signal
Susceptibility blooming	Large haemorrhage, tumour	Size exaggeration on GRE/SWI; true size on SE/TSE; blooming increases with TE
Cross-talk	Disc desiccation, Schmorl’s node	At slice edges only; disappears with interleaved acquisition; no morphologic defect
RF inhomogeneity	Focal liver lesion, infarct	Follows coil sensitivity profile; improves with intensity correction; no mass effect

Artifact-aware protocol design

Field strength selection

The choice between 1.5 T and 3 T involves trade-offs that directly affect artifact burden. At 3 T, SNR doubles, enabling higher resolution or faster scanning. However, susceptibility artifacts worsen by a factor of 2, chemical shift doubles, and RF inhomogeneity becomes significant. For patients with metal implants, 1.5 T is often preferable. For high-resolution brain imaging, 3 T is superior when susceptibility is not the primary concern.

Sequence selection hierarchy

When artifact is anticipated, follow this hierarchy: (1) Use spin-echo or fast spin-echo instead of gradient-echo to reduce susceptibility. (2) Use fat suppression (STIR, Dixon) to eliminate chemical shift. (3) Use fast or single-shot sequences to reduce motion. (4) Use cardiac or respiratory gating for pulsatile or respiratory motion. (5) Use saturation bands to suppress unwanted signal.

Parameter optimisation

Bandwidth selection is a critical but often overlooked parameter. Higher bandwidth reduces chemical shift and susceptibility artifacts but decreases SNR. A bandwidth of 200-300 Hz/pixel is typical for T1-weighted imaging, while 500-800 Hz/pixel may be needed for regions with significant fat-water interfaces. Matrix size should be sufficient to resolve the structures of interest while avoiding excessive scan time. For brain imaging, 256×256 is usually adequate; for high-resolution musculoskeletal imaging, 384×384 or 512×512 may be necessary.

Coil selection and positioning

Coil choice significantly affects artifact burden. Surface coils provide high SNR but limited FOV and non-uniform sensitivity. Volume coils provide uniform excitation but lower SNR. Phased-array coils combine the benefits of both but require careful sensitivity calibration. For abdominal imaging, anterior-posterior phased arrays are standard. For brain imaging, 32-channel or 64-channel head coils enable high parallel imaging factors with minimal g-factor penalty.

MRI quality assurance for artifact prevention

A comprehensive QA programme is essential for maintaining artifact-free imaging. The following schedule represents best practice:

Daily checks

• RF noise monitoring: Run a noise floor scan before first patient to detect zipper artifacts from new interference sources.
• Phantom imaging: Acquire ACR phantom images daily to verify image uniformity, spatial resolution, and geometric accuracy.
• Coil functionality: Verify all coil elements are operational using built-in diagnostics.

Weekly checks

• Shim verification: Measure field homogeneity across the phantom and patient volumes.
• Gradient linearity: Verify geometric accuracy using grid phantoms.
• SAR monitoring: Confirm specific absorption rate calculations are within safe limits.

Monthly checks

• Comprehensive phantom analysis: Evaluate SNR, uniformity, ghosting, and slice profile using standardised phantoms.
• Coil sensitivity calibration: Update parallel imaging calibration data for all coil combinations.
• RF shielding integrity: Verify door seals, waveguide integrity, and penetration panel condition.

Annual checks

• Full system calibration: Manufacturer service visit for comprehensive hardware verification.
• Magnet stability: Monitor cryogen levels, helium pressure, and field drift.
• Software updates: Evaluate and install vendor software updates that may include new artifact correction algorithms.

Artifact considerations in special populations

Paediatric MRI

Motion artifacts dominate paediatric MRI, with up to 30% of examinations requiring sedation or general anaesthesia in children under 6 years. Protocol optimisation strategies include: feed-and-wrap techniques for neonates, propofol sedation protocols for toddlers, and child-friendly environments (decoration, music, video goggles) for cooperative older children. Fast sequences like HASTE and SSFSE enable sub-second acquisitions that “freeze” motion. PROPELLER and BLADE sequences acquire k-space in rotating blades, making them inherently motion-insensitive.

Emergency MRI

In the emergency setting, scan time minimisation is paramount. Protocols should prioritise fast, motion-robust sequences: DWI for stroke, GRE/SWI for haemorrhage, and FLAIR for general parenchymal assessment. Susceptibility artifacts from emergency airway equipment, cervical collars, and monitoring devices must be anticipated. Non-ferrous alternatives should be used whenever possible, and sequences should be selected to minimise susceptibility (TSE over GRE when metal is present).

Pregnancy

While MRI has no ionising radiation and is considered safe in pregnancy, artifact management is complicated by physiological changes. Fetal motion is rapid and unpredictable, requiring single-shot sequences. Maternal respiration and cardiac output changes affect abdominal imaging. Amniotic fluid creates susceptibility artifacts at the maternal-fetal interface. Protocols should use SSFSE for fetal imaging and navigator-gated sequences for maternal abdominal imaging.

Implanted devices

Patients with cardiac pacemakers, cochlear implants, neurostimulators, and orthopaedic implants present unique artifact challenges. MR-conditional devices allow scanning under specific conditions, but the device itself creates susceptibility artifacts. SEMAC and MAVRIC sequences are specifically designed for imaging near implants. View Angle Tilting (VAT) reduces in-plane distortion. For non-conditional devices, risk-benefit analysis must consider both safety and diagnostic utility in the presence of severe artifacts.

Future directions in artifact management

The next decade promises transformative advances in MRI artifact reduction. Artificial intelligence will enable real-time motion prediction and correction before artifacts form. Self-gating techniques will eliminate the need for external monitoring devices. Ultra-high field systems (7 T and above) will require new approaches to dielectric effects, potentially using travelling wave excitation or metamaterial structures to control RF field distribution.

Quantitative MRI techniques that measure T1, T2, and proton density directly will be less susceptible to artifacts that affect conventional weighted imaging. Synthetic MRI generates multiple contrast-weighted images from a single quantitative acquisition, reducing the opportunity for motion between sequences. Compressed sensing and deep learning reconstruction will enable high-quality imaging from highly undersampled data, reducing scan time and motion sensitivity simultaneously.

For radiology departments, the key to staying ahead of artifact challenges is continuous education, investment in QA infrastructure, and close collaboration between radiologists, radiographers, physicists, and vendor engineers. The radiologist who understands artifact physics will always deliver more reliable diagnoses than one who relies solely on image interpretation.

Summary of MRI artifacts and remedies

Artifact	Primary Cause	Key Remedy	Clinical Priority
Motion (ghosting)	Patient movement	Fast sequences, navigators, immobilisation	Critical
Chemical shift	Fat-water frequency difference	Fat suppression, Dixon, high bandwidth	High
Susceptibility	Field inhomogeneity	Spin echo, lower field, shimming	High
Gibbs (truncation)	Finite k-space sampling	Higher matrix, apodisation	Medium
Aliasing (wrap)	FOV undersampling	Phase oversampling, saturation	High
Partial volume	Voxel size limitation	Thin slices, 3D acquisition	Medium
Flow	Moving proton phase shifts	Flow compensation, gating	High
Cross-talk	Slice overlap excitation	Interleaved acquisition, gap	Low
RF inhomogeneity	Non-uniform B₁ field	Dielectric pads, multi-transmit	High
Zipper	External RF interference	RF shielding, interference hunt	Medium
Magic angle	Collagen fibre orientation	Longer TE, repositioning	Medium
Dielectric effect	Standing RF waves	Dielectric pads, multi-transmit	High (3T+)
N/2 ghost	EPI gradient asymmetry	Gradient calibration, phase correction	Medium
Random noise	Thermal/electronic noise	Larger voxels, more averages, better coils	Medium
Chemical misregistration	Fat-water image misalignment	Consistent parameters, field map	Low

Further reading

1. SATLine: Advanced MRI Quality Assurance Protocols — SATMED Health
2. SATDrape: Patient Positioning and Immobilisation Systems — SATMED Health
3. SATPro: 3T and 7T MRI Sequence Optimisation — SATMED Health
4. SATSurgical: MRI Protocols for Orthopaedic Implant Imaging — SATMED Health
5. SATJect: Contrast Protocol Optimisation for MRI — SATMED Health

Conclusion

Magnetic resonance imaging artifacts represent a diverse and often subtle challenge that demands deep understanding of physics, sequence design, and clinical context. This guide has examined 18 distinct MRI artifact categories, from the ubiquitous motion ghosts that plague every MRI department to the esoteric magic angle phenomenon that can mislead even experienced musculoskeletal radiologists.

The management of MRI artifacts follows a hierarchical approach: prevention through optimised sequence parameters and patient preparation; recognition through radiologist education and pattern familiarity; and correction through advanced hardware and software solutions. Unlike CT, where artifacts often require repeat scanning with adjusted parameters, many MRI artifacts can be mitigated within the same examination by switching sequences or adjusting acquisition geometry.

Modern MRI technology has dramatically expanded the radiologist’s arsenal against artifacts. Parallel imaging reduces scan times and motion sensitivity. Multi-transmit coils tame RF inhomogeneity at 3T. Advanced shimming algorithms automatically optimise field homogeneity. Motion correction techniques track and compensate for patient movement in real-time. Dixon methods elegantly separate fat and water while eliminating chemical shift. Yet the fundamental physics—Larmor precession, Fourier sampling, and magnetic susceptibility—remain the foundation upon which all solutions are built.

For radiographers, understanding artifact mechanisms empowers real-time decision-making at the scanner console: when to swap phase and frequency directions, when to add saturation bands, when to switch from GRE to TSE. For radiologists, artifact recognition prevents diagnostic errors that can lead to unnecessary biopsies, missed malignancies, or incorrect treatment. For MRI physicists and hospital administrators, systematic QA programmes and continuing education ensure that the substantial investment in MRI technology yields maximum diagnostic return.

As MRI continues to evolve toward higher field strengths (7T clinical systems are now FDA-approved), ultra-fast sequences, and AI-assisted reconstruction, the artifact landscape will transform. New challenges—such as the severe dielectric effects and RF power deposition at 7T—will emerge alongside new solutions. The radiology professional who masters the principles outlined in this guide will be well-equipped to navigate these advances while maintaining the highest standards of diagnostic quality.

References

1. Haacke EM, Brown RW, Thompson MR, Venkatesan R. Magnetic resonance imaging: physical principles and sequence design. 2nd ed. New York: Wiley-Liss; 1999.
2. Stark DD, Bradley WG Jr. Magnetic resonance imaging. 3rd ed. St. Louis: Mosby; 1999. ISBN 9780815185185.
3. McRobbie DW, Moore EA, Graves MJ, Prince MR. MRI from picture to proton. 2nd ed. Cambridge: Cambridge University Press; 2007.
4. Hashemi RH, Bradley WG Jr, Lisanti CJ. MRI: the basics. 4th ed. Philadelphia: Wolters Kluwer; 2018.
5. Bernstein MA, King KF, Zhou XJ. Handbook of MRI pulse sequences. Amsterdam: Elsevier Academic Press; 2004.
6. Plewes DB, Kucharczyk W. Physics of MRI: a primer. J Magn Reson Imaging. 2012;35(5):1038-1054.
7. Robson MD, Gatehouse PD, Bydder M, Bydder GM. Magnetic resonance: an introduction to ultrashort TE (UTE) imaging. J Comput Assist Tomogr. 2003;27(6):825-846.
8. Reeder SB, Pineda AR, Wen Z, et al. Iterative decomposition of water and fat with echo asymmetry and least-squares estimation (IDEAL): application with fast spin-echo imaging. Magn Reson Med. 2005;54(3):636-644.
9. Ma J. Breath-hold water and fat imaging using a dual-echo two-point Dixon technique with an efficient and robust field-map generation scheme. Magn Reson Med. 2004;52(2):415-419.
10. Sehgal V, Delproposto Z, Haacke EM, et al. Clinical applications of neuroimaging with susceptibility-weighted imaging. J Magn Reson Imaging. 2005;22(4):439-450.
11. Haacke EM, Mittal S, Wu Z, Neelavalli J, Cheng YC. Susceptibility-weighted imaging: technical aspects and clinical applications, part 1. AJNR Am J Neuroradiol. 2009;30(1):19-30.
12. Mittal S, Wu Z, Neelavalli J, Haacke EM. Susceptibility-weighted imaging: technical aspects and clinical applications, part 2. AJNR Am J Neuroradiol. 2009;30(2):232-252.
13. Chavhan GB, Babyn PS, Jankharia BG, Cheng HL, Shroff MM. Steady-state MR imaging sequences: physics, classification, and clinical applications. Radiographics. 2008;28(4):1147-1160.
14. Morelli JN, Runge VM, Ai F, et al. An image-based approach to understanding the physics of MR artifacts. Radiographics. 2011;31(3):849-866.
15. Zhuo J, Gullapalli RP. MR artifacts, safety, and quality control. Radiographics. 2006;26(1):275-297.
16. Kanal E, Maravilla K, Rowley HA. Gadolinium contrast agents for CNS imaging: current concepts and clinical evidence. AJNR Am J Neuroradiol. 2014;35(12):2215-2226.
17. Kuhl CK, Träber F, Gieseke J, et al. Whole-body high-spatial-resolution MR imaging with parallel imaging at 3.0 T: feasibility study. Radiology. 2008;246(3):954-964.
18. Griswold MA, Jakob PM, Heidemann RM, et al. Generalized autocalibrating partially parallel acquisitions (GRAPPA). Magn Reson Med. 2002;47(6):1202-1210.
19. Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. Parallel magnetic resonance imaging. Phys Med Biol. 2007;52(7):R15-R55.
20. Roemer PB, Edelstein WA, Hayes CE, Souza SP, Mueller OM. The NMR phased array. Magn Reson Med. 1990;16(2):192-225.
21. Foo TK, Bernstein MA, Aisen AM, et al. Improved ejection fraction and flow velocity estimates with use of view sharing and uniform repetition time excitation with fast cardiac techniques. Radiology. 1995;195(2):471-478.
22. Markl M, Chan FP, Alley MT, et al. Time-resolved three-dimensional phase-contrast MRI. J Magn Reson Imaging. 2003;17(4):499-506.
23. Gatehouse PD, Keegan J, Crowe LA, et al. Applications of phase-contrast flow and velocity imaging in cardiovascular MRI. Eur Radiol. 2005;15(10):2172-2184.
24. Bammer R. Basic principles of diffusion-weighted imaging. Eur J Radiol. 2003;45(3):169-184.