---

MRI ARTIFACTS - Complete Reference (50 Marks)

PHYSICS FOUNDATION

1. Spin excitation - RF pulse at Larmor frequency (f = γ × B₀, where γ = 42.57 MHz/T for protons)
2. Spatial encoding - gradient fields along frequency-encode (FE), phase-encode (PE), and slice-select directions
3. Signal detection - raw data fills k-space; Fourier transform converts k-space to image

---

CATEGORY 1 - MOTION-RELATED ARTIFACTS

1. Motion Artifact (Ghosting)

• MRI acquires multiple phase-encoding lines sequentially over time (each TR acquires one k-space line in conventional spin echo).
• If tissue moves between phase-encoding steps, each k-space line is acquired from a slightly different tissue position.
• This produces phase inconsistency in k-space.
• Fourier reconstruction treats these phase differences as if they arise from different spatial frequencies, projecting "ghost" copies of moving structures displaced in the PE direction.
• Ghost spacing = FOV / (motion frequency × TR × number of PE steps).

• Patient education: instruct to hold still, practice breath-holding
• Sedation or general anesthesia for uncooperative patients
• Physical restraint (foam pads, wrapping)
• Swap PE direction to move ghosts away from area of interest
• Increase NSA/NEX (averages) - reduces random motion effects
• Cardiac gating (ECG-triggered or peripheral pulse)
• Respiratory gating (bellows, navigator echo - PACE technique)
• Ultrafast sequences (HASTE, EPI, TrueFISP) - freeze motion in 2-5 seconds
• PROPELLER/BLADE technique - oversample k-space center, detect and reject bad data
• Rest slabs / saturation bands over moving structures
• Antiperistaltic agents (e.g., Buscopan for bowel)
• Respiratory compensation (ROPE, COPE algorithms)

---

2. Nyquist Ghost (N/2 Ghost)

• EPI acquires all k-space lines in a single TR using alternating gradient polarity (each odd line is read left-to-right, each even line right-to-left).
• Any timing errors, eddy currents, or B₀ inhomogeneity cause a phase offset between even and odd echo lines.
• This even-odd phase inconsistency creates a N/2 frequency modulation in k-space.
• After Fourier transform, this appears as a ghost shifted exactly FOV/2.

• Reference scan (navigator scan without phase encoding) acquired before each EPI acquisition to measure and correct the phase offset
• Improved gradient hardware (faster rise times, better eddy current pre-compensation)
• Ghost correction algorithms in reconstruction software
• Parallel imaging (GRAPPA, SENSE) reduces the number of EPI echoes needed

---

3. Flow Artifact / Pulsatile Flow Artifact

• Flowing protons experience different RF pulses and gradient pulses than stationary tissue.
• Inflow enhancement (Time-of-flight effect): fully magnetized fresh spins entering the slice produce brighter signal than partially saturated stationary tissue.
• Pulsatile flow: the velocity of arterial blood changes with the cardiac cycle. Different k-space lines are acquired at different phases of systole/diastole, producing phase inconsistency → ghosts in PE direction.
• Phase dispersion: spins moving through a gradient acquire phase relative to stationary spins. If velocities vary across the vessel, phase spread causes signal loss (especially in GRE).

• Cardiac gating (ECG-trigger data acquisition to same cardiac phase)
• Saturation bands (REST slabs) placed superior/inferior to slice to pre-saturate flowing blood
• Flow compensation (gradient moment nulling / GMN): adds additional gradient lobes that rephase first-order (velocity) motion
• Swap PE direction to redirect ghosts
• Use spin echo rather than GRE sequences (SE refocuses phase errors)
• VENC (velocity encoding) matched appropriately in phase-contrast MRI

---

CATEGORY 2 - K-SPACE / FOURIER ARTIFACTS

4. Gibbs Ringing / Truncation Artifact

• The Fourier transform requires an infinite number of frequencies to represent a sharp discontinuity (step function) perfectly.
• When k-space is truncated (limited matrix size, especially in PE direction), high-spatial-frequency data is absent.
• The incomplete Fourier series produces the Gibbs phenomenon: oscillations (ringing) near the edge.
• The number of rings = half the number of k-space lines.
• More prominent in PE direction because fewer PE steps are used to shorten scan time.

• Increase matrix size (more phase encoding steps) - directly reduces artifact
• Increase FOV in PE direction (oversampling)
• Apply k-space filtering (Hanning window, Fermi filter) - smooths edges but slightly blurs image
• Zero-filling / zero padding in k-space (interpolation, not true resolution improvement)

---

5. Zipper Artifact

• External RF interference enters the scanner room (improperly shielded Faraday cage, RF door left open, electronic equipment inside).
• This extraneous RF signal at a frequency within the receiver bandwidth is treated by the system as if it originated from a specific spatial location (determined by the gradient at that moment).
• The system plots it as a line of false signal at that frequency-encode position.

• Ensure RF shielding (Faraday cage) is intact - door must be fully closed during scanning
• Remove electronic devices from scanner room
• Identify and eliminate RF-emitting sources
• Hardware inspection if persistent

---

6. Herringbone / Crisscross Artifact

• A single corrupted k-space data point (spike) from a hardware malfunction, electronic transient, or data spike in raw data.
• Because k-space and image space are Fourier pairs, a single delta function (spike) anywhere in k-space transforms to a sinusoidal wave spanning the entire image.
• Multiple spikes or a corrupted k-space line generates a superimposed grid pattern.

• Identify and remove/replace the corrupted k-space line (spike correction algorithms)
• Repeat the sequence
• Hardware maintenance (gradient amplifier, ADC board checks)
• Raw data filtering

---

7. Data Clipping Artifact

• The received MRI signal is amplified and digitized by the Analog-to-Digital Converter (ADC).
• If the signal amplitude exceeds the dynamic range of the ADC (receiver gain set too high), the signal is "clipped" - the waveform is truncated at the maximum value.
• This is equivalent to multiplying the true signal by a rectangular window in the time domain, which Fourier transforms to convolution with a sinc function → ringing and edge artifacts.
• Also occurs with incorrectly set prescan gain.

• Proper prescan calibration (auto-prescan)
• Reduce receiver gain (manual correction)
• Do not place high-signal structures (e.g., pads soaked in contrast material) inside the FOV unnecessarily

---

CATEGORY 3 - FREQUENCY/PHASE ENCODING ARTIFACTS

8. Chemical Shift Artifact (Type 1)

• Water and fat protons precess at slightly different Larmor frequencies due to electron shielding differences.
• At 1.5T: fat resonates ~220 Hz lower than water (3.5 ppm difference × 63.87 MHz = ~224 Hz).
• At 3T: the difference is ~440 Hz.
• The frequency-encoding gradient is used by the system to assign spatial location by frequency.
• Fat signal at a given location is assigned to the wrong pixel (shifted by ΔHz / gradient amplitude in Hz/mm).
• At fat/water interface: where they overlap → bright band; where gap appears → dark band.

• Increase receiver bandwidth (reduces pixel shift, but also reduces SNR)
• Fat suppression (FATSAT, STIR, Dixon) - eliminates fat signal entirely
• Swap to phase-encode direction (shifts artifact but doesn't eliminate)
• Use higher bandwidth sequences
• Avoid narrow bandwidth sequences near fat-water interfaces

---

9. Chemical Shift Artifact (Type 2) / India Ink / Black Boundary Artifact

• In GRE sequences, water and fat spins are initially in phase at TE=0.
• Because fat precesses 3.5 ppm slower, they progressively dephase.
• At specific TEs (called "out-of-phase" TEs), water and fat spins in the same voxel point in exactly opposite directions and cancel each other.
• At 1.5T: in-phase TE ~4.6 ms, 9.2 ms; out-of-phase TE ~2.3 ms, 6.9 ms.
• At 3T: in-phase TE ~2.3 ms; out-of-phase TE ~1.15 ms.
• Voxels at fat-water interfaces containing both fat and water protons lose signal → black rim.

• Choose in-phase TE values (avoid out-of-phase TEs) to minimize artifact
• Out-of-phase imaging is diagnostically used to detect intracellular lipid (adrenal adenoma, fatty liver)
• Dixon technique uses both in-phase and out-of-phase images to separately map fat and water

---

10. Aliasing / Wrap-Around / Foldover Artifact

• The Nyquist sampling theorem requires the sampling rate to be at least twice the highest frequency present.
• In the PE direction, the sampling rate is determined by the number of PE steps × TR.
• If anatomy extends beyond the FOV, it contains spatial frequencies exceeding the Nyquist limit.
• These aliased frequencies are indistinguishable from lower frequencies and are plotted on the opposite side of the image.
• In FE direction: oversampling is trivial (zero cost), so FE aliasing is automatically suppressed.
• PE direction aliasing is the problem because adding PE steps increases scan time.

• Increase FOV in PE direction (encompasses all anatomy)
• Phase oversampling (No Phase Wrap / anti-aliasing): acquires more PE steps beyond FOV but discards them from display
• Swap PE to a direction where anatomy doesn't extend beyond FOV
• Saturation bands on tissue outside FOV
• Parallel imaging methods

---

11. Fat-Water Swap Artifact

• Dixon imaging generates fat-only and water-only images by acquiring two echoes (in-phase and out-of-phase) and computing: Water = (IP + OOP)/2; Fat = (IP - OOP)/2.
• Phase errors from B₀ inhomogeneity cause fat and water phases to be misidentified.
• The algorithm applies a region-growing or phase unwrapping algorithm to determine which is fat and which is water; if the algorithm fails in a region (e.g., near air-tissue interfaces), it assigns the labels incorrectly.

• Improved B₀ shimming before scan
• Use 3-point or multi-point Dixon (better phase unwrapping)
• Apply region of interest constraints
• Manual post-processing correction

---

CATEGORY 4 - SUSCEPTIBILITY / FIELD-RELATED ARTIFACTS

12. Magnetic Susceptibility Artifact

• Magnetic susceptibility (χ) describes how much a material becomes magnetized in B₀.
• Ferromagnetic/paramagnetic materials (metal, hemosiderin, air-tissue interfaces) have χ very different from surrounding tissue.
• This creates local B₀ field inhomogeneity (ΔB).
• Local field distortion causes: a) Frequency shift → spatial misregistration (wrong pixel assignment) b) Phase dispersion within a voxel → signal loss (T2* decay) c) GRE sequences are far more sensitive because no 180° refocusing pulse is applied. d) SE sequences partially compensate via 180° pulse (refocuses static dephasing).
• The "blooming effect" is an exaggeration of the true size of susceptibility sources (hemosiderin, calcification, air).

• Use spin echo instead of gradient echo
• Lower field strength (artifact scales with B₀)
• Shorter TE
• Increase readout bandwidth
• Metal Artifact Reduction Sequences (MARS): view-angle tilting, SEMAC, MARS-SE
• Titanium implants (lower susceptibility than stainless steel or cobalt-chromium)
• STIR rather than CHESS fat suppression near metal

---

13. Blooming Artifact

• Extension of susceptibility artifact where the field perturbation extends beyond the physical boundaries of the susceptibility source.
• T2* decay from dephasing spreads signal void beyond the true dimensions.
• Clinically useful: makes microhemorrhages, calcifications, air emboli visible that would otherwise be invisible.
• SWI (Susceptibility Weighted Imaging) deliberately maximizes blooming for hemosiderin detection.

• Reduce by using SE sequences (see susceptibility above)
• Deliberately exploit with GRE/SWI to detect microbleeds in traumatic brain injury, cavernomas, amyloid angiopathy

---

14. Dielectric Artifact / Dielectric Effect

• The RF wavelength (λ = c / f) at 3T (128 MHz) in tissue is approximately 26 cm.
• When the patient's body dimension approaches or exceeds λ/2, standing wave interference patterns form inside the body.
• This causes constructive interference (bright regions) in the center and destructive interference (dark regions) at the periphery, or vice versa.
• Called "dielectric effect" because tissue permittivity influences wave propagation.
• Also contributes to non-uniform flip angles across the FOV.
• More severe at 3T and 7T than 1.5T.

• Dielectric pads (bags of high-permittivity gel placed on patient) - alter the standing wave pattern
• RF transmit optimization (B1+ shimming with multi-channel transmit coil)
• 3D RF pulses for better B1 uniformity
• Work at 1.5T for large patients

---

15. Eddy Currents Artifact

• Rapidly switching gradient fields induce currents in conducting parts of the scanner (cryostat, RF shields, gradient coil housing) by Faraday's law of induction.
• These induced currents generate their own secondary magnetic fields that oppose the intended gradient (Lenz's law).
• Eddy fields decay with their own time constants (τ).
• If not corrected, they cause: incorrect slice positions, image warping, phase errors in EPI.
• In diffusion imaging: eddy currents from strong diffusion gradients cause image-to-image geometric distortion that varies with diffusion direction.

• Active gradient shielding (shielded gradient coils): secondary winding cancels eddy field
• Pre-emphasis (pre-distortion) of gradient waveforms: gradient amplifier adds corrective pre-current
• Hardware: improved cryostat design with reduced eddy conductance
• Post-processing: eddy current correction in diffusion imaging (FSL eddy tool, FEE)
• Use of shorter, more efficient gradient pulses (VERSE)

---

CATEGORY 5 - PARALLEL IMAGING / ACCELERATION ARTIFACTS

16. Parallel Imaging Artifact (Residual Aliasing / g-factor Noise)

• Parallel imaging undersamples k-space (acquires every Rth PE line, where R = acceleration factor).
• Undersampling causes R-fold aliasing by Nyquist rules.
• SENSE: uses coil sensitivity maps to "unfold" aliased images. Where coil sensitivity profiles are poorly differentiated (far from coils), unfolding fails → residual aliasing.
• GRAPPA: uses k-space autocalibration signal (ACS) lines to synthesize missing k-space. Errors in fitting produce noise amplification characterized by g-factor.
• Higher R → shorter scan time but worse SNR and more potential artifact.

• Reduce acceleration factor R
• Use coils with more elements (better coil sensitivity variation)
• Improve ACS calibration (more ACS lines for GRAPPA)
• Regularization in reconstruction (SENSE-REGULARIZED)
• Use combined GRAPPA + SENSE or compressed sensing

---

17. Moire Fringes

• When two periodic signals with slightly different spatial frequencies superimpose, a low-frequency "beat" pattern appears - this is the Moire fringe.
• In MRI, occurs when aliased signal from outside FOV interacts with in-FOV signal, or from wrap-around in conjunction with field inhomogeneity.
• Also seen in sequences where periodic k-space sampling interacts with oscillating physiological signals.
• Can occur in large FOV coronal/axial acquisitions where signal from both sides of the body wraps.

• Phase oversampling to prevent wrap-around
• Increase FOV
• Improved B₀ shimming
• Surface coils with limited coverage

---

CATEGORY 6 - GEOMETRIC / SPATIAL ARTIFACTS

18. Gradient Non-Linearity Artifact

• Ideal gradient fields must be perfectly linear (G = constant × position) for correct spatial encoding.
• Real gradient coils only achieve perfect linearity near the isocenter.
• Toward the edges of the gradient FOV, the gradient field deviates from linearity.
• Non-linear gradients assign incorrect spatial positions to peripheral voxels.
• Objects near the edge appear displaced or distorted (barrel or pincushion distortion).
• Particularly significant in stereotactic planning and radiation therapy target definition.

• 3D gradient warp correction (GradWarp) applied in post-processing using known gradient coil geometry
• Scan at isocenter (center of gradient linear region)
• Gradient coil design improvements (larger linear FOV)
• Dedicated gradient linearity calibration phantoms

---

19. Stairstep / Staircase Artifact (2D Multislice)

• 2D multislice acquisitions have discrete, thick slices with gaps between them.
• When reformatted in an orthogonal plane, the finite slice thickness creates a blocky stepped appearance rather than a smooth curve.
• Unlike isotropic 3D acquisitions, 2D slices have inherently anisotropic voxels (slice thickness >> in-plane resolution).
• Reformat resolution in the z-direction is limited to slice thickness.

• Use thin slices with small or zero gaps
• Use 3D volumetric acquisition (isotropic voxels → smooth reformats in any plane)
• Overlap slices (gap = negative)
• Post-processing interpolation (though adds apparent not true resolution)

---

20. Partial Volume Artifact

• MRI assigns a single signal value to each voxel.
• If a voxel contains two tissues (e.g., fat and tumor, or fluid and soft tissue), the signal is the volume-weighted average.
• Thick slices and large in-plane voxels increase the probability of partial volume.
• A large voxel spanning a tissue edge will register the average of both tissues.
• Small lesions smaller than voxel size may be completely obscured.

• Use thinner slices
• Increase in-plane matrix (smaller voxels)
• 3D acquisition with isotropic voxels
• Align slices perpendicular to the interface of interest
• Avoid interpolation as a substitute for true resolution

---

CATEGORY 7 - SEQUENCE-SPECIFIC ARTIFACTS

21. Magic Angle Artifact

• In ordered structures (collagen fibers in tendon, ligament), the T2 relaxation is highly anisotropic.
• The dipolar coupling between adjacent protons in collagen depends on the orientation of the fiber to B₀.
• The dipolar Hamiltonian contains a (3cos²θ - 1) term. When θ = 54.74° (the "magic angle"), this term = 0.
• At the magic angle, dipolar coupling vanishes, T2 effectively lengthens dramatically (from ~1 ms to ~10-20 ms).
• On sequences with short TE (<20 ms): T1W, PD, GRE - the tendon appears bright.
• On long TE T2W sequences, the extended T2 still decays before signal acquisition → no artifact.

• Use long TE sequences (T2W): artifact disappears
• Reposition the body part to change fiber orientation
• Be aware at 55° to B₀ (typically curved tendons passing around pulleys)

---

22. Cross-Talk / Slice Cross-Excitation Artifact

• Slice selection RF pulses are not perfectly rectangular in frequency profile - they have imperfect edges and side lobes.
• Adjacent slices share overlapping RF excitation if the slice gap is insufficient.
• A proton in the "tail" of slice 1's profile is partially excited, and before it recovers (T1 recovery), it is excited again by slice 2.
• This partial saturation reduces signal in overlapping regions.
• More severe with closely spaced or contiguous slices.

• Use interleaved slice acquisition (acquire odd slices first, then even - separates simultaneously acquired neighbors)
• Maintain adequate slice gap (typically 10-20% of slice thickness)
• Use 3D volumetric sequences (no inter-slice cross-talk)
• Sinc RF pulses with better slice profiles (or optimized pulses)

---

23. Projection Artifact / Incomplete Saturation

• In 2D imaging, the slice-selective gradient limits excitation to a chosen plane.
• If out-of-plane signal is not properly suppressed (inadequate spoiling, incomplete saturation), residual magnetization from surrounding tissue contributes signal.
• Particularly seen when FOV encompasses high-signal structures (e.g., bowel with oral contrast, fat) that "project" through the slice.
• Can also occur in 3D sequences at slab boundaries.

• Ensure adequate spoiling (gradient spoilers, RF spoiling)
• Saturation bands placed superior/inferior to slab
• Adequate TR to allow T1 recovery discrimination
• Correct slab thickness and position

---

24. Surface Coil Intensity Variation / Surface Coil Flare

• Surface coils are receive-only coils with a spatially varying sensitivity profile.
• SNR and signal intensity decay approximately as 1/r³ from the coil center.
• Near the coil: very high signal (bright).
• Away from coil: rapidly decreasing sensitivity.
• The non-uniform B₁⁻ receive sensitivity is not compensated unless explicitly corrected.

• Intensity normalization (prescan normalization, N3/N4 correction) - post-processing algorithms
• Sensitivity encoding correction (divide image by coil sensitivity map)
• Use volume coils (more uniform) for structures requiring uniformity
• Normalization filters in scanner software

---

25. Saturation Band Artifact / Dark Band

• Pre-saturation pulses (90° pulses followed by crusher gradients) are applied to suppress signal from specific regions (superior/inferior to imaging slab, fat, vessels).
• If the saturation band overlaps with the imaging volume due to poor planning, anatomy in the overlap zone is saturated and appears dark.
• Similarly, inversion recovery pulses (STIR, FLAIR) null signal at a specific TI; tissues with similar T1 to null-point will be suppressed.
• Dark band artifact also seen in balanced SSFP (bSSFP/TrueFISP) sequences from banding artifacts at points where B₀ causes off-resonance phase of exactly 180°.

• Careful placement of saturation bands (avoid overlap with ROI)
• Review saturation band positions on localizer before prescribing
• For bSSFP banding: improved B₀ shimming, phase-cycling techniques, or use alternative sequences

---

26. Signal Loss Artifact

• Susceptibility dephasing (see above)
• Poor coil placement - receiver coil too far from ROI
• Through-plane motion during echo time
• Flow void - fast-flowing blood dephases
• T2 decay* from iron deposition, hemosiderin, air
• Inadequate flip angle at that location (B1 non-uniformity)

---

CATEGORY 8 - POST-PROCESSING / RECONSTRUCTION ARTIFACTS

27. Gibbs/Ringing in Image Stitching / Composing Artifact

• Large anatomy (full spine, legs) requires multiple overlapping acquisitions with different coil positions/stations.
• Differences in B₁ uniformity, coil sensitivity, patient position shift between stations, or normalization errors create intensity mismatches at the stitch boundaries.
• Automated blending algorithms attempt to smooth transitions but residual seam lines may remain.

• Adequate overlap between stations
• Consistent normalization across stations
• Table-feed (continuous table movement) acquisitions (whole-body MRI) reduce station boundaries
• Use system-integrated AutoAlign or CARE protocols

---

28. Spatially Dependent Phase Artifact

• B₀ inhomogeneity across the FOV causes spatially varying resonance frequency offsets.
• During phase-sensitive acquisitions (phase contrast MRA, susceptibility mapping, Dixon), the background phase from B₀ variation adds to the desired phase signal.
• This "background phase" is spatially dependent and can mimic or obscure the signal of interest.

• High-order B₀ shimming before acquisition
• Background phase correction (polynomial fit and subtraction)
• Reference phase acquisition for subtraction
• PREBASELINE correction algorithms

---

CATEGORY 9 - CONTRAST AGENT ARTIFACTS

29. Gadolinium Pseudo-Layering Artifact

• Gadolinium chelates diffuse into fluid-containing spaces and may concentrate at the dependent portion due to their higher density.
• Gd shortens T1 of adjacent tissue/fluid; concentrated Gd creates T1 shortening at a specific depth.
• On T1W images this appears as a layer of bright signal resembling a fluid level.
• Can mimic intra-articular loose bodies, hemorrhagic layering, or complex cyst.

• Clinical correlation - recognize the phenomenon
• Repeat imaging after repositioning patient (true fluid levels shift; pseudo-layering also shifts but differently)
• Delay post-contrast timing

---

30. Radiation-Induced Artifact (MR-Linac / Hybrid MRI)

• In MR-guided radiotherapy (MRgRT), MRI is acquired simultaneously with radiation delivery.
• The radiation beam interacts with the main magnetic field - secondary electron trajectories are deflected by the Lorentz force (F = qv × B).
• The radiation ionizes gas and generates RF emission that contaminates MRI signal.
• Gradient eddy currents from rapidly switching MR gradients affect dosimetry.

• Temporal interleaving of MRI acquisition and beam delivery
• RF shielding optimization within the MR-Linac bore
• Hardware filtering; specialized reconstruction algorithms
• Sequence optimization for MR-Linac environment

---

CATEGORY 10 - ADDITIONAL ARTIFACTS

31. Signal-to-Noise Ratio (SNR) Artifacts

• Small voxels (thin slice, high matrix, small FOV)
• High receiver bandwidth
• Low field strength
• Poor coil-patient coupling
• High body temperature (Johnson noise)

• Larger voxels (compromise resolution)
• More averages (longer scan time)
• Narrow bandwidth (more chemical shift artifact)
• Better coil design
• Higher field strength

---

32. Starry Sky Artifact

• Phase encoding errors from multiple small motion events, gradient imperfections, or timing inconsistencies.
• In EPI-DWI: each bright "star" is caused by a localized eddy current or motion event affecting a single or few k-space lines.
• Corrupt k-space lines → point spread function artifacts.

• EPI correction
• Navigator echoes to detect and reject corrupted shots
• Multi-shot DWI with motion correction

---

33. Lipid Suppression Failure Artifact

• CHESS (Chemical Shift Selective) fat saturation uses a frequency-selective 90° pulse tuned to the fat resonance (3.5 ppm from water).
• B₀ inhomogeneity shifts the local resonance frequency away from the CHESS pulse center frequency.
• In regions where B₀ is inhomogeneous (near air-tissue interfaces, metallic implants, curved surfaces), fat is no longer at the expected frequency → CHESS pulse fails to saturate fat.
• Water protons may instead be saturated, reducing water signal.

• Shimming before fat-sat sequences
• STIR (Short Tau Inversion Recovery): T1-based fat suppression - not frequency dependent → more robust near metal and at field edges
• Spectral STIR or water-only excitation (SPAIR, IDEAL)
• Dixon-based fat separation (most robust, frequency-independent)

---

34. The "Annefact" / Antenna Effect

• Conductive wires resonant with the RF transmit frequency accumulate RF energy (antenna effect).
• The wire tip acts as a high-impedance source → local tissue heating AND local B₁ enhancement → abnormal signal.
• Length of resonance: ~λ/2 at Larmor frequency (at 1.5T: ~26 cm for half-wavelength antenna).

• Avoid long conductive leads inside the scanner without MR-conditional clearance
• Use MR-conditional pacemaker protocols
• Reduce RF power (SAR)
• Position leads to minimize resonant coupling

---

35. Zebra Stripe Artifact

• bSSFP sequences maintain a steady-state transverse magnetization between TR cycles.
• The steady-state signal is critically dependent on the resonance offset frequency.
• At off-resonance frequencies of ±1/(2TR), the steady-state signal drops to zero (banding nulls).
• B₀ inhomogeneity causes spatial variation in resonance offset → alternating bright (on-resonance) and dark (off-resonance/banding) bands.
• Band spacing = 1/TR in frequency; physically separated by distance proportional to B₀ gradient.

• Very short TR (pushes banding nulls farther apart in Hz - wider "pass band")
• High-order B₀ shimming
• Phase cycling methods (rotate RF phase to shift banding locations, combine images)
• Use non-bSSFP alternative if banding unacceptable

---

36. BLADE/PROPELLER Residual Artifact

• BLADE (Siemens)/PROPELLER (GE) acquires k-space as rotating blades/propellers, oversampling the k-space center.
• The iterative registration algorithm detects and corrects in-plane rotation and translation.
• If motion is through-plane, or if the blade itself contains motion, the correction is incomplete.
• Over-correction can over-smooth the final image.

• Motion during data acquisition minimized (patient coaching)
• Increased number of blades (more overdetermination for correction)
• BLADE-specific reconstruction parameters (weighting of outlier blades)

---

SUMMARY TABLE - QUICK REVISION

---

HIGH-YIELD EXAM POINTS

1. Chemical shift type 1 - frequency-encode direction, increases with lower bandwidth, increases with field strength (3T worse than 1.5T)
2. Gibbs ringing - pseudo-syrinx in spine - increase PE steps or use k-space filter
3. Magic angle - always use T2W to confirm/deny tendon pathology
4. Aliasing - PE direction only (FE oversampled automatically)
5. Susceptibility - GRE >> SE; scales with B₀; STIR better than CHESS near metal
6. Nyquist ghost - EPI specific; FOV/2 displacement; corrected by reference scan
7. Dielectric artifact - 3T-specific; central brightening; dielectric pads help
8. Zipper - always FE direction; always RF contamination; fix the shielding
9. bSSFP banding - short TR is the primary solution (moves nulls to wider frequency separation)
10. Parallel imaging noise - g-factor amplification; reduce R or use more coil elements

• Cummings Otolaryngology Head and Neck Surgery, pp. 149-150
• Bradley and Daroff's Neurology in Clinical Practice (GRE susceptibility, SWI sections)
• MRI Questions - Motion Artifacts
• Medical Professionals - MRI Artifact Guide