External magnetic field 𝑩0 is not a problem: homogeneous (a few ppm) Local internal distortions of the magnetic field are a problem Reason: all substances get magnetized when placed in a magnetic field Susceptibility 𝜒: • tells how easy substances can be magnetized: 𝑴 = 𝜒 𝑯 • 𝑯 is the “real” magnetic field and 𝑩 is in fact the magnetic induction • Relationship: 𝑩 = 𝜇𝑯 = 𝜇0 1 + 𝜒 𝑯 ( 𝜇 = permeability) • Diamagnetic substances: −𝜒 ≪ 1 → 𝜇 ≅ 𝜇0 Most tissues are diamagnetic • Paramagnetic substances: 𝜒 < 1 → 𝜇 > 𝜇0 (attracted by 𝑩0 ) Examples: gadolinium (Gd) and deoxyhemoglobin • Ferromagnetic substances: 𝜒 ≫ 1 → 𝜇 ≫ 𝜇0 (strongly attracted by 𝑩0 ) Examples: iron (Fe), cobalt (Co) and nickel (Ni)
Susceptibility artifacts occur at interfaces of different 𝜒, such as tissue-air interfaces:
• Differences in 𝜒 lead to distortions in the local field • Causes dephasing of spins (signal loss) and mismapping artifacts GRE is more sensitive to these artifacts than SE Artifacts Artifacts RF and gradient artifacts RF-pulses: • Crosstalk in multi-slice imaging: slice gap needed • Imperfect 180°-pulses: spoiling needed Gradients: • Eddy currents distort gradient waveforms • Nonlinearities cause local magnetic field distortions Artifacts Signal processing artifacts Aliasing: due to undersampling in k-space (object larger than FOV) • Solutions: – Oversampling: o FOV and 𝑁𝑥 are doubled with constant ∆𝑥 o Data outside FOV are discarded o Along read out: identical sampling time 𝑇𝑥 = 𝑁𝑥 ∆𝑡 and identical SNR o Along phase encode: longer 𝑇𝑎𝑐𝑞 → halfscan: identical 𝑇𝑎𝑐𝑞 and SNR – Saturation of signals outside FOV o Saturation recovery and spoiler gradient o Surface coil Artifacts Gibbs ringing • Spurious ringing around sharp edges • Reason: o Sharp edges contain very high spatial frequencies o They can not be measured in practice: truncation o Object convolved with sinc-function → ringing • Solutions: o Filtering data → reduced spatial resolution o 𝑁 ↑ with constant FOV ( 𝑓𝑜𝑠𝑐 ↑ as 𝑁 ↑ ) • Gibbs ringing most pronounced along phase encode (𝑁 determines 𝑇𝑎𝑐𝑞 ) Artifacts Chemical shift Chemical shift: difference in 𝑓0 experienced in different chemical environments Example: ∆𝑓𝑐𝑠 = 𝑓𝐹 − 𝑓𝑊 = −3.5 ppm Read out: 𝑓 ↔ 𝑥 thus misregistration Gives bright bands (overlap) and dark bands (subtraction) Displacement (in pixels): ∆𝑓𝑐𝑠 ∆𝑓𝑝𝑖𝑥𝑒𝑙 = 𝜎𝐵0 𝐺𝑥 ∆𝑥 Reduction: • 𝐺𝑥 ↑ or ∆𝑥 ↑ • fat sat pulses EPI: most prominent along phase encode direction (cumulative phase shifts) Artifacts Motion and flow Physiologic motion: Gross body movements, Respiratory motion, Cardiac motion, Peristalsis, Blood flow, Cerebrospinal fluid (CSF) Two time scales in MRI Read out: 𝑇𝑥 = 𝑁𝑥 ∆𝑡 Phase encode: 𝑇𝑅 Read out: blurring Phase encode: ghosting Reason: Phase encode cycle not synchronous with motion Motion leads to magnetization changes (magnitude and/or phase shifts) which vary from view to view (view = measurement of a profile) FT cannot reconstruct this in a correct manner (Modulation-shift theorem) Ghosts (partial copies) of the object Ghost separation in pixels (periodic motion): ∆= 𝑇𝑅 𝑁𝑦 𝑁𝐸𝑋 𝑇𝑝 = 𝑇𝑎𝑐𝑞 𝑇𝑝 Ghosts can be bright and dark, they become fainter with increasing distance from original structure Nonperiodic or random motions: ghost pattern more complex Artifacts Solutions • Pulsatile flow: spatial presaturation of inflowing protons (no signal, no ghosts) • Periodic motion: cardiac gating and ECG triggering • Respiratoty motion: – Monitor chest wall motion → reorder scan in k-space (central lines important) – Breath hold – Navigators • Fast scan imaging (EPI,…)
» A: free, B: ECG-triggered, C: navigator
» D: with T2-prepulse better contrast for coronaries