High Field MRI: Technology, Applications, Safety, and Limitations R. Jason Stafford, Ph.D. Department Department of of Imaging Imaging Physics Physics The The University University of of Texas Texas M. M. D. D. Anderson Anderson Cancer Cancer Center Center Houston, Houston, TX TX Brief overview of high-field MRI • Introduction • Technical/Safety Issues 16T, 32mm vertical bore – Main field – RF Field – Gradients • Contrast changes – – – – Spin lattice relaxation (T1) Spin-Spin relaxation (T2) Transverse relaxation (T2*) Spectral Resolution • Major Applications Berry MV & Geim AK, "Of Flying Frogs and Levitrons", Levitrons", Euro J Phys 18: 18: 307307-313 (1997). AAPM AAPM 2005 2005 The promise of high-field MRI • Trade SNR increase into higher resolution/speed – Higher resolution imaging • More detail & less partial volume averaging – Faster Imaging • Breath holds, minimize motion, kinetics • Exploration of new/altered contrast mechanisms • Potential to significantly advance anatomic, functional, metabolic and molecular MR imaging AAPM AAPM 2005 2005 1 The move to higher fields • 3.0T whole body scanners – FDA approved for commercial use in 2002 – Accounted for 8.5% of high-field revenue in 2003 Signa Excite Intera Achieva General Electric Philips MAGNETOM Trio Siemens • Whole body 4-9.4T scanners: in evaluation AAPM AAPM 2005 2005 High-field SNR: 7T versus 3T in brain 7T white matter SNR =65 Gray matter SNR = 76 3T white matter SNR =26 Gray matter SNR = 34 TSE, 11 echoes, 7 min exam, 20cm FOV, 512x512 (0.4mm x 0.4mm), 3mm thick slices AAPM AAPM 2005 2005 Images courtesy of SIEMENS Medical Systems The future? … 9.4 T whole body imaging Early test image of a kiwi GE Medical Systems 9.4 T @ Univ. Illinois Chicago AAPM AAPM 2005 2005 http://www.uic.edu/depts/paff/newsbureau/mriphotos.htm http://www.uic.edu/depts/paff/newsbureau/mriphotos.htm 2 The static magnetic field (B0) • Modern superconducting magnet design – Type II superconductors – Niobium titanium (NbTi) windings • Critical field limits upper field (< 10 T) • Bypass by cooling < 4.2 K – Niobium tin (Nb3Sn) for higher fields • Brittle and difficult to wind • Expensive to use – Fields above 10 T likely to interleave both windings AAPM AAPM 2005 2005 High-field siting challenges • To keep constant homogeneity, as B0 ↑ magnet – size increases – weight increases – cryogen volume and consumption increases • energy stored in windings increases – stray field lines are extended • Costs and siting concerns can be significant – Modern 3T scanners weigh 2x as much as 1.5T – >3T : 20 tons with cryogens + 100 tons shielding AAPM AAPM 2005 2005 High-field siting challenges • Challenge: minimize these costs while maintaining field homogeneity? • Magnet winding circuits • Tighter/more windings to reduce length • Reduced length => reduced cryogen volume/use • • • • Conductor formats and joining techniques Filament alloys Shimming Shielding AAPM AAPM 2005 2005 3 Magnet shimming • Need higher performing, automated shims to maintain homogeneity • Several stages – Magnet => δ < 125 ppm – Superconducting shims: δ < 1.5 ppm – Passive + Room Temperature: δ < 0.2 ppm 0.5 ppm 3T 1.5T 0 AAPM AAPM 2005 2005 -0.5 ppm Magnetic field homogeneity • Often stated as the δ (in Hz or ppm) across a given diameter of spherical volume (DSV). • Homogeneity desired is often application dependent • For 1.5T: – Routine imaging: – Fast imaging (EPI): – Spectroscopy: < 5 ppm at 35 cm DSV < 1 ppm at 35 cm DSV < 0.5 ppm at 35 cm DSV AAPM AAPM 2005 2005 Inhomogeneities/susceptibility errors • Off-resonance displacements are in the frequency encode direction δ ⋅ γ B ⋅ FOV ∆x = ppm 0 BW – Minimize errors by • Increased bandwidths (BW) • Decreased field of view (FOV) and/or slice thickness • Increase encoding matrix • Don’t forget slice select geometric distortions! – Increase RF bandwidth (if possible) AAPM AAPM 2005 2005 4 Metal implants at 3 Tesla 3T 1.5T metal prosthetic BW=125 kHz BW=50kHz AAPM AAPM 2005 2005 Magnet Shielding • Reduces problems of siting MRI in a confined space – 5 G line reduced from 10-13 m => 2-4 m • Passive Shielding – high permeability material, such as iron, provides return path for stray field lines of B0 decreasing the flux away from the magnet. – can be quite heavy and expensive • Active Shielding – secondary shielding coils produce a field canceling fringe fields generated by primary field coils – typically coils reside inside the magnet cryostat – Commercial 3T scanners rely on this to minimize weight AAPM AAPM 2005 2005 Isogauss plot of 1.5T actively shielded magnet Fringe Fields: 1.5T Fault condition: 5m radial x 7m axial for t<2s AAPM AAPM 2005 2005 5 Isogauss plot of 3.0T actively shielded magnet Fringe Fields: 3.0T Fault condition: 7.5 m for t<100s 6mx AAPM AAPM 2005 2005 FDA B0 Field Safety limits Guidance for Industry and FDA Staff: Criteria for Significant Risk Investigations of Magnetic Resonance Diagnostic Devices, July 14th, 2003. http://www.fda.gov/cdrh/ode/guidance/793.pdf AAPM AAPM 2005 2005 B0 field safety concerns • Ferromagnetic projectiles • Medical devices – Translation – Torque – Interference • Magnetohydrodynamic effects AAPM AAPM 2005 2005 6 Main field Safety: Torques and Force • Torque (L) on an object in magnetic field L ∝ m ⋅ B0 ⋅ sin θ ∝ B02 ⋅ sin θ • Translational force on object in magnetic field r r F ∝ ∇(m × B0 ) ∝ B0∇B0 • Torque and translational force also proportional to susceptibility and volume of material AAPM AAPM 2005 2005 Magnetic Field safety: Torques and Force • Equipment formally designated as “MR Safe” at 1.5T may not be at 3T • Force on a paramagnetic object at 3T can be about 5x the force at 1.5T • Force on a ferromagnetic object can be about 2.5x the force at 1.5T • Effects are worse in “short bore” • The “list” of tested devices – www.mrisafety.com www.simplyphysics.com www.simplyphysics.com AAPM AAPM 2005 2005 3.0 0.050 0.045 1.5T 3.0T 1.5T: 5G 3.0T: 5G 2.5 1.5T 3.0T 1.5T: 5G 3.0T: 5G Fringe Field Force: 1.5T versus 3.0T Field Strength 2.0 0.040 0.035 0.030 0.025 1.5 0.020 1.0 0.015 0.010 0.5 0.005 0.0 0.000 350 0 2 4 6 250 4.0 2 3 4 5 6 3.5 1.5T 1.5T: 5G 3.0T 3.0T: 5G 300 Relative Force 8 7 8 1.5T 1.5T: 5G 3.0T 3.0T: 5G 3.0 2.5 200 2.0 150 1.5 100 1.0 50 0.5 0 0.0 0 AAPM AAPM 2005 2005 2 4 6 8 2 3 4 5 6 7 8 Axial Distance from Isocenter (m) 7 Magnetohydrodynamic Effects • Electrically conductive fluid flow in magnetic field induces current and a force opposing the fluid flow • Effects greatest when flow perpendicular to field – Potential across vessel ~ B0 – Force resisting flow ~ B02 • T-wave swelling – – – – ECG distortions during highest aortic flow Induced potentials ~ 5 mV/Tesla Effect exacerbated at high-fields Challenge: obtaining reliable ECG’s at higher fields AAPM AAPM 2005 2005 Magnetohydrodynamic Effects • Increased blood pressure due to additional work needed to overcome magnetohydrodynamic force has a negligible effect on blood pressure – < 0.2% at 10 Tesla • Hypothesized that field strengths ranging from 18 Tesla are needed before a significant risk is seen in humans. AAPM AAPM 2005 2005 Transient effects from the static field • Phenomena reported in association with patients moving in/out of high field magnets – – – – – – Nausea (slight) Vertigo Headache Tingling/numbness Visual disturbances (phosphenes) Pain associated with tooth fillings • Effects transient and cease after leaving magnet – actively shielded & short bore high-field magnets • larger spatial gradient – reduced or avoided by moving slowly in the main field AAPM AAPM 2005 2005 8 Radiofrequency at high-field • B1 field sensitivity goes as B0 • RF propagation becomes increasingly inhomogeneous – – – – Wavelength now on order of patient size Permittivity, conductivity and patient conformation Reduced penetration Increased dielectric effects • RF phase and magnitude function of position • Significant imaging challenges lie ahead … AAPM AAPM 2005 2005 B1 inhomogeneity • Hyperintensity in middle of imaged volume – Dielectric effects become more significant as B0↑ – Oil filled phantoms more homogeneous – Challenges for brain and body homogeneity 3T Profile 1.5T profile 1.5T 3.0T AAPM AAPM 2005 2005 B1 inhomogeneity 3T 1.5T T1-W Spin Echo Images using torso array coil Destructive interference in the pelvis can lead to persistent “black hole” artifacts. Use a dielectric pad to help correct. New coil designs to help minimize effects. AAPM AAPM 2005 2005 9 RF Safety: Specific Absorption Rate (SAR) • Conductivity (σ) in body gives rise to E field from RF – SAR ~ σΕ2/ρ • SAR = RF Power Absorbed per unit mass (W/kg) – 1 W/kg => 1°C/hr heating in an insulated tissue slab • RF power deposition causes heating – Primary concern: whole body and localized heating – Significant concern at high-fields – Don’t forget about local heating of medical devices! SAR ∝ B02 ⋅ (flip angle) 2 ⋅ (RF duty cycle) ⋅ (Patient Size) AAPM AAPM 2005 2005 FDA SAR limits Guidance for Industry and FDA Staff: Criteria for Significant Risk Investigations of Magnetic Resonance Diagnostic Devices, July 14th, 2003. http://www.fda.gov/cdrh/ode/guidance/793.pdf AAPM AAPM 2005 2005 International Electrotechnical Commission • 3T MR unit operating modes are set by recommended IEC guidelines – Scanners report SAR in real-time – notify users of operating thresholds • Normal Mode – SAR < 2 W/kg over 6 minutes, (∆T < 0.5 °C) • First Level controlled Mode – Requires medical supervision – SAR < 4 W/kg over 6 minutes, (∆T < 1.0 °C) • Second level controlled mode – IRB is needed to scan humans AAPM AAPM 2005 2005 10 SAR limits on imaging SAR ∝ B02 ⋅ (flip angle) 2 ⋅ (RF duty cycle) ⋅ (Patient Size) • Puts restrictions on – – – – Pulse repetition time Number of RF pulses in a multi-echo sequence (FSE) Slice efficiency in multi-slice imaging Ability to use high SAR pulses for contrast • Fat saturation • Magnetization transfer pulses • Inversion pulses AAPM AAPM 2005 2005 Ways to work around SAR limitations • RF pulse design – Reduced flip angle (particularly for fast spin echo) • RF coil design – Use of arrays • Transmit-receive arrays to reduce power • Parallel imaging techniques (SENSE, SMASH) • Imaging parameters – Rectangular field of view – Increased TR – Less slice in multi-slice imaging (lower efficiency) AAPM AAPM 2005 2005 Partially Parallel Imaging (PPI) • PPI will be increasingly important in the development of high field imaging • Uses apriori localized sensitivity information from multiple receiver array coils to recover full image from undersamped k-space acquisition • Standard software on new generation scanners – SENSitivity Encoding (SENSE) • Pruessmann KP, et al, Magn Reson Med. 42(5):952-62 (1999). – SiMultaneous Acquistion of Spatial Harmonics (SMASH) • Sodickson DK, et al, Magn Reson Med. 38(4):591-603 (1997). AAPM AAPM 2005 2005 11 Low Resolution Fully encoded image Aliased Images 43 2 Un-aliased Image 1 unfold SENSE 3 2 1 unalias SMASH Corrected k-space phase-encode direction Undersampled k-space phase-encode direction 4 Advantages of parallel imaging • Facilitates faster acquisition by collecting less lines of k-space – Doesn’t compromise resolution – SNR reduced by AT LEAST a factor of √2 • Less # of echoes => Less SAR • Reduces T2/T2* blurring for echoecho-train sequences • Reduction of acoustic noise AAPM AAPM 2005 2005 Gradients at higher-fields • High performance gradients needed to take advantage of increased SNR for high resolution/speed • Max amplitude ~ 20-50 mT/m • Max slew rates ~ 120-200 T/m/s • Increased reactive (inductive and capacitive) coupling to bore/shims/RF coils – increased eddy currents and non-linearities – self-inductance limits maximum amplitude and slew rate – lower inductance designs the easiest fix AAPM AAPM 2005 2005 12 Gradient safety at higher fields • Physiological constraints on dB/dt to prevent peripheral nerve stimulation limit gradient performance – One strategy for overcoming: shorten linearity volume • Acoustic noise – force on the coils scales with the main field • NOTE: Higher performance gradients are usually linear over a more restricted FOV – Increased geometric distortions AAPM AAPM 2005 2005 Field Strength and Image Quality • What happens to SNR and contrast with increased main field? AAPM AAPM 2005 2005 Signal as a function of field strength • Where does the increase in signal come from? • Sample magnetization proportional to B0 M 0 ∝ ∆N ↑↓ ≈ N s hγ B0 kT • Faraday’s Law: Induced e.m.f. in coil proportional to time rate of change of transverse magnetization Larmor Precession Frequency = ω 0 = γ B0 AAPM AAPM 2005 2005 13 Higher fields … how much SNR? • Signal versus field strength Signal ∝ ω 0 M 0 ∝ B02 • Noise versus field strength 2 2 Noise ∝ σ coil aB01/ 2 + bB02 + system + σ sample ∝ AAPM AAPM 2005 2005 High-field signal-to-noise ratio SNR ∝ ⎧⎪ B07 / 4 (low field limit) ∝⎨ aB01/ 2 + bB02 ⎪⎩ B0 (mid-field regime) B02 • At high-field, B1(B0) is no longer easily quantifiable • SNR is still “nearly” linear with B0 in this regime AAPM AAPM 2005 2005 T1 relaxation as a function of B0 T1 (ms) T1 (ω 0 ) ≅ Aω 0b a gr ym at ite wh r te m at r te adipose B0 (T) AAPM AAPM 2005 2005 Bottomley PA, et al, Med Phys (1987) 14 T1 relaxation as a function of B0 • Spin lattice relaxation both lengthens and converges for most tissues with increased field strength • Consequences – Contrast and SNR reduction • Need longer TR and/or prepatory pulses – Longer inversion times needed • STIR and FLAIR – Tissue and blood more easily saturated – Reduced Ernst angles in gradient echo imaging AAPM AAPM 2005 2005 T1-weighted imaging • Can use SNR boost for higher resolution – Keep similar scan time • Spin-echo T1-W imaging will be SAR limited – Number of slices – Fat saturation • Solutions – – – – – Use an array head coil Reduce number of slices Rectangular field of view Longer TR Multiple acquisitions 1.5T 3.0T T2 and T2* relaxation as a function of B0 • T2 can decrease slightly at fields > 3T • T2* decreases significantly at higher fields – Changes vary strongly with tissue environment – Effects • Increased T2* contrast from contrast agents or blood • Decreased signal on gradient echo images due to susceptibility effects – Use of shorter TE • T2* filtering of echo trains in EPI – Use of shorter echo trains (multi-shot or PPI) AAPM AAPM 2005 2005 15 T2-weighted imaging • Benefits from higher SNR – Higher bandwidth => longer acceptable echo-trains – Potential for higher resolution in similar time • Longer TR to compensate for T1 lengthening – Overcome time penalty with longer echo-train and rectangular field of view acquisitions AAPM AAPM 2005 2005 T2-weighted brain using 8 channel array 3T 1.5T 7200 TR 3500 62.5 BW 31.25 24 ETL 8 2 Navg 1 24x18 FOV 20x20 512x256 256x224 1:55 Time 1:38 ∆x = 0.47mm x 0.94mm ∆x = 0.78mm x 0.89mm AAPM AAPM 2005 2005 Spectral resolution at higher fields • Larger spectral separation between different chemical species – MR spectroscopy applications will obviously benefit from this and SNR increase • Chemical shift between fat/water increases – 220 Hz @ 1.5T => 440 Hz @ 3T – faster accrual of phase between water/fat for a given TE • In-phase, out-of-phase TE timings change – exasperates chemical shift artifacts • Use higher bandwidths AAPM AAPM 2005 2005 16 Imaging applications • What are some of the major applications that will receive the highest boost from higher field imaging? AAPM AAPM 2005 2005 Spectroscopy • Increased spectral resolution and SNR – Brain, prostate, breast (+ other body applications) – Multi-nuclear: 31P, 19F, 23Na, 13C 1.5T Cr NAA 3.0T NAA SNRCr increased Cr by factor of ~2 at 3T Cho Cho AAPM AAPM 2005 2005 2hz, 131 deg, 99% 4hz, 140 deg, 99% Prostate & breast spectroscopy 3T Prostate MRS Cunningham, CH, et al, Magn Reson Med 53:1033–1039 (2005) 4T Breast MRS Bolan, PJ, et al, Magn Reson Med. 50(6):1134-43 (2003) AAPM AAPM 2005 2005 17 Blood Oxygen Level Dependent imaging • Functional MRI relies on the BOLD effect • BOLD facilitates neuronal activation measurements without using exogenous contrast agents • Activation: Oxy-blood increases while Deoxy-blood (paramagnetic) decreases – T2* is lengthened => signal increase – BOLD contrast increased due to smaller T2* values – SNR increase also leads to higher sensitivity • CNR increases by factor of 1.8-2.2 from 1.5T to 3.0T – These net effects, and reasons for them, are complicated AAPM AAPM 2005 2005 Krasnow, B, et al, NeruoImage (2003) Gradient- echo BOLD fMRI: 1.5T vs 3.0T 1.5T: p < 7.9 x10-5 3.0T: p < 1x10-10 Right Hand Sensorimotor Task Diffusion Weighted Imaging • SNR is crucial – Thinner slices • Reduce partial volume artifacts – Higher b-values 1.5T 3.0T • Diffusion Tensor Imaging (DTI) – Same benefits as DWI – Faster acq.=> minimize motion • Shortened T2* – limits benefits – Use parallel imaging techniques 3.0T Diffusion and Diffusion Tensor Imaging 18 Body Diffusion MRI at 3T: Breast Ax T2 Anisotropic ADC Isotropic AAPM AAPM 2005 2005 Perfusion imaging • Arterial Spin Labeling (ASL) – Uses and inversion pulse to “tag” blood – Images acquired as tagged blood perfuses into tissue – Long T1 results in better tagging • Dynamic Susceptibility Contrast (DSC) – Bolus of paramagnetic agent • T2* contrast – T2* effect increased by field 1.5T 3.0T AAPM AAPM 2005 2005 Contrast Enhanced imaging • Higher SNR • Longer tissue T1 vs. little change in contrast agent T1 1.5T 3.0T – Better contrast – Use less contrast • Perform higher resolution dynamic imaging • Applications: brain, breast and body imaging Dynamic contrast enhanced imaging AAPM AAPM 2005 2005 19 MR Angiography at higher fields • In general SNR => better spatiotemporal resolution • Time of flight (TOF) – Relies on saturated normal tissue and bright inflow – Longer T1 time => better background tissue saturation • Magnetization Transfer Contrast can further suppress – Must be careful of SAR limits – Higher-field => increased inflow signal • Contrast bolus – Better T1-contrast • Phase contrast – More sensitive to slow flow 3D TOF Gibbs, GF, et al, AJNR (2004) (www.medical.philips.com) Cardiac imaging at higher fields • Speed is king in cardiac imaging – Use parallel imaging techniques to their fullest • CINE imaging – SAR => reduce flip angles for SSFP (trueFISP/FIESTA) • T2 weighting and SNR loss • Black blood imaging (double inversion recovery) – Increased T1 of blood (+ 30% ) => longer inversion time needed – More SNR and slow T1 relaxation • Chance to increase the limited slice efficiency of method • Cardiac Tagging methods – Persistance of tagging • Emerging techniques: Perfusion, Delayed Enhancement AAPM AAPM 2005 2005 Cardiac imaging at 3T SSFP CINE Cardiac Tagging 1.5T +SENSE 1.5T 3T systole diastole 3T AAPM AAPM 2005 2005 Gutberlet, M, Eur Radiol 15: 1586–1597 (2005). 20