Reduction of Metal Artefacts in MRI of Hip Arthroplasty The following presentation was given at the Radiological Society of North America (RSNA) metalions@ucl.ac.uk Reduction of Metal Artefacts in MRI of Hip Arthroplasty D. McRobbie, A. Papadaki, K. Satchithananda, A Mitchell, A Hart, J Henckel and C Todd

Reduction of Metal Artefacts in MRI of Hip Arthroplasty TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Reduction of Metal Artefacts in MRI of Hip Arthroplasty D. McRobbie, A. Papadaki, K. Satchithananda, A Mitchell, A Hart, J Henckel and C Todd

TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Introduction Implant-derived metal debris from metal-on-metal hip replacements is thought to be responsible for pseudotumours, a fluid or solid mass of tissue visible on cross-sectional imaging. These pseudotumours have been linked to pain and tissue destruction but it is not known how common they are in well functioning hips imaged by MRI. This is key to understanding their clinical significance. Metal implants cause susceptibility artefacts in MRI. We will describe ways to reduce these artefacts to visualization of structures adjacent to implants.

Parameters and techniques that can affect TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Parameters and techniques that can affect the appearance of metal implants in MRI: TITLE Implant Type Size/geometry Direction in relation to Bo Bo effect Magnetic field strength Sequence and sequence parameters Type of sequence TE Slice Thickness Read out bandwidth Matrix FOV Phase/Frequency encode direction Sequence Design View angle tilting (VAT) and Multiple VAT Slice encoding for Metal Artifact Correction in MRI (SEMAC) Multi-acquisition variable-resonance image combination (MAVRIC)

Methods TITLE Co-Cr Titanium METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Methods To investigate the effect of some of the sequence parameters, a Co-Cr hip implant (Figure 1) was inserted in water and held in place by strings. A Titanium implant was also imaged for comparison. Images of the phantom were acquired mostly on a 1.5T MRI system. A 3T system was also used for comparison. Images of patients were obtained on a 1.5 T MRI system Co-Cr Titanium

Orientation relative to Bo TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Orientation relative to Bo To reduce artefact, optimal direction of the long axis of the implant is parallel to Bo Bo

TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Type of implant Titanium cause less artefacts than cobalt-chromium and stainless steel. Co-Cr Titanium Theory: The different magnetic susceptibility  of the implant from tissue results in unwanted field gradients that dephase the MR signal and lead to image distortions. CoCr has  ~ 900 ppm, about 5 times  for titanium [1].

TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES High field MRI systems Susceptibility artefacts are worse on 3T scanners compared to 1.5T 1.5 T 3T Theory : Field inhomogeneities ΔΒ scale with B0. The distortion is due to the size of ΔΒ relative to the frequency encode gradient GFE.. T2* is also reduced at higher B0.

Sequence Parameters to change TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Sequence Parameters to change TITLE INTRODUCTION These are sequence parameters that can be changed easily on any commercially available clinical MRI system [2] Type of sequence TE Slice Thickness Read out bandwidth Matrix FOV Phase/Frequency encode direction Parallel imaging SEQUENCE PARAMETERS VAT METHODS fMRI RESULTS EEG RESULTS fMRI OTHER CONCLUSION SUMMARY

Type of sequence Choose SE or TSE/FSE over gradient echo TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Type of sequence Choose SE or TSE/FSE over gradient echo Gradient Echo SE or TSE Theory: Spin echo and turbo spin echo sequences are preferred because they employ 180o RF refocusing pulses that can compensate for field inhomogeneities caused by metal implants.

Type of sequence Choose TSE/FSE over SE TITLE TE 6.5 TE 97 TE 6.5 METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Type of sequence Choose TSE/FSE over SE TE 6.5 TE 97 TE 6.5 TE 97 Turbo/fast spin echo Spin echo Theory: Longer TE in spin echo leads to more signal dephasing and uneven distortion between the echoes. Consistent image quality is maintained between echoes for TSE/FSE.

For spin echo In SE reduce TE DRAWBACK TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES For spin echo TITLE In SE reduce TE DRAWBACK Does not work well for T2-weighting INTRODUCTION TE 50 ms TE 6.5 ms TE 25 ms TE 100 ms SEQUENCE PARAMETERS VAT METHODS fMRI RESULTS EEG SE OTHER RESULTS fMRI Theory: Decreasing TE will allow less time for signal dephasing and reduce the artefact. Increasing bandwidth will allow shorter TE. For T2 –weighted images TSE should be used. The shortest inter-echo spacing should be used. Increasing the bandwidth will allow shorter echo spacing. SUMMARY CONCLUSION

For Fast/turbo spin echo TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES For Fast/turbo spin echo IES 5.06 ms IES 7.36 ms IES 9.42 ms reduce inter-echo spacing DRAWBACK Increase SAR Theory: The dephasing caused by the metal is substantially refocused by the train of 180 degree pulses. Distortions will still occur due to frequency misregistration but T2* signal voids are reduced.

For Fast/turbo spin echo TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES For Fast/turbo spin echo TE 22 ms reduce inter-echo spacing TE has little effect DRAWBACK Increase SAR TE 94 ms Theory: The dephasing caused by the metal is substantially refocused by the train of 180 degree pulses. Distortions will still occur due to frequency misregistration but T2* signal voids are reduced.

For Fast/turbo spin echo TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES For Fast/turbo spin echo reduce inter-echo spacing TE has little effect Echo train length or turbo factor has no effect. DRAWBACK Increase SAR TF 15 TF 7 TF 3 Theory: The dephasing caused by the metal is substantially refocused by the train of 180 degree pulses. Distortions will still occur due to frequency misregistration but T2* signal voids are reduced.

Decrease slice thickness TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Decrease slice thickness TITLE DRAWBACKS Signal-to-noise ratio (SNR) is decreased More slices required for coverage INTRODUCTION SEQUENCE PARAMETERS VAT METHODS fMRI METHODS fMRI METHODS fMRI RESULTS EEG RESULTS EEG RESULTS EEG OTHER RESULTS fMRI RESULTS fMRI RESULTS fMRI Theory: Decreasing the slice thickness increases the slice select gradient so more frequencies are included within the same voxel and artefact spreads over less voxels. Below a certain limit, the slice thickness may be controlled by the RF transmission bandwidth and no further improvement occurs. SUMMARY CONCLUSION CONCLUSION CONCLUSION

Decrease slice thickness TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Decrease slice thickness TITLE DRAWBACKS Signal-to-noise ratio (SNR) is decreased More slices required for coverage INTRODUCTION 7mm 5mm 3mm SEQUENCE PARAMETERS VAT METHODS fMRI METHODS fMRI METHODS fMRI RESULTS EEG RESULTS EEG RESULTS EEG RESULTS fMRI OTHER RESULTS fMRI RESULTS fMRI Theory: Decreasing the slice thickness increases the slice select gradient so more frequencies are included within the same voxel and artefact spreads over less voxels. Below a certain limit, the slice thickness may be controlled by the RF transmission bandwidth and no further improvement occurs. CONCLUSION SUMMARY CONCLUSION CONCLUSION

Increase bandwidth DRAWBACK Signal-to-noise ratio (SNR) is decreased TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Increase bandwidth TITLE INTRODUCTION DRAWBACK Signal-to-noise ratio (SNR) is decreased SEQUENCE PARAMETERS SEQUENCE PARAMETERS VAT METHODS fMRI METHODS fMRI METHODS fMRI RESULTS EEG RESULTS EEG RESULTS EEG RESULTS fMRI OTHER RESULTS fMRI RESULTS fMRI CONCLUSION SUMMARY CONCLUSION Theory: Increasing the pixel bandwidth increases the frequency encode gradient so more frequencies are included within the same voxel and artefact spreads over less voxels. CONCLUSION

Increase bandwidth DRAWBACK Signal-to-noise ratio (SNR) is decreased TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Increase bandwidth DRAWBACK Signal-to-noise ratio (SNR) is decreased 781 Hx/px 400 Hz/px 600 Hz/px 199 Hz/px Theory: Increasing the pixel bandwidth increases the frequency encode gradient so more frequencies are included within the same voxel and artefact spreads over less voxels.

Increase matrix or reduce field-of-view TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Increase matrix or reduce field-of-view DRAWBACK Increased matrix will decrease SNR and increase scan time Theory: Increasing the matrix for constant FOV increases the frequency encode gradient so more frequencies are included within the same voxel and artefact spreads over less voxels. Decreasing FOV has the same effect although is not often changed as anatomy of interest should be covered.

Increase matrix or reduce field-of-view TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Increase matrix or reduce field-of-view TITLE The reduction of artefact is not significant if both the matrix and bandwidth are changed [3] Hint: Select resolution required for clinical scan and optimise bandwidth. DRAWBACK Increased matrix will decrease SNR and increase scan time 128 x 76 512 x 306 320 x 192 256 x 153 INTRODUCTION SEQUENCE PARAMETERS VAT METHODS fMRI METHODS fMRI METHODS fMRI RESULTS EEG RESULTS EEG RESULTS EEG OTHER RESULTS fMRI RESULTS fMRI RESULTS fMRI SUMMARY CONCLUSION Theory: Increasing the matrix for constant FOV increases the frequency encode gradient so more frequencies are included within the same voxel and artefact spreads over less voxels. Decreasing FOV has the same effect although is not often changed as anatomy of interest should be covered. CONCLUSION CONCLUSION

Frequency encode direction TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Frequency encode direction The metal artefact is worse in the frequency encode direction, generally select frequency encode parallel to the long axis of the implant Phase encode LR (coronal plane) Phase encode HF (coronal plane)

TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Fat suppression Use STIR instead of spectral fat saturation or water excitation T2 tse fat sat STIR Theory: STIR relies upon nulling the signal from fat due to its T1relaxation time following an inversion pulse. This is unaffected by inhomogeneities caused by metal. Fat-sat, water-excitation and Dixon depend upon the resonant frequencies of water and fat. These are altered by the presence of the metal.

Parallel Imaging Parallel imaging has little effect on the artefact TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES Parallel Imaging TITLE INTRODUCTION Parallel imaging has little effect on the artefact Use it if necessary to reduce scan time VAT METHODS fMRI SEQUENCE DEVELOPMENT RESULTS EEG R=1 R=2 R=3 R=4 OTHER RESULTS fMRI Theory: Parallel imaging reduces the number of phase encode lines acquired using array coil element sensitivities either in image space or k-space. Distortion is still present due to frequency encoding. CONCLUSION SUMMARY

Standard MARS Protocol Optimised MARS protocol TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETERS REFERENCES NEW SEQUENCES OPTIMISED SEQUENCE T1w coronal TSE Standard MARS Protocol TE/TR = 7/679 ms Matrix 320 x 192 Bandwidth 744 Hz/px Slice thickness = 5mm PE direction HF Optimised MARS protocol TE/TR = 7.7/694 ms Matrix 512 x 384 Bandwidth 751 Hz/px Slice thickness = 3mm PE direction RL

TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETRS REFERENCES NEW SEQUENCES Sequence development New sequences under development offer the prospect of further reducing the metal artefact. VAT – view angle tilting [4] MAVRIC – Multi-acquisition variable-resonance image combination [1] SEMAC - Slice encoding for metal artefact correction in MRI [5] Theory: View Angle Tilting partially corrects the in-plane distortion, but at the cost of introducing blurring. To unwrap the distortion extreme tilt angles are necessary. Butts et al [4] have proposed refinements to the VAT method to reduce the effect of the blurring[. Several non-standard methods are in development but are as yet unavailable on commercial scanners: MAVRIC [1] and SEMAC [5] utilise modified 3D methods taking advantage of the insensitivity of the phase encoding process to magnetic field distortions. Single point scanning [6] involves the use of 3 phase encodes but results in impractically long scan times.

TITLE METHODS EEG METHODS fMRI RESULTS EEG CONCLUSION INTRODUCTION METHODS PHYSICAL PARAMETERS SEQUENCE PARAMETRS REFERENCES NEW SEQUENCES REFERENCES: Koch KM, Lorbiecki JE, Hinks RS,King KF, A multispectral three-dimensional acquisition technique for imaging near metal implants, Magn Reson Med, 2009, 61: 381-390 Vandevenne JE, Vanhoenacker FM, Parizel PM, Pauly KB, Lang PK, Reduction of metal artefacts in musculoskeletal MR Imaging, JBR-BTR,2007, 90: 345-349 Toms AP, Smith-Bateman C, Malcolm PN, Cahir J, Graves M Optimization of metal artefact reduction (MAR) sequences for MRI of total hip prostheses, Clin Radiol, 2010, 65: 447-452 Butts K, Pauly JM, Gold GE, Reduction of blurring in View Angle Tilting MRI, Magn Reson Med, 2005, 53: 418-424 Lu W, Pauly KB, Gold GE, Pauly JM, Hargreaves BA, SEMAC: Slice Encoding for Metal Artifact Correction in MRI, MRM, 2009, 62: 66-76 Ramos-Cabrer P, Duynhoven JPM, van der Toorn A, Nicolay K (2004) MRI of hip prostheses using single-point methods: in vitro studies towards the artefact free imaging of individuals with metal implants, Magn. Reson. Imag. 22:1097-1103