Postoperative Spinal Imaging Metal Artefact Reduction Using Ultrashort Echo Time MRI: A Feasibility Study
Amy Ming-Chun Tsai Sevao1, Alistair Young1, Benjamin Schmitt2, Hament Pandya3, Karen Billington3, Anthony Doyle3, David Grodzki4, and Brett Cowan5

1Anatomy with Medical Imaging, University of Auckland, Auckland, New Zealand, 2Sydney, Australia, 3Auckland, New Zealand, 4Erlangen, Germany, 5University of Auckland, Auckland, New Zealand

Synopsis

Post-operative spine imaging with metal implants in situ are problematic because of the significant metal artefacts, in both CT due to beam hardening, and MRI fom signal loss. Ultrashort Echo Time (UTE) MRI has potential to significantly reduce metal artefacts because of its method of acquisition. Significant differences in metal artefacts between conventional MR and UTE are found in our study, imaging sheep spine with spinal fusion hardware in situ, with importnat clinical implications.

Purpose

MRI is a common modality for imaging of the spine because of the excellent visualization of the spinal cord, nerve roots and intervertebral discs it provides. However its use is limited in the post-surgical context due to metal artefact. In conventional MRI sequences, the presence of metallic implants can cause T2* dephasing, image distortion and signal loss. These artefacts combine to make assessment of the structures near the metal implants difficult, and the inability to visualise important post-operative anatomy may adversely affect patient management1. Ultrashort Echo Time imaging (UTE) is a MRI sequence designed to visualise short T2 structures, which has been used in musculoskeletal imaging research and has immense potential in clinical MSK imaging. Theoretically, key properties of UTE such as reduced TE time and 3D radial acquisition can also reduce or eliminate some factors that cause metal artefacts2. The purpose of this study is to evaluate the effectiveness of UTE MRI and UTE using Pointwise Encoding Time Reduction with Radial Acquisition (PETRA), another sequences from the same family, in minimising metal artefacts of spinal fusion metal hardware, compared to that of conventional MR T1 and T2 sequences with WARP.

Methods

Pre-instrumentation MRI scans (T1, T2, UTE, UTE-PETRA) of the lumbar spine of four sheep carcasses were performed using a 3T clinical MRI scanner at (Siemens; Skyra) with a 18-channel body matrix. Posterior 3-level fusion of the lower lumbar regions were carried out by an orthopaedic surgical trainee, following routine operative protocols. Two sets of titanium and cobalt chromium pedicle screws and connecting rods were used respectively (Medtronic CD Horizon Solera 6.5 and Legacy 6.5) to represent the two most common metals in spine fusion hardware. Post-instrumentation MRI scanning used the institutional routine T1 and T2 protocols with WARP, and relevant post-instrumentation UTE MRI sequence parameters were: UTE (TE 0.08, TR 6, flip angle 8, FOV 240, radial spoke 64000) and UTE-PETRA (TE 0.08, TR 4, flip angle 6, FOV 240, radial spoke 90000). Two MSK radiologists assessed the visibility of neural foramina and nerve roots and the degree of anatomical distortion at the levels associated with the spinal fusion (Fig 1,2). They graded the images on a modified 5-point scale3, 0 = no visibility of anatomical structures or severe distortion, up to 4 = complete visibility of structures or no distortion present. Quantitative analysis is performed by manually outlining the regions of signal loss and blooming from the metal artefacts on each sequence4, and building 3D volume models from the corresponding ROIs . Both qualitative and quantitative measures were compared using the paired t test to find statistically significant differences between the metal artefact volume and the grades of anatomical visibility and distortion of the different sequences.

Results

T1 imaging (286.5 + 11.1cm3) demonstrated the maximum average volume of metal artefacts, followed by T2 (282.1 + 18.1cm3), UTE-PETRA (202 + 34.6cm3) and UTE the minimum (180.6 + 25.5cm3). While the difference in mean volume between T1 and T2 is not significant (p=0.78), the differences between T1 and UTE (p=0.004), T1 and UTE-PETRA (p=0.01), T2 and UTE (p=0.01), T2 and UTE-PETRA (p=0.04), and UTE and UTE-PETRA (p=0.03) were.The neural foramina was least visible on T2 imaging (2.6 + 0.4), followed by T1 (3.2 + 0.4), UTE-PETRA (3.5 + 0.4) and most visible on UTE (3.6 + 0.3). The differences in grades were significant when comparing T1 with UTE (p<0.05), T2 with UTE (p<0.001), and T2 with UTE-PETRA (p<0.001). The nerve roots were the least visible on T2 (2.1 + 0.5), followed by T1 (2.4 + 0.5), UTE-PETRA (3.3 + 0.4) and most visible on UTE (3.4 + 0.4). The differences were significant when comparing T1 with UTE (p<0.01), T1 with UTE-PETRA (p=0.02), T2 with UTE and PETRA respectively (p<0.01). Image distortion was the most severe on T1 (1.2 + 0.3), followed by T2 (1.5 + 0.2), UTE-PETRA (3.2 + 0.2) and least on UTE (3.3 + 0.2). The differences were significant when comparing between T1, T2, UTE and UTE-PETRA sequences (p<0.001).

Conclusion

Our results show that UTE significantly reduces metal artefacts, improving visibility of anatomical structures, minimising image distortion and reducing area of signal loss. The differences are significant when compared to conventional MRI sequences with WARP, the current gold standard in postoperative spine imaging. With further development, UTE could improve patient outcome by providing key information to guide clinical management in the post-operative setting that are currently unavailable through any other type of imaging.

Acknowledgements

We thank Siemens AG (Healthcare Sector, Erlangen, Germany) for providing the work-in-progress software package including UTE and UTE-PETRA. Special thanks are given to the staff at the Animal Lab (VJU), University of Auckland who provided the sheep carcasses and assisted with their management throughout the study.

References

1. Chang EY, Bae WC. Imaging the Knee in the Setting of Metal Hardware. Magn Reson Imaging Clin NA. Elsevier Inc; 2014;22(4):765–86.

2. Grodzki DM, Jakob PM, Heismann B. Ultrashort echo time imaging using pointwise encoding time reduction with radial acquisition (PETRA). Magn Reson Med. 2012 Feb;67(2):510–8.

3. Kretzschmar M, Nardo L, Han MM, Heilmeier U, Sam C, Joseph GB, et al. Metal artefact suppression at 3 T MRI: comparison of MAVRIC-SL with conventional fast spin echo sequences in patients with Hip joint arthroplasty. Eur Radiol [Internet]. 2015;25(8):2403–11. Available from: http://link.springer.com/10.1007/s00330-015-3628-0

4. Lazik A, Landgraeber S, Schulte P, Kraff O, Lauenstein TC, Theysohn JM. Usefulness of metal artifact reduction with WARP technique at 1.5 and 3T MRI in imaging metal-on-metal hip resurfacings. Skeletal Radiol [Internet]. 2015;44(7):941–51. Available from: http://link.springer.com/10.1007/s00256-015-2128-2

Figures

Fig. 1. Nerve root appearances on T1 (A), T2 (B), UTE (C) and UTE-PETRA (D) at similar levels, adjacent to titanium screws in the same sheep. Nerve roots on T1 and T2 (red arrowhead) are indistinct whereas the same nerve roots on UTE and UTE-PETRA are easily seen (red arrows).

Fig 2. The image of the spinal cord is distorted near the level of the pedicle screws on T1(A) and T2(B), outlined by the yellow dashed lines, but the image have minimal distortion on UTE(C) and PETRA(D), outlined by green dashed lines.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
4476