Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Jan 1.
Published in final edited form as: Magn Reson Imaging. 2023 Nov 7;105:57–66. doi: 10.1016/j.mri.2023.11.007

Metal Artifact Reduction Around Cervical Spine Implant Using Diffusion Tensor Imaging at 3T: A Phantom Study

Slimane Tounekti 1, Mahdi Alizadeh 2, Devon Middleton 1, James S Harrop 2, Hiba Bassem 3, Laura Krisa 4, Choukri Mekkaoui 5,6,7, Feroze B Mohamed 1
PMCID: PMC10841892  NIHMSID: NIHMS1943566  PMID: 37939969

Abstract

Purpose:

Diffusion MRI continues to play a key role in non-invasively assessing spinal cord integrity and pre-operative injury evaluation. However, post-operative Diffusion Tensor Imaging (DTI) acquisition of patients with metal implants results in severe geometric distortion. We propose and demonstrate a method to alleviate the technical challenges facing the acquisition of DTI on post-operative cases and longitudinal evaluation of therapeutics.

Material and Methods:

The described technique is based on the combination of the reduced Field-Of-View (rFOV) strategy and the phase segmented EPI, termed rFOV-PS-EPI. A custom-built phantom based on a cervical spine model with metal implants was used to collect DTI data at 3 Tesla scanner using: rFOV-PS-EPI, reduced Field-Of-View single-shot EPI (rFOV-SS-EPI), and conventional full FOV techniques including SS-EPI, PS-EPI, and readout-segmented EPI (RS-EPI). Geometric distortion, SNR, and signal void were assessed to evaluate images and compare the sequences. A two-sample t-test was performed with p-value of 0.05 or less to indicate statistical significance.

Results:

The reduced FOV techniques showed better capability to reduce distortions compared to the Full FOV techniques. The rFOV-PS-EPI method provided DTI images of the phantom at the level of the hardware whereas the conventional rFOV-SS-EPI is useful only when the metal is approximately 20 mm away. In addition, compared to the rFOV-SS-EPI technique, the suggested approach produced smaller signal voids area as well as significantly reduced geometric distortion in Circularity (p < 0.005) and Eccentricity (p < 0.005) measurements. No statistically significant differences were found for these geometric distortion measurements between the rFOV-PS-EPI DTI sequence and conventional structural T2 images (p > 0.05).

Conclusion:

The combination of rFOV and a phase-segmented acquisition approach is effective for reducing metal-induced distortions in DTI scan on spinal cord with metal hardware at 3T.

Keywords: Diffusion Tensor Imaging, Metal implant, Metal Artifacts, Geometric Distortion, 3 Tesla

Introduction

Diffusion Tensor Imaging (DTI) has emerged as a key tool for in-vivo investigation of the central nervous system (i.e. brain and spinal cord (SC)) integrity and white matter (WM) connectivity mapping 1,2. It has the potential of probing pathology induced changes in neural system microstructure by assessing the diffusion properties of water molecules inside the biological tissues in multiple directions, providing insights into the organization and orientation of structures 3. DTI-derived metrics Fractional Anisotropy (FA), Mean Diffusivity (MD), Radial Diffusivity (RD), and Axial Diffusivity (AD), can be computed and used as a reliable imaging biomarkers for describing the SC microstructure changes with certain pathologies 4. In fact, numerous studies have demonstrated that there is a correlation between the computed metrics and the standard clinical scales used for assessing the severity of physical disability such as the Japanese Orthopaedic Association (JOA) or the modified JOA (mJOA) score 5,6 as well the INSCSCI score.

Spinal Cord disorders such as spinal cord injury (SCI) and Degenerative Spondylotic Myelopathy (DSM) can affect the entire nervous system and lead to tissue/axonal damage, i.e., demyelination, transection, and atrophy, resulting in serious clinical complication including motor and/or sensory systems dysfunction, and partial or complete paralysis. In this context, the microstructural damage such as demyelination and axonal loss associated with SCI cause changes in the diffusivity of water in the spinal cord, and fiber bundles density, resulting in changes in DTI-derived metrics according to the level and severity of the damage 7-9. Surgical intervention for SCI treatment entail implantation of metallic hardware (Anterior or Posterior) for maintaining SC stabilization and avoid further short-term or long-term complications 10.

Structural MRI techniques, i.e., T1- and T2-weighted imaging, and DTI have gained popularity in clinical practice compared to conventional radiology modalities (CT, X-Ray) for pre-operative diagnosis and assessment of SCI. Kara et al have demonstrated the feasibility of using DTI-derived indices as robust biomarkers for the early detection of DSM in patients with normal-appearing T2-weighted images 11. A recent study has shown that the change of DTI-metrics correlates with the change in the mJOA score as well as with the mJOA recovery rate, showing evidence that pre-operative DTI has prognostic potential in predicting surgical outcomes 12,13. However, DTI is hampered by its intrinsic low-sensitivity, as well as the low-spatial resolution and high-sensitivity to motion, leading to image distortions and a weak signal-to-noise ratio (SNR). In addition, performing DTI on SC is technically challenging mainly due to the small dimensions of the SC, the physiological motion (e.g., heart, lungs, and throat in its proximity) and the susceptibility-induced distortions. However as described below, the last decade has seen numerous developments in newer pulse sequences coupled with spine specific post processing capabilities to overcome these barriers enabling reproducible and reliable data collection of the human spinal cord.

DTI is often performed using the single-shot Echo Planar Imaging pulse sequence (SS-EPI) due to its fast acquisition speed. However, this acquisition method is prone to various limitations, including eddy-currents and high-sensitivity to magnetic field inhomogeneity, inducing artifacts dramatically affecting the image quality. To address some of these technical issues, two dimension radiofrequency (2DRF) pulses have been introduced and coupled with the SS-EPI readout for collecting reduced Field-Of-View (rFOV) dMRI data of the SC 14-16. The rFOV-SS-EPI acquisition technique has been shown to provide high-resolution, reduced-distortion dMRI of SC and has been implemented on most vendors’ MR scanners, namely ZOOMIT (Siemens), IZOOM (Philips), and FOCUS (GE). In addition, Phase-segmented EPI (PS-EPI) and readout-segmented EPI (RS-EPI) pulse sequences have been introduced recently for performing high-resolution and distortion reduced DTI on the brain 17-19. However, the application of these techniques for SC diffusion imaging remains limited due to their high sensitivity to motion.

Postsurgical metallic implants typically induce dramatic magnetic field inhomogeneities, leading to severe image distortions. The metal-induced artifact depends on the implant hardware (material, size, shape), the magnetic field B0 strength, and MRI sequence 20. Given these technical challenges limiting the ability of collecting images near the metal hardware, the use of DTI for evaluating post-operative clinical outcomes remain an unexplored field and the post-surgery SCI assessment is still heavily based on structural MRI techniques and a surgeon’s skills. Studies have attempted various acquisition approaches to demonstrate the feasibility of performing metal-artifacts reduction for DTI scan near metal implant 21-23. Despite these recent technical developments, the potential to effectively suppress metal-induced artifacts for the diffusion-MRI scan is still unreached. This is mainly due to the limitations of the proposed techniques, including through-plane distortion, image blurring, and low SNR. In addition, this can be due to the high B0 field inhomogeneity in proximity to the metal, specifically at ultra-high field (UHF) such as 3 or 7 Tesla.

In this study, we combine the rFOV technique with the PS-EPI pulse sequence, so called (rFOV-PS-EPI), to address geometric distortions near the metal in DTI scan of SC at 3T. Distortion-reduced DTI data were collected on a custom-built cervical spine phantom model with a metal implant configuration commonly used in spine surgery. The efficacy of the proposed pulse sequence in reduction of metal artifacts was visually and quantitatively evaluated compared to the gold standard rFOV-SS-EPI pulse sequence as well as the full-FOV approaches: SS-EPI, PS-EPI, and RS-EPI.

Material and Methods:

1. Protocol Setup

The experiments in this study were conducted on 3T Magnetom Prisma system (Siemens Healthineers, Erlangen, Germany) with a maximum gradient amplitude of 80 mT/m and a maximum slew-rate of 200 T/m/s. The phantom was placed in the vendor 64-channel receive head-neck coil, with the MR body coil used for radiofrequency (RF) transmission. In this phantom-based study, no human or animal subjects were involved in phantom construction or data collected. Therefore, approval from the ethics committee was not required.

2. Cervical Spine Phantom:

Figure. 1 illustrates the spine model as well as the final phantom used in the experiments. A cervical spine model with MRI compatible Titanium Alloy implants as currently used in the clinics was constructed and (NuVasive, CA, USA) used in this study (Figure 1. a). Ten grams of agar-agar powder was diluted in one liter of water to the phantom was suspended in this solution. An asparagus stalk was used as a surrogate spinal cord to provide an anisotropic structure to contrast with the surrounding gel medium. It was placed in the spinal canal of the spine model (Figure 1. b). The spinal phantom model was centered in a cylindric plastic container. After filling the container with the agar solution, it was left in ambient temperature to set (Figure 1. c).

Figure 1:

Figure 1:

Open in a new tab

The cervical Spine phantom used in this study. (a) Spine model with MRI compatible Titanium Alloy implants. (b) An asparagus was placed in the spinal canal of the spine model. (c) The spinal phantom model was centered in a cylindrical plastic container and filled with the agar solution.

3. Pulse sequence design

Figure.2. illustrates the EPI readout sampling scheme of the SS-EPI, PS-EPI, and the RS-EPI. The gold standard SS-EPI pulse sequence allows collecting the data in one excitation pulse (Figure 2. a). However, the PS-EPI (Figure 2. b) and RS-EPI (Figure 2. c) require multiple RF shots for fully sampling the Fourier-Space.

Figure 2:

Figure 2:

Open in a new tab

(a)The readout sampling scheme of the single-shot EPI (SS-EPI), (b) Phase-Segmented EPI (PS-EPI), and (c) the Readout-Segmented EPI (PS-EPI). The SS-EPI pulse sequence need only one RF pulse for acquiring one slice. However, the PS-EPI and RS-EPI (Figure 2. b and c) require multiple RF shots for fully sampling the Fourier-Space.

In this study, the gold standard SS-EPI technique was modified to build the PS-EPI MR pulse sequence using the IDEA pulse sequence environment (Siemens Medical Solutions, Erlangen, Germany). Then, the 2DRF excitation pulse was implemented into the developed PS-EPI sequence to create the proposed rFOV-PS-EPI approach. The diagram of the rFOV-PS-EPI pulse sequence used for data collection is shown in Figure 3. The 2DRF excitation pulse was used to acquire a rFOV along the Phase Encoding (PE) direction. It consists of 45 PE lines with an echo spacing of 0.37 ms, resulting in a total duration of 16.93 ms. The conventional Stejskal-Tanner diffusion preparation scheme with one refusing pulse of 180 degrees is used to acquire diffusion weighted images 24. Although the 2DRF aims to acquire a reduced number of acquired lines at each RF pulse, the Phase segmentation sampling scheme allows achieving a significantly shorter EPI readout train, addressing the susceptibility-induced geometric distortions. The fat saturation technique, not shown in the diagram, was applied prior the RF excitation pulse in order to suppress the unwanted fat signal and address the chemical-shift related artifacts in the images.

Figure 3:

Figure 3:

Open in a new tab

The diagram of the developed rFOV-PS-EPI pulse sequence used in this study. The 2DRF excitation pulse consists of 45 phase encoding blips with an echo spacing of 0.37 ms, resulting in a total duration of 16.93 ms.

4. Acquisition Parameters

Twelve axial diffusion-weighted images were acquired with a b-value of 600 s/mm2 applied along twelve non-collinear diffusion encoding directions using the vendor provided gradient direction file under the MDDW diffusion mode. One non-DW image with b-value of 0 s/mm2 was collected prior the dMRI images. The detailed acquisition parameters of all diffusion MR pulse sequences used in this phantom study are given in Table.1.

Table.1:

DTI pulse sequence parameters used on Phantom

Parameters SS-EPI PS-EPI RS-EPI rFOV-SS-EPI rFOV-PS-EPI
TR (ms) 5200 2600 5370 4500 2900
TE (ms) 90 43 75 62 60
Number of slices 30 30 30 30 30
Slice thickness (mm) 5 5 5 5 5
FOV (mm2) 120×120 120×120 120×120 120×57 120×57
Matrix size (mm2) 134×134 134×134 134×134 134×64 134×64
Spatial resolution (mm3) 0.9×0.9×5 0.9×0.9×5 0.9×0.9×5 0.9×0.9×5 0.9×0.9×5
PE direction R>L R>L R>L R>L R >L
Phase Partial Fourier 6/8 6/8 × 6/8 6/8
Read Partial Fourier × × 6/8 × ×
Echo spacing (ms) 1 1.11 0.38 0.98 1.1
Bandwidth (Hz/Px) 1244 1382 691 1166 1286
EPI factor 134 19 134 64 8
Number of segments 1 7 7 1 8
Fat region Thickness (mm) × × × 20 20
Scan Time (min) 1:13 4:36 6 1:03 5:07
Open in a new tab

In addition, conventional anatomical 3D T2-weighted images were collected using a SPACE pulse sequence. The spatial resolution was set to 0.9×0.9×5 mm3 to fit the voxel size of the DTI scan. The other scan parameters were TR= 1500 ms, TE= 110 ms, Flip angle =120°, bandwidth= 620 Hz/Pixel, Slice thickness= 5 mm, FOV = 120×120 mm2, matrix size=128×128 mm2, number of slices= 30, phase oversampling= 30%, slice Partial Fourier = 6/8, scan time= 1:23 minutes.

5. Data analysis

The collected images were first converted to the Neuroimaging Informatics Technology Initiative (NIFTI) format. Diffusion tensor maps FA, colored FA, MD, AD, and RD were extracted using DTIFIT tool in FSL (fsl.fmrib.ox.ac.uk/fsl).

5. 1. SNR Measurement:

A Region-Of-Interest (ROI)-based SNR measurement was performed on b0-images collected using the rFOV-PS-EPI and rFOV-SS-EPI to quantify the SNR improvement between the two pulse sequences. The ROIs were manually selected at the center of the asparagus using the Medical Image Processing, Analysis, and Visualization (MIPAV) software (mipav.cit.nih.gov). The measurement was not performed at the first and last slices to avoid effects from signal dropout induced by RF pulse imperfection. Slices 10 to 18 were excluding from measurement due to the presence of high-intensity geometric distortions in data acquired with the standard rFOV-SS-EPI pulse sequence.

5. 2. Signal-void Measurement:

A manually drawn ROI was used to measure the extent of the signal void area around the hardware at the sagittal view of the b0-images acquired with the rFOV-PS-EPI as well as the rFOV-SS-EPI.

5. 3. Distortion Evaluation:

To quantify the geometric distortion induced by the presence of the metal, slice-by-slice ROIs were manually selected following the asparagus edge on T2 image and the ADC maps computed from the rFOV-PS-EPI, rFOV-SS-EPI using MIPAV by an imaging expert. Supplementary Figure 1 shows an example of the selected ROI on rFOV-PS-EPI and T2 images. ROI-based parameters: Area, Perimeter, Circularity, and Eccentricity were then extracted for pairwise comparison between the structural and diffusion data.

5. 4. Statistical Analysis:

In order to assess the difference between the gold standard rFOV-SS-EPI and the developed rFOV-PS-EPI measurement, a two-sample t-test was performed using SPSS (IBM SPSS, Chicago) with a two-sided p-value of 0.05 or less used to indicate a significant result.

Results:

Figure 4 displays the b0-image and the Trace maps obtained using the standard SS-EPI (top row), the PS-EPI (second row), and the RS-EPI (bottom row) of the same slice shown in the T2-image. The selected slice is located at the metal hardware level and indicated by the green line (cross hair) in the sagittal view of the T2-image. The gold standard SS-EPI sequence suffers from severe susceptibility artifacts which affect the shape of the phantom and completely hinders the visibility of the asparagus. The PS-EPI sequence significantly reduces the geometric distortions compared to the SS-EPI and RS-EPI where the asparagus-model is clearly visible as highlighted in the zoomed red box. However, some residual geometric distortion can be seen as indicated by the red arrows. The fourth and fifth rows of Figure 4 show the comparison between the rFOV-SS-EPI, commonly used for SC imaging, and the developed rFOV-PS-EPI pulse sequence. The Spinal canal as well as the asparagus cylindrical shape are clearly visible in the images obtained with the proposed rFOV-PS-EPI method (zoomed green box).

Figure 4:

Figure 4:

Open in a new tab

The b0-images and the Trace maps obtained using the standard SS-EPI, the PS-EPI, the RS-EPI, rFOV-SS-EPI and the developed rFOV-PS-EPI pulse sequence of the same slice. The selected slice is located at the metal hardware level as indicated by the green line in the sagittal view of the T2-weighted image. The SS-EPI sequence suffers from high-intensity susceptibility artifacts. The asparagus can be detectable with the PS-EPI pulse sequence but there are some remaining distortions as highlighted in the zoomed red box. However, Spinal Canal as well as the asparagus spherical shape are clearly visible in the images obtained with the proposed rFOV-PS-EPI method (zoomed green box).

In order to accurately assess the potential of suggested method in providing distortion-reduced images near the metal, Figure 5 compares the b0-images and MD maps obtained at various distances from the metal (metal level, 5, 10, 15, 20 mm) using the rFOV-PS-EPI and rFOV-SS-EPI pulse sequences. The selected slices are indicated by the dashed colored lines on the sagittal view of the T2-image and displayed on the right. The ability of the developed method to obtain a true shape of the asparagus at b0 and MD maps at a slice level close to the metal implant is highlighted by the colored zoomed boxes.

Figure 5:

Figure 5:

Open in a new tab

Axial views of the b0-images and MD maps obtained at 0 mm, 5 mm, 10, 15, 20 mm from the metal using the rFOV-PS-EPI and rFOV-SS-EPI pulse sequences. The selected slices are indicated by the dashed colored lines on the sagittal view of the T2-image and displayed on the right. The capacity of the developed method to obtain a true shape of the asparagus at b0 and MD maps at a slice level close to the metal implant is highlighted by the red zoomed boxes. The rFOV-SS-EPI provide an image of the asparagus at a distance of 20 mm away from the metal hardware.

SNR Measurement:

Table 2 shows the slice-by-slice ROI-based SNR values extracted from b0-images collected using the rFOV-SS-EPI and rFOV-PS-EPI images. The mean and standard deviation (SD) of signal comparison shows that the conventional rFOV-SS-EPI method provide near 20% higher SNR than the proposed rFOV-PS-EPI technique. A t-test comparison also shows a significant difference between the SNR values of the two sequences with a p-value of 0.029 (< 0.05).

Table.2:

SNR measurement extracted using manually selected ROI on b0-images acquired using the rFOV-SS-EPI and rFOV-PS-EPI pulse sequence.

rFOV-SS-EPI rFOV-PS-EPI
Slice Index ROI size SNR ROI size SNR
2 41 57.5 41 63.11
3 43 57.51 43 61.85
4 40 57.49 40 54.58
5 43 57.39 43 48.92
6 41 57.15 41 51.93
7 22 57.51 37 54.16
8 21 57.51 37 47.51
9 9 56.81 13 39.39
19 13 57.21 45 30.90
20 44 42.60 44 31.08
21 31 45.92 31 25.71
22 27 29.38 27 22.97
23 43 36.48 43 25.09
24 50 38.58 50 31.02
25 34 50.42 34 45.63
26 57 56.45 57 42.03
27 53 57.14 53 48.03
Mean/SD - 51.36/13.6 - 42.61/12.88
Open in a new tab

Signal-void Measurement:

The signal-void measurement shows that hypo-signal region around the metal is nearly 30 % smaller with the rFOV-PS-EPI with an Area of 976 mm2 compared to the rFOV-SS-EPI where the measured Area was 1513 m2, Figure 6.

Figure 6:

Figure 6:

Open in a new tab

ROI-based signal void area measurement at the sagittal view of the b0-images acquired with the rFOV-SS-EPI and the rFOV-PS-EPI.

Distortion Evaluation:

Figure 7 displays a direct comparison between the FA and colored FA maps computed from data collected using the rFOV-SS-EPI and rFOV-PS-EPI acquisition methods. Significant reduction in geometric distortions can be seen with data acquired using the proposed technique, allowing a better detection of the asparagus model with its cylindrical shape and the blue color at the colored FA map.

Figure 7:

Figure 7:

Open in a new tab

The FA and colored FA maps computed from data collected using the rFOV-SS-EPI and rFOV-PS-EPI acquisition methods. Significant reduction in geometric distortions can be seen with data acquired using the proposed technique.

Figure 8 shows the slice-by-slice comparison of ROI-based parameters: Area, Perimeter, Circularity, and Eccentricity extracted from manually drawn ROI on the asparagus phantom. The asparagus cross-section Area measurement illustrates a great similarity between the T2 and the rFOV-PS-EPI values even (SI) near the metal level.

Figure 8:

Figure 8:

Open in a new tab

The slice-wise ROI-parameters extracted from the selected ROI: Area, Perimeter, Circularity, and Eccentricity. Great affinity between the rFOV-PS-EPI and the reference T2 metrics, demonstrating the capacity of the developed technique to provide distortion-reduced DTI images. In addition, no measurements were extracted from data collected using the rFOV-SS-EPI method at the site of the Hardware due to the presence of signal void and high-intensity geometric distortions.

In addition, the Perimeter of the selected ROI on the asparagus displays that the rFOV-SS-EPI values are different compared to the T2 and rFOV-PS-EPI due to the geometric distortion. However, the T2 and rFOV-PS-EPI values are similar except for the SI-9 to SI-14 due to the signal dropout induced by the metal.

The evaluation of the Circularity of the asparagus using the three pulse sequences demonstrates that the proposed acquisition approach provides better circular-shaped phantom with reduced distortion compared to the standard rFOV-SS-EPI method. The measurement of circularity shows good agreement between the rFOV-PS-EPI and the reference T2 data. The loss of the circularity visible at the SI-9 to SI-18 using the suggested technique is mainly due to the signal void where only a portion of the asparagus is detectable, however the geometric distortion remains low as shown in the top row of Figure 5.

Furthermore, the Eccentricity parameter extracted from the selected ROI shows that the standard acquisition method rFOV-SS-EPI is vulnerable to metal-induced artifacts compared to the reference T2 and the rFOV-PS-EPI imaging techniques. The ROI-based extracted parameters demonstrate that the gold standard imaging technique rFOV-SS-EPI is completely hampered by metal artifacts at regions close to the hardware (SI-10 to SI-18), prohibiting carry out any reliable measurements at these slices.

Table 3 exhibits the t-test result of the pairwise comparison between the (rFOV-SS-EPI vs rFOV-PS-EPI), (rFOV-PS-EPI vs T2), and (rFOV-SS-EPI vs T2) for each ROI parameter for slices below (SI-2 to SI-9) and above (SI-19 to SI-28) the metal implant. Significant difference was found for the pair-comparison of the Circularity and Eccentricity of ROI between the rFOV-SS-EPI vs rFOV-PS-EPI and rFOV-SS-EPI vs T2 with (p-value <0.05). The rFOV-PS-EPI vs T2 shows no significant difference for the same parameters, demonstrating that the rFOV-PS-EPI is able to provide a distortion reduction comparable to the structural method.

Table 3:

The P-values of the pairwise comparison of parameters: Area, Perimeter, Circularity, and Eccentricity below and above the metal hardware between the rFOV-PS-EPI vs rFOV-SS-EPI, rFOV-PS-EPI vs T2, and rFOV-SS-EPI vs T2.

Area Perimeter Circularity Eccentricity
Below Above Below Above Below Above Below Above
rFOV-PS-EPI vs rFOV-SS-EPI 0.36403 0.57665 0.23180 0.40079 0.00051 0.00957 0.00298 0.00816
rFOV-PS-EPI vs T2 0.03508 0.40707 0.01279 0.56356 0.88637 0.47230 0.98795 0.03561
rFOV-SS-EPI vs T2 0.04922 0.80364 0.81142 0.26442 0.00041 0.00801 0.00039 0.00082
Open in a new tab

Discussion

The potential of DTI in pre-operative SCI has been demonstrated in several clinical studies 4,5,11,12,25. However, its capability to provide quantitative information of the microstructure on post-surgery SCI cases with metallic hardware remains limited 4. Severe geometric distortions arise due to the presence of metal in the imaging plane or in the adjacent plane, also known respectively as in-plane and through-plane artifacts 20,26. It has been shown that the size, the type and the intensity of metal-induced artifacts increase with magnetic field strength and employed pulse sequence parameters such as the echo spacing and the receiver bandwidth 27. Therefore, in this phantom-based study, a reduced FOV acquisition approach, termed rFOV-PS-EPI, is proposed to mitigate the metal-induced artifacts and to demonstrate the feasibility of performing DTI scans near metallic hardware at ultra-high field strength of 3T.

The b0-images and Trace maps obtained using the standard full FOV pulse sequence (SS-EPI) are dramatically distorted by the susceptibility artifacts (Figure 4, first row), showing evidence of the vulnerability of this method to the metal-induced distortions and its inappropriacy for post-surgery SCI imaging. The readout segmented (RS-EPI) and phase segmented (PS-EPI) Fourier-space sampling scheme allow achieving shorter echo spacing and high bandwidth per pixel, reducing the severity of image distortions in the presence of metal (Figure 4). The asparagus shape detected and zoomed in the red box in Figure 4 illustrates that the PS-EPI pulse sequence addresses the geometric distortions compared to SS-EPI and RS-EPI techniques. However, it suffers from residual susceptibility artifact and geometrics deformations as indicated by arrows in Figure 4.

In order to achieve a shorter echo-spacing and further improve the image quality, the reduced FOV technique was implemented into the PS-EPI method to create the proposed rFOV-PS-EPI pulse sequence. Despite some signal dropout at the level of the metal implant, the developed method shows great potential in obtaining reduced artifacts and high image quality of the asparagus model compared to the imaging techniques: rFOV-SS-EPI, SS-EPI, PS-EPI, and the RS-EPI.

The efficacy of this technique is demonstrated by a significant reduction of in-plane and through-plane image distortion at 5 mm slice thickness, commonly used in imaging protocols for the spinal cord 28. The b0-images and MD maps show that using the rFOV-SS-EPI the asparagus shape can be detected only starting from the fifth slice away from the metal (~ 20 mm) which is in line with previous finding 23. However, it can be partially seen at the level of the spinal hardware and fully visible at the previous slice using the proposed rFOV-PS-EPI method as shown in the red boxes in Figure 5.

In addition, the validity of the proposed pulse sequence to collect metal reduced DTI images on spinal model with metallic hardware was demonstrated by providing less geometric distortion DTI maps and smaller signal void area around the site of the metal (Figure 6 and Figure 7), providing evidence of its potential for post-operative SCI application to extract information at site of injury.

The parameters extracted from the selected ROI at each slice on the asparagus cross-section: Perimeter, Area, Circularity, and Eccentricity display great affinity between the rFOV-PS-EPI and the reference T2 metrics, demonstrating the capacity of the proposed approach to collect distortion-free DTI data. In addition, the measurements show that the data collected using the standard imaging method does not allow extracting indices at the site of the metallic implant due to the presence of signal void and severe geometric distortions. Although there was increase in the loss of Circularity and Eccentricity in regions near the metal (SI-9 to SI-19), the rFOV-PS-EPI method provides visualization of a portion of the asparagus with low geometric distortion (Figure 8).

The pairwise comparison between the T2 vs rFOV-SS-EPI and the rFOV-SS-EPI vs rFOV-PS-EPI Circularity and Eccentricity parameters show a significant difference (p-value < 0.05) below and above the metal hardware Table 3. However, no significant differences were found for these parameters for the comparison between the rFOV-PS-EPI and T2 technique (p-value > 05). In addition, the cross-section area parameter is significantly different at T2 data when it is compared to ADC maps extracted from rFOV-SS-EPI and rFOV-PS-EPI data. This is mainly due to distortion induced by the high-intensity diffusion gradients.

Although the capability of the developed imaging approach in addressing the metal-induced artifacts for DTI scans around the metallic implant, there remain technical limitations that should be highlighted. The ROI-based SNR comparison at b0-images shows that the rFOV-SS-EPI provides significantly better SNR efficiency compared to the proposed rFOV-PS-EPI method with p-value of 0.029 (Table 2), demonstrating the validity of the rFOV-SS-EPI method for pre-operative spinal cord DTI. In addition, several studies have demonstrated that the multi-shot acquisition strategy is a reliable approach for high-resolution DTI, although it has high-sensitivity to motion-induced artifacts 18,29-32. The Inter-shot phase variations can be amplified by the high-intensity diffusion gradients, resulting in severe ghosting artifacts. Multiples correction strategies, including navigator-based sequences 33-36, have been considered for ghosting artifact correction.

The other major drawback of the proposed imaging strategy is the prolonged scan time associated with the multi-shot acquisition scheme 37,38. In this phantom experiment, the scan time of the 8-shot DTI scan was five times longer than the rFOV-SS-EPI. In addition, the cardiac gating technique is commonly used for in vivo DTI scan of the SC for reducing the contribution of the cardiovascular pulsations in signal corruption, resulting in increased total scan time. Furthermore, high angular-resolution DTI as well as advanced diffusion models, such as Diffusion Kurtosis Imaging (DKI) and Neurite Orientation Density Dispersion Imaging (NODDI), 9,39 to fully characterize the SC microstructure enables thorough SC post-operative assessment, albeit an extended acquisition time. Therefore, acceleration acquisition techniques such as the Simultaneous Multi-Slice 40 must be implemented for reducing the scan time for clinical applications.

Additional correction techniques such as View-Angle-Tilting (VAT) or Slice Encoding for Metal Artifact Correction (SEMAC) could be implemented into the proposed pulse sequence for improving the capacity of the in-plane and through slices metal artifacts suppression, providing better measurement of DTI metrics at the site of injury.

Conclusion

While the reduced FOV (rFOV-SS-EPI) acquisition technique has been conventionally used to perform high resolution DTI on SC, its efficacy for post-operative SCI evaluation remains limited by the presence of artifacts due to metal implants. Here, we introduce a technique based on the combination of the reduced FOV strategy and multi-shot acquisition scheme. The phantom-based results are very promising and demonstrate the benefit of the proposed acquisition method in achieving high-resolution DTI with reduced geometric distortion near the metal-based spinal hardware at 3T. Future work will involve validation of this sequence in post-operative patients with metal implants as well as optimizations to address the remaining technical limitations.

Supplementary Material

1

Supplementary Figure 1: Example of the selected ROI on rFOV-PS-EPI and T2 images to extract the parameters: Area, Perimeter, Circularity, and Eccentricity for pairwise comparison between the structural and diffusion data.

Acknowledgments

We gratefully acknowledge Maya Polackal for assistance in writing and editing the manuscript.

Funding

This work was supported by the National Institute of Neurological Disorders and Stroke (NINDS) of the National Institutes of Health (NIH) under award number R01NS111113 (Thomas Jefferson University, Philadelphia, PA).

Abbreviations

MRI

Magnetic Resonance Imaging

DTI

Diffusion Tensor Imaging

SC

Spinal Cord

FA

Fractional Anisotropy

MD

Mean Diffusivity

AD

Axial Diffusivity

RD

Radial Diffusivity

SCI

Spinal Cord Injury

SNR

Signal-to-Noise Ratio

EPI

Echo Planar Imaging

SS-EPI

Single-Shot EPI

PS-EPI

Phase-Segmented EPI

RS-EPI

Readout-Segmented EPI

rFOV

Field-Of-View

ROI

Region-Of-Interest

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Competing interests

We have no conflict of interests to declare.

Data Availability

Collected T2, Diffusion MRI images, computed maps such as FA, MD will be made available by Slimane Tounekti upon request.

References:

  • 1.Saksena S, Mohamed FB, Middleton DM, et al. Diffusion Tensor Imaging Assessment of Regional White Matter Changes in the Cervical and Thoracic Spinal Cord in Pediatric Subjects. J Neurotrauma. 2019;36(6):853–861. doi: 10.1089/neu.2018.5826 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Muller J, Middleton D, Alizadeh M, et al. Hybrid diffusion imaging reveals altered white matter tract integrity and associations with symptoms and cognitive dysfunction in chronic traumatic brain injury. NeuroImage Clin. 2021;30(October 2020):102681. doi: 10.1016/j.nicl.2021.102681 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Pierpaoli C, Basser PJ. Toward a quantitative assessment of diffusion anisotropy. Magn Reson Med. 1996;36(6):893–906. doi: 10.1002/mrm.1910360612 [DOI] [PubMed] [Google Scholar]
  • 4.Shabani S, Kaushal M, Budde MD, Wang MC, Kurpad SN. Diffusion tensor imaging in cervical spondylotic myelopathy: a review. J Neurosurg Spine. 2020;33(1):65–72. doi: 10.3171/2019.12.SPINE191158 [DOI] [PubMed] [Google Scholar]
  • 5.Gao SJ, Yuan X, Jiang XY, et al. Correlation study of 3T-MR-DTI measurements and clinical symptoms of cervical spondylotic myelopathy. Eur J Radiol. 2013;82(11):1940–1945. doi: 10.1016/j.ejrad.2013.06.011 [DOI] [PubMed] [Google Scholar]
  • 6.Ellingson BM, Salamon N, Woodworth DC, Yokota H, Holly LT. Reproducibility, temporal stability, and functional correlation of diffusion MR measurements within the spinal cord in patients with asymptomatic cervical stenosis or cervical myelopathy. J Neurosurg Spine. 2018;28(5):472–480. doi: 10.3171/2017.7.SPINE176 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Mamata H, Jolesz FA, Maier SE. Apparent diffusion coefficient and fractional anisotropy in spinal cord: Age and cervical spondylosis-related changes. J Magn Reson Imaging. 2005;22(1):38–43. doi: 10.1002/jmri.20357 [DOI] [PubMed] [Google Scholar]
  • 8.Toktas ZO, Tanrikulu B, Koban O, Kilic T, Konya D. Diffusion tensor imaging of cervical spinal cord: A quantitative diagnostic tool in cervical spondylotic myelopathy. J Craniovertebr Junction Spine. 2016;7(1):26–30. doi: 10.4103/0974-8237.176617 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Liu Z, Bian B, Wang G, et al. Evaluation of microstructural changes in spinal cord of patients with degenerative cervical myelopathy by diffusion kurtosis imaging and investigate the correlation with JOA score. BMC Neurol. 2020;20(1):1–8. doi: 10.1186/s12883-020-01752-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Nouh MR. Spinal fusion-hardware construct: Basic concepts and imaging review. World J Radiol. 2012;4(5):193. doi: 10.4329/wjr.v4.i5.193 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Kara B, Celik A, Karadereler S, et al. The role of DTI in early detection of cervical spondylotic myelopathy: A preliminary study with 3-T MRI. Neuroradiology. 2011;53(8):609–616. doi: 10.1007/s00234-011-0844-4 [DOI] [PubMed] [Google Scholar]
  • 12.Maki S, Koda M, Kitamura M, et al. Diffusion tensor imaging can predict surgical outcomes of patients with cervical compression myelopathy. Eur Spine J. 2017;26(9):2459–2466. doi: 10.1007/s00586-017-5191-7 [DOI] [PubMed] [Google Scholar]
  • 13.Krisa L, Middleton DM, Saksena S, et al. Clinical Utility of Diffusion Tensor Imaging as a Biomarker to Identify Microstructural Changes in Pediatric Spinal Cord Injury. Top Spinal Cord Inj Rehabil. 2022;28(2):1–12. doi: 10.46292/sci21-00048 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Rieseberg S, Frahm J, Finsterbusch J. Two-dimensional spatially-selective RF excitation pulses in echo-planar imaging. Magn Reson Med. 2002;47(6):1186–1193. doi: 10.1002/mrm.10157 [DOI] [PubMed] [Google Scholar]
  • 15.Finsterbusch J High-resolution diffusion tensor imaging with inner field-of-view EPI. J Magn Reson Imaging. 2009;29(4):987–993. doi: 10.1002/jmri.21717 [DOI] [PubMed] [Google Scholar]
  • 16.Saritas EU, Cunningham CH, Lee JH, Han ET, Nishimura DG. DWI of the spinal cord with reduced FOV single-shot EPI. Magn Reson Med. 2008;60(2):468–473. doi: 10.1002/mrm.21640 [DOI] [PubMed] [Google Scholar]
  • 17.Li Z, Wu G, Zhao X, Luo F, Li SJ. Multiecho segmented EPI with z-shimmed background gradient compensation (MESBAC) pulse sequence for fMRI. Magn Reson Med. 2002;48(2):312–321. doi: 10.1002/mrm.10219 [DOI] [PubMed] [Google Scholar]
  • 18.Porter DA, Heidemann RM. High resolution diffusion-weighted imaging using readout-segmented echo-planar imaging, parallel imaging and a two-dimensional navigator-based reacquisition. Magn Reson Med. 2009;62(2):468–475. doi: 10.1002/mrm.22024 [DOI] [PubMed] [Google Scholar]
  • 19.Chang HC, Sundman M, Petit L, et al. Human brain diffusion tensor imaging at submillimeter isotropic resolution on a 3Tesla clinical MRI scanner. Neuroimage. 2015;118:667–675. doi: 10.1016/j.neuroimage.2015.06.016 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Germann C, Falkowski AL, Von Deuster C, Nanz D, Sutter R. Basic and Advanced Metal-Artifact Reduction Techniques at Ultra-High Field 7-T Magnetic Resonance Imaging - Phantom Study Investigating Feasibility and Efficacy. Invest Radiol. 2022;57(6):387–398. doi: 10.1097/RLI.0000000000000850 [DOI] [PubMed] [Google Scholar]
  • 21.Tung YH, In MH, Ahn S, Speck O. Rapid Geometry-Corrected Echo-Planar Diffusion Imaging at Ultrahigh Field: Fusing View Angle Tilting and Point-Spread Function Mapping. Magn Reson Med. 2022;88(5):2074–2087. doi: 10.1002/mrm.29360 [DOI] [PubMed] [Google Scholar]
  • 22.Li S, Wang Y, Hu Z, et al. High-fidelity diffusion tensor imaging of the cervical spinal cord using point-spread-function encoded EPI. Neuroimage. 2021;236(January). doi: 10.1016/j.neuroimage.2021.118043 [DOI] [PubMed] [Google Scholar]
  • 23.Yang L, Liu Y, Kong X, et al. Diffusion tensor magnetic resonance imaging of the postoperative spine with metallic implants. NMR Biomed. 2020;33(8):1–12. doi: 10.1002/nbm.4321 [DOI] [PubMed] [Google Scholar]
  • 24.Stejskal EO, Tanner JE. Spin diffusion measurements: Spin echoes in the presence of a time-dependent field gradient. J Chem Phys. 1965;42(1):288–292. doi: 10.1063/1.1695690 [DOI] [Google Scholar]
  • 25.Matsuda Y, Miyazaki K, Tada K, et al. Increased MR signal intensity due to cervical myelopathy. J Neurosurg. 2009;74(6):887–892. doi: 10.3171/jns.1991.74.6.0887 [DOI] [PubMed] [Google Scholar]
  • 26.Sutter R, Ulbrich EJ, Jellus V, Nittka M, Pfirrmann CWA. Reduction of metal artifacts in patients with total hip arthroplasty with slice-encoding metal artifact correction and view-angle tilting MR imaging. Radiology. 2012;265(1):204–214. doi: 10.1148/radiol.12112408 [DOI] [PubMed] [Google Scholar]
  • 27.Germann C, Nanz D, Sutter R. Magnetic Resonance Imaging Around Metal at 1.5 Tesla: Techniques From Basic to Advanced and Clinical Impact. Invest Radiol. 2021;56(11):734–748. doi: 10.1097/RLI.0000000000000798 [DOI] [PubMed] [Google Scholar]
  • 28.Cohen-Adad J, Alonso-Ortiz E, Abramovic M, et al. Generic acquisition protocol for quantitative MRI of the spinal cord. Nat Protoc. 2021;16(10):4611–4632. doi: 10.1038/s41596-021-00588-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Dai E, Liu S, Guo H. High-resolution whole-brain diffusion MRI at 3T using simultaneous multi-slab (SMSlab) acquisition. Neuroimage. 2021;237(May):118099. doi: 10.1016/j.neuroimage.2021.118099 [DOI] [PubMed] [Google Scholar]
  • 30.Miller KL, Pauly JM. Nonlinear phase correction for navigated diffusion imaging. Magn Reson Med. 2003;50(2):343–353. doi: 10.1002/mrm.10531 [DOI] [PubMed] [Google Scholar]
  • 31.Tounekti S, Troalen T, Bihan-Poudec Y, et al. High-resolution 3D diffusion tensor MRI of anesthetized rhesus macaque brain at 3T. Neuroimage. 2018;181(March):149–161. doi: 10.1016/J.NEUROIMAGE.2018.06.045 [DOI] [PubMed] [Google Scholar]
  • 32.Chang HC, Guhaniyogi S, Chen NK. Interleaved diffusion-weighted improved by adaptive partial-Fourier and multiband multiplexed sensitivity-encoding reconstruction. Magn Reson Med. 2015;73(5):1872–1884. doi: 10.1002/mrm.25318 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Ordidge RJ, Helpern JA, Qing ZX, Knight RA, Nagesh V. Correction of motional artifacts in diffusion-weighted MR images using navigator echoes. Magn Reson Imaging. 1994;12(3):455–460. doi: 10.1016/0730-725X(94)92539-9 [DOI] [PubMed] [Google Scholar]
  • 34.Butts K, Crespigny A De, Pauly JM, Moseley M. Diffusion-Weighted interleaved Echo-Planar Imaging with a Pair of Orthogonal Navigator Echoes. Magn Reson Imaging. 1996;35:763–770. doi: 10.1002/mrm.1910350518 [DOI] [PubMed] [Google Scholar]
  • 35.Pipe JP. Motion correction with {PROPELLER} {MRI}: {A}pplication to head motion and free-breating cardiac imaging. Magn Reson Med. 1999;42(5):963–969. [DOI] [PubMed] [Google Scholar]
  • 36.Liu W, Zhao X, Ma Y, Tang X, Gao JH. DWI using navigated interleaved multishot EPI with realigned GRAPPA reconstruction. Magn Reson Med. 2016;75(1):280–286. doi: 10.1002/mrm.25586 [DOI] [PubMed] [Google Scholar]
  • 37.Frost R, Jezzard P, Clare S, Porter DA, Miller KL. Scan Time Reduction for Readout-Segmented EPI Using Simultaneous Multislice Acceleration: Diffusion-Weighted Imaging at 3 and 7 Tesla. 2015;149:136–149. doi: 10.1002/mrm.25391 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Chen N kuei, Guidon A, Chang HC, Song AW. A robust multi-shot scan strategy for high-resolution diffusion weighted MRI enabled by multiplexed sensitivity-encoding (MUSE). Neuroimage. 2013;72:41–47. doi: 10.1016/j.neuroimage.2013.01.038 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Yu J, Sun Y, Cao G, Zheng X, Jing Y, Li C. Diffusional kurtosis imaging in evaluation of microstructural changes of spinal cord in cervical spondylotic myelopathy feasibility study. Medicine (Baltimore). 2020;99(47):e23300. doi: 10.1097/MD.0000000000023300 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Setsompop K, Fan Q, Stockmann J, et al. High-resolution in vivo diffusion imaging of the human brain with generalized slice dithered enhanced resolution: Simultaneous multislice (gSlider-SMS). Magn Reson Med. 2018;79(1):141–151. doi: 10.1002/mrm.26653 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1

Supplementary Figure 1: Example of the selected ROI on rFOV-PS-EPI and T2 images to extract the parameters: Area, Perimeter, Circularity, and Eccentricity for pairwise comparison between the structural and diffusion data.

NIHMS1943566-supplement-1.tif (3.8MB, tif)

Data Availability Statement

Collected T2, Diffusion MRI images, computed maps such as FA, MD will be made available by Slimane Tounekti upon request.

RESOURCES