Note: This copy is for your personal, non-commercial use only. To order presentation-ready copies for distribution to your colleagues or clients, cont

Transcription

1 Note: This copy is for your personal, non-commercial use only. To order presentation-ready copies for distribution to your colleagues or clients, contact us at Jang Gyu Cha, MD Wook Jin, MD Min Hee Lee, MD Dong Hoon Kim, MD Jai Soung Park, MD Won Han Shin, MD Boem Ha Yi, MD Reducing Metallic Artifacts in Postoperative Spinal Imaging: Usefulness of IDEAL Contrastenhanced T1- and T2-weighted MR Imaging Phantom and Clinical Studies 1 Purpose: To prospectively compare the effectiveness of iterative decomposition of water and fat with echo asymmetry and least-squares estimation (IDEAL) T2-weighted and contrast material enhanced T1-weighted magnetic resonance (MR) imaging with that of a conventional MR imaging protocol in minimizing metallic artifacts in phantoms and in patients with metallic hardware after spinal surgery. ORIGINAL RESEARCH n TECHNICAL DEVELOPMENTS 1 From the Departments of Radiology (J.G.C., M.H.L., D.H.K., J.S.P., B.H.Y.) and Neurosurgery (W.H.S.), Soonchunhyang University Hospital, 1174 Jung-Dong, Wonmi-Gu, Bucheon-Si, Gyeonggi-Do Korea; and Department of Radiology, East-West Neomedical Center, Seoul, Korea (W.J.). Received September 25, 2010; revision requested October 27; revision received November 23; accepted December 3; fi nal version accepted December 15. Address correspondence to J.G.C. ( mj4907@schmc.ac.kr ). q RSNA, 2011 Materials and Methods: Results: Conclusion: Institutional review board approval and informed consent were obtained for this study. Coronal T1- and T2-weighted MR images of six titanium alloy pedicle screws in an oil bath containing tubes filled with diluted gadolinium contrast medium were obtained with frequency-selective fat saturation (FSFS) and IDEAL. Axial T2-weighted and contrast-enhanced T1-weighted MR imaging of the spine was performed with FSFS and IDEAL at 22 lumbar levels in 19 patients. Two musculoskeletal radiologists qualitatively analyzed the images in terms of the visualization of paravertebral muscle and the spinal canal region, uniformity of fat saturation, and noise. The paired images were rated by using a five-point scale. For the quantitative study with phantoms, the short- and long-axis lengths of metallic artifacts were determined on signal intensity profiles. In the phantom study, metallic artifact size was markedly decreased in the IDEAL T2-weighted and contrast-enhanced T1-weighted images ( P,.001). In the clinical study, IDEAL T2-weighted and contrast-enhanced T1-weighted images enabled significantly improved visualization of the dural sac ( P,.001), spinal muscles ( P,.05), uniformity of fat saturation ( P,.001), and noise ( P,.05). IDEAL T2-weighted and contrast-enhanced T1-weighted MR imaging examinations effectively reduce the degree of tissueobscuring artifacts produced by spinal fixation hardware and improve image quality compared with FSFS T2-weighted and contrast-enhanced T1-weighted MR imaging. q RSNA, 2011 Supplemental material: /suppl/doi: /radiol /-/dc1 Radiology: Volume 259: Number 3 June 2011 n radiology.rsna.org 885

2 Magnetic resonance (MR) imaging plays a vital role in the postoperative imaging of patients who have undergone spinal instrumentation procedures ( 1 ). However, magnetic susceptibility artifacts from metallic devices remain problematic in evaluating the spine postoperatively, even in patients with newly developed implants made of titanium alloys that produce fewer artifacts ( 2 ). Incomplete fat saturation represents an additional challenge in evaluating the postoperative spine with metallic hardware ( 3 ). To resolve these problems, a number of techniques to minimize susceptibility to artifacts have been proposed ( 2,4,5 ). The recently developed iterative decomposition of water and fat with echo asymmetry and least-squares estimation (IDEAL) technique is a three-point waterfat separation method that uses asymmetric echoes and least-squares fitting to achieve the maximum possible signalto-noise ratio (SNR) performance ( 3,6,7 ). IDEAL can be used for robust separation of fat and water with very high SNR efficiency ( 7 ). In contrast to conventional fat-saturation methods, IDEAL is insensitive to magnetic field (B0 and B1) inhomogeneity ( 8 ). The present study was performed to prospectively compare the effectiveness of IDEAL T2-weighted and contrast material enhanced T1-weighted imaging with that of a conventional MR imaging Advances in Knowledge n The main reason for artifact reduction in the iterative decomposition of water and fat with echo asymmetry and leastsquares estimation (IDEAL) fast spin-echo MR images was the large decrease in the peripheral rim of high signal intensity. n IDEAL imaging was useful in obtaining a clearer view of the dural sac and paravertebral muscles. n In contrast to short inversion time inversion-recovery imaging, IDEAL imaging can be used with gadolinium contrast enhancement. protocol in minimizing metallic artifacts in phantoms and in patients with metallic hardware after spinal surgery. Materials and Methods GE Healthcare (Waukesha, Wis) provided research support in the development of our IDEAL test sequence. Phantom Study A phantom study was performed by using six different titanium alloy pedicle screws that are currently used for treatment at our institution (Soonchunhyang University Bucheon Hospital) (Table E1 [online]). The phantom was constructed as a dual box-in-a-box structure (Appendix E1 [online]). The screws were placed at the center of the inner container between the second and third rows of test tubes ( Fig 1 ). The phantom study consisted of seven MR imaging studies one without a screw and the remaining six with one of six different pedicle screws each. Clinical Study Patients. Our Institutional Review Board approved the research proposal. For the clinical study, all patients were recruited from the Department of Neurosurgery at our institution between November 2008 and May All subjects gave their informed consent after receiving a full explanation of the nature of the study. The inclusion criteria were as follows: Patients had to (a) have undergone lumbar spinal instrumentation surgery with titanium alloy pedicle screws and (b) be 18 years of age or older. The exclusion criteria were as follows: (a) scoliosis involving a vertebral column curvature greater than 15, (b) ex- Implication for Patient Care n The reduction in image artifacts achieved with the IDEAL technique enhances visualization of the spine and surrounding structures, thereby potentially increasing the amount of pertinent clinical information acquired, which may subsequently aid in patient care. tensive vertebral fractures resulting in difficulty in assessing axial images, and (c) severe postoperative infection of a kind that could disrupt the border of the dural sac and paraspinal muscle. Between November 2008 and May 2009, 19 patients (eight men [mean age, 58.8 years; range, years] and 11 women [mean age, 60.6 years; range, years]) who had undergone spinal surgery with metallic hardware and were undergoing contrast-enhanced spinal MR imaging at our institution were recruited for the prospective clinical study. The mean postoperative interval between the date of surgery and the date of imaging was 31.6 months (range, 2 71 months). Twenty-two spinal levels in 19 patients with metallic hardware were included in this clinical study: one level of T12-L1, one level of L1-L2, four levels of L3-L4, 13 levels of L4-L5, and three levels of L5-S1. Data acquisition. The phantom and clinical studies were performed by using a 1.5-T MR imaging unit (Signa HDxt; GE Medical Systems, Milwaukee, Wis) and a dedicated phased-array spine coil. All MR images were obtained parallel to the long axis of the fixation screw. All MR imaging studies consisted of a conventional group, in which frequency-selective fatsuppressed T1- and T2-weighted fast spinecho (FSE) sequences were performed, Published online before print /radiol Radiology 2011; 259: Abbreviations: FSE = fast spin-echo IDEAL = iterative decomposition of water and fat with echo asymmetry and least-squares estimation SNR = signal-to-noise ratio Author contributions: Guarantors of integrity of entire study, J.G.C., M.H.L.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript fi nal version approval, all authors; literature research, J.G.C., M.H.L., D.H.K.; clinical studies, J.G.C., W.J., M.H.L., D.H.K., W.H.S.; experimental studies, J.G.C., B.H.Y.; statistical analysis, J.G.C.; and manuscript editing, J.G.C., D.H.K., J.S.P. Potential confl icts of interest are listed at the end of this article. 886 radiology.rsna.org n Radiology: Volume 259: Number 3 June 2011

3 Figure 1 Figure 1: Phantom model. (a) Photograph and (b) coronal IDEAL T2-weighted MR image of phantom show its box-in-a-box structure. Twenty test tubes containing diluted gadolinium solution were placed in the inner container box, which was fi lled with soybean oil to produce fat-saturated images. The outer container box was fi lled with normal saline. and an IDEAL group, in which IDEAL T1- and T2-weighted FSE sequences were performed. Axial MR imaging was performed in the phantom study, and coronal MR imaging was performed in the clinical study. Table E2 (online) summarizes the imaging parameters of all sequences. The frequency encoding direction was right to left in the phantom study and anterior to posterior in the clinical study. Quantitative Study For each imaging sequence, a center image bearing the center axis of the screw was selected among the coronal images, and identical regions of interest (ROIs) were manually located by J.G.C. (with 8 years of experience in musculoskeletal imaging) over each test tube on the phantom. ROIs varied in size from 49 to 50 mm 2 (mean, 49.8 mm 2 ). So that we could quantitatively assess the ability of IDEAL to preserve signal intensities, we compared the SNRs of the 20 test tubes in the center coronal image with and that without the pedicle screws for each MR imaging sequence. The SNR was calculated by dividing the signal intensity by the mean value of the background intensity ( 9 ), which was measured from the ROI outside the phantom in a region free of artifacts. Artifacts were measured on the coronal images, which allowed determination of the distribution of signal loss along the screw surface ( Fig 2 ). Five continuous sections of the coronal images were selected for measurement of screw length. In each image, the short-axis length of the pedicle screw was measured at three different levels of the head and midbody. Long-axis length was also measured. Then, profiles of the signal intensity along the short axis of the head and body of the screw and along the long axis were created by using threedimensional software (Lucion; Infinitt Healthcare, Seoul, Korea). The metallic artifacts were defined as areas of signal void surrounded by a rim of high signal amplitude that followed the shape of the titanium screw ( 2 ). Therefore, we measured the width between the sharp signal spikes on the profile, which were seen as peripheral rims of high signal intensity on MR images. This measurement by using profiles allowed window width and center independent measurements of artifact size ( 10 ) ( Fig 3 ). All measurements were performed by one radiologist (M.H.L., with 1 year of experience in musculoskeletal imaging) by using this technique. Qualitative Study Qualitative assessment of all images was performed independently by two musculoskeletal radiologists with expertise in MR imaging (M.H.L. and W.J. [with 10 years of experience in musculoskeletal imaging]). The readers were blinded as to the sides on which the IDEAL and conventional groups were located. Analysis of each set was performed with an independent Digital Imaging and Communications in Medicine viewer (PiView Star, version 5; Infinitt Healthcare). Three axial sections including the level of the end plate, the inter vertebral disc, and the pedicle were selected for qualitative eval uation. The readers independently reviewed the following two sets of axial MR images in random order: (a) IDEAL gadolinium-enhanced T1- and T2-weighted images (IDEAL set) and (b) fat-saturated gadoliniumenhanced T1- and T2-weighted images (conventional set). To minimize learning bias, patients names, ages, and identification numbers and the imaging parameters were hidden during the review. Intervals of at least 2 weeks separated the reviews of the two sets of images. For qualitative analysis, four key components visibility of paravertebral muscle, visibility of the dural sac, Radiology: Volume 259: Number 3 June 2011 n radiology.rsna.org 887

4 Figure 2 Figure 2: Phantom profile of a pedicle screw. (a) Coronal IDEAL T1-weighted MR image of phantom shows the six short-axis lines perpendicular to the pedicle screw (arrow) and one long-axis line parallel to the screw used to obtain the profi le view. (b) Profi le view of the midportion of the screw head selected shows the metallic artifact, defi ned as the distance between the two points (arrows) of peak signal intensity determined along the line through the phantom. Figure 3 Figure 3: Coronal MR images of pedicle screws. (a) Fat-saturated T1-weighted image. (b) IDEAL T1-weighted image. (c) Fat-saturated T2-weighted image. (d) IDEAL T2-weighted image. Note that b and d show markedly decreased metallic artifacts as compared with a and c. A major reason for the metallic artifact reduction was improvement in the peripheral area of high signal intensity around the pedicle screw. 888 radiology.rsna.org n Radiology: Volume 259: Number 3 June 2011

5 homo geneity of fat suppression, and noise in all images were evaluated by using five-point scales. (a) The visibility of paravertebral muscle in the MR image, with separate evaluation of the right and left paravertebral muscles, was graded as follows: A grade of 0 indicated no visibility; a grade of 1, image distortion in more than 75% of the paravertebral muscles; a grade of 2, image distortion in 50% 75% of the paravertebral muscles; a grade of 3, image distortion in 25% 50% of the paravertebral muscles; and a grade of 4, no distortion. (b) The visibility of the dural sac was graded as follows: A grade of 0 indicated no visibility; a grade of 1, that only one quadrant was visible; a grade of 2, that two quadrants were visible; a grade of 3, that three quadrants were visible; and a grade of 4, that the whole dural sac was visible. (c) Homogeneity of fat suppression was graded as follows: A grade of 0 indicated incomplete fat saturation in the whole image; a grade of 1, incomplete fat saturation in one quadrant; a grade of 2, incomplete fat saturation in two quadrants; a grade of 3, incomplete fat saturation in three quadrants; and a grade of 4, homogeneous fat saturation in the whole image. (d) Noise in addition to the metallic artifact was graded as follows: a grade of 0 indicated that there was no artifact-free quadrant; a grade of 1, that there was one artifact-free quadrant; a grade of 2, that there were two artifact-free quadrants; a grade of 3, that there were three artifact-free quadrants; and a grade of 4, that the whole image was free of artifacts. Table 1 Results of Comparison of Mean Lengths of Metallic Artifact at Long and Short Axis of Pedicle Screws between Conventional Fat-suppressed FSE MR Images and IDEAL FSE MR Images Pulse Sequence Head Length of Short Axis (mm) Length of Long Axis (mm) Fat-saturated T1 weighted (46.06, 51.92) (17.01, 20.19) (93.01, 98.51) IDEAL T1 weighted (23.00, 24.66) * 9.88 (9.15, 10.62) * (68.62, 73.10) * Fat-saturated T2 weighted (44.30, 50.12) (12.62, 15.76) (83.94, 89.43) IDEAL T2 weighted (24.25, 26.59) * 9.14 (8.24, 10.04) * (67.80, 72.77) * Body Note. Data are means, with 95% confi dence intervals in parentheses. * P,. 05 for comparison with data for conventional protocol. Table 2 Comparison of Differences in SNR of Contrast Agent filled Vials between FSE MR Imaging with and That without Pedicle Screws for Each Pulse Sequence Name of Pedicle Screw and Sequence T1-weighted Imaging T2-weighted Imaging MEGA Conventional 0.47 (0.26, 0.68) 2.69 (2.29, 3.09) IDEAL 0.34 (0.17, 0.52) * 0.36 (0.23, 0.49) * GSS Conventional 0.51 (0.31, 0.71) 1.76 (1.48, 2.04) IDEAL 0.24 (0.14, 0.33) * 0.31 (0.21, 0.40) * Perfi x Conventional 0.66 (0.38, 0.95) 2.01 (1.82, 2.33) IDEAL 0.14 (0.07, 0.21) * 0.29 (0.16, 0.41) * Novel Conventional 0.43 (0.27, 0.59) 2.22 (2.04, 2.40) IDEAL 0.34 (0.26, 0.42) 0.22 (0.15, 0.30) * Legacy Conventional 1.35 (0.73, 1.96) 1.70 (0.80, 2.60) IDEAL 0.32 (0.22, 0.43) * 0.14 (0.07, 0.21) * Optima Conventional 0.53 (0.37, 0.69) 1.52 (1.18, 1.86) IDEAL 0.37 (0.20, 0.53) * 1.17 (0.09, 0.25) * Note. Data are means, with 95% confi dence intervals in parentheses. * P,. 05 for comparison with data for conventional protocol. Statistical Analysis Quantitative analysis. The pedicle screw widths in the test tubes were compared between the conventional and IDEAL image sets by using two-tailed paired t tests with 95% confidence limits. SNR differences between the two image sets were compared by using the paired signed rank test. Qualitative analysis. Interobserver agreement for the four categories of visibility of the paravertebral muscle, visibility of the dural sac, homogeneity of fat suppression, and noise in addition to the metallic artifact was assessed with the weighted statistic. The k values were interpreted as suggested by Landis and Koch ( 11 ). The individual scores assigned by the two radiologists for the T1- and T2-weighted imaging sequences performed with and those performed without use of the IDEAL technique were also compared in a pairwise manner with respect to the above five categories by using the paired t test. All statistical tests were performed with statistical software (Stata, version 9.0; Stata, College Station, Tex). All reported P values are from two-sided tests. In all analyses, P,.05 was considered to indicate a statistically significant difference. Results No patients were excluded from the study on the basis of our exclusion criteria or the quality of the MR imaging examination. Radiology: Volume 259: Number 3 June 2011 n radiology.rsna.org 889

6 Table 3 Image Quality Scores for Comparison of Fat-saturated Contrast-enhanced T1-weighted Images Obtained with IDEAL Technique versus Those Obtained without Rated Feature, Side, and Pulse Sequence Reviewer Reader 1 Reader 2 Visualization of paravertebral muscle Right Fat-saturated contrast-enhanced T1 weighted 2.35 (2.14, 2.56) 2.17 (1.98, 2.36) 0.58 IDEAL contrast-enhanced T1 weighted 2.98 (2.87, 3.10) 2.79 (2.64, 2.93) 0.42 Left Fat-saturated contrast-enhanced T1 weighted 2.39 (2.14, 2.64) 2.09 (1.87, 2.31) 0.49 IDEAL contrast-enhanced T1 weighted 2.97 (2.86, 3.08) 2.73 (2.58, 2.87) 0.42 Visualization of dural sac Fat-saturated contrast-enhanced T1 weighted 1.70 (1.24, 2.16) 2.05 (1.58, 2.51) 0.76 IDEAL contrast-enhanced T1 weighted 3.65 (3.45, 3.86) 3.70 (3.52, 3.87) 0.48 Uniformity of fat saturation Fat-saturated contrast-enhanced T1 weighted 3.02 (2.69, 3.34) 3.03 (2.70, 3.36) 0.60 IDEAL contrast-enhanced T1 weighted 3.86 (3.75, 3.97) 3.78 (3.65, 3.92) 0.47 Noise Fat-saturated contrast-enhanced T1 weighted 2.09 (2.01, 2.17) 2.03 (1.97, 2.09) 0.58 IDEAL contrast-enhanced T1 weighted 3.55 (3.37, 3.72) 3.67 (3.52, 3.82) 0.55 k Value * Note. Unless otherwise specified, data are means, with 95% confidence intervals in parentheses. Images were rated on a five-point scale, as described in the Materials and Methods section. * The k values were reported as follows: k, 0.20 indicated poor agreement; k = , fair agreement; k = , moderate agreement; k = , substantial agreement; and k = , almost perfect agreement. P,. 001 for comparison with data for conventional protocol by using paired t test. Table 4 Image Quality Scores for Comparison of Fat-saturated T2-weighted Images Obtained with IDEAL Technique versus Those Obtained without Rated Feature, Side, and Pulse Sequence Reviewer Reader 1 Reader 2 Visualization of paravertebral muscle Right Fat-saturated T2 weighted 2.5 (2.31, 2.69) 2.14 (1.91, 2.36) 0.60 IDEAL T2 weighted 2.88 (2.79, 2.97) 2.73 (2.61, 2.85) 0.46 Left Fat-saturated T2 weighted 2.42 (2.20, 2.64) 2.03 (1.79, 2.27) 0.51 IDEAL T2 weighted 3.20 (3.03, 3.36) 2.89 (2.70, 3.09) 0.48 Visualization of dural sac Fat-saturated T2 weighted 1.38 (0.95, 1.81) 1.29 (0.88, 1.69) 0.91 IDEAL T2 weighted 3.53 (3.26, 3.80) 3.82 (3.65, 3.98) 0.52 Uniformity of fat saturation Fat-saturated T2 weighted 1.77 (1.38, 2.16) 1.89 (1.51, 2.28) 0.52 IDEAL T2 weighted 3.90 (3.82, 3.99) 3.88 (3.78, 3.98) 0.69 Noise Fat-saturated T2 weighted 3.09 (2.85, 3.33) 2.90 (2.69, 3.13) 0.67 IDEAL T2 weighted 3.71 (3.57, 3.85) 3.85 (3.73, 3.96) 0.49 k Value * Note. Unless otherwise specifi ed, data are means, with 95% confi dence intervals in parentheses. Images were rated on a fi vepoint scale, as described in the Materials and Methods section. * The k values were reported as follows: k, 0.20 indicated poor agreement; k = , fair agreement; k = , moderate agreement; k = , substantial agreement; and k = , almost perfect agreement. P,. 001 for comparison with data for conventional protocol by using paired t test. Quantitative Study Table 1 summarizes the results of pedicle screw length in the phantom. The lengths of the pedicle screws measured on MR images were greater than those of the actual screws for all MR imaging sequences. The mean lengths of pedicle screws on the IDEAL FSE images were significantly shorter than those on the conventional fat-suppressed FSE images ( P,.001). In comparison, the IDEAL FSE images showed marked improvement in reducing metallic artifacts. Table 2 shows the differences in the SNRs of the contrast agent filled vials between FSE imaging with and that without pedicle screws for each pulse sequence. Changes in the SNRs of vials on the IDEAL T1- and T2-weighted FSE images were significantly less than those on the fat-suppressed T1- and T2-weighted FSE images ( P,.05), except for one pedicle screw on one T1- weighted image ( P =.3), for which IDEAL imaging showed a decreased metallic artifact. Qualitative Study Tables 3 and 4 show image quality scores for comparison of the IDEAL T2-weighted and contrast-enhanced T1-weighted im ages with conventional fat-saturated T2-weighted and contrastenhanced T1-weighted images. The IDEAL contrast-enhanced T1- and T2- weighted FSE images had significantly better scores for visualization of both sides of the paravertebral muscle than the conventional fat-saturated contrastenhanced T1- and T2-weighted FSE images, according to both readers ( P,.001) ( Fig 4 ). Visibility of the dural sac and noise were markedly improved in the IDEAL contrast-enhanced T1- and T2-weighted FSE images compared with the fatsaturated contrast-enhanced T1- and T2-weighted FSE images ( P,.001) (Figs 5, 6). For uniformity of fat saturation, the mean scores were for fat-saturated T1-weighted FSE imaging, which were significantly lower than the scores for IDEAL T1-weighted imaging ( , P,.001), while the mean scores for fat-saturated T2- weighted FSE imaging were , 890 radiology.rsna.org n Radiology: Volume 259: Number 3 June 2011

7 Figure 4 Figure 4: Axial MR images in 69-year-old man with interbody fusion and posterior fi xation in L4-L5. (a) Fat-saturated contrast-enhanced T1-weighted image does not show nerve roots (arrows) in the neural foramina owing to metallic artifacts, whereas (b) the nerve roots (arrows) are clearly seen on an IDEAL contrast-enhanced T1- weighted image. (c) Fat-saturated T2-weighted image demonstrates failure of fat suppression in paravertebral muscle (arrows) caused by metallic devices. (d) IDEAL T2-weighted image shows preservation of the signal intensity of the paravertebral muscle (arrows). Figure 5 Figure 5: Axial MR images in 61-year-old woman who underwent posterior fi xation surgery in L3-L4 because of spinal stenosis. (a) Fat-saturated contrast-enhanced T1-weighted and (c) fatsaturated T2-weighted images show signal loss in the dural sac at the L3-L4 level (arrow). In contrast, the whole area of the dural sac (arrow) is well depicted on (b) IDEAL contrast-enhanced T1-weighted image and (d) IDEAL T2-weighted image. compared with for IDEAL T2-weighted imaging ( P,.001) ( Fig 6 ). Interobserver agreement between the two radiologists was moderate to substantial for qualitative evaluation of fat-suppressed contrast-enhanced T1- weighted and IDEAL T1-weighted FSE images. There was moderate to almost perfect interobserver agreement for fat-suppressed T2-weighted and IDEAL T2-weighted FSE images. Discussion Metallic artifacts are induced by differences in the magnetic susceptibilities of Radiology: Volume 259: Number 3 June 2011 n radiology.rsna.org 891

8 Figure 6 Figure 6: Axial MR images in 49-year-old woman with 5-year-old posterior fi xation in L4-L5. (a) Fat-saturated contrastenhanced T1-weighted image shows a pulse artifact (arrow) that is not seen in the same location (arrow) on (b) an IDEAL contrast-enhanced T1-weighted image. Failure of fat saturation is seen on (c) a fat-saturated T2-weighted image. In contrast, (d) an IDEAL T2-weighted image shows homogeneous fat saturation. adjacent tissues and implants ( ). Short inversion time inversion-recovery (STIR) imaging has been pro posed as an alternative solution to decrease these susceptibility artifacts because it is relatively independent of the homogeneity of the main magnetic field ( 12,15 ). However, application of STIR is restricted compared with application of contrast-enhanced T1-weighted imaging because of the risk of suppressing short-t1 tissues with a T1 similar to that of fat ( 3,16 ). Recently, IDEAL water-fat separation was introduced to achieve uniform fat saturation by compensating for the effects of field inhomogeneity while main- taining a high SNR ( 3 ). In this study, we examined the clinical utility of the IDEAL technique in the spines of patients with metallic hardware. Our phantom study results showed that the width of the metallic artifacts was significantly reduced in IDEAL imaging. The main reason for artifact reduction in the IDEAL images was the large decrease in the peripheral rim of high signal intensity owing to improvement of the mismapping of a disproportionate number of spins to that location ( 12 ). However, this technique showed a little improvement for geographic distortion. The phantom study results also revealed lesser SNR changes of contrast material with most of the pedicle screws in IDEAL imaging. Therefore, our study demonstrated that the IDEAL technique is highly effective in maintaining its own signal intensity against metallic artifacts, which in conventional imaging may pose diagnostic difficulties when anatomic structures around metallic devices are represented as unexpected regions of high signal intensity. IDEAL imaging can provide consistent tissue contrast even in the presence of metallic hardware, allowing for accurate interpretation of postoperative spinal MR images. Our qualitative study indicated marked improvement of visibility of the dural 892 radiology.rsna.org n Radiology: Volume 259: Number 3 June 2011

9 sac and paravertebral muscles in IDEAL imaging. The dural sac and paravertebral muscles were included as the main target areas of qualitative analysis because they are commonly involved in postoperative complications, such as hematomas or abscesses, and they are frequently obscured by artifacts from pedicle screws, delaying diagnosis and treatment of postoperative complications. As seen in our results, the IDEAL technique provides uniform fat saturation, even in the presence of metallic hardware, because it can directly measure field inhomogeneities to demodulate their effects from the acquired source images, which renders these methods insensitive to both B0 and B1 inhomogeneities ( 7,12,15 19 ). IDEAL has very high SNR efficiency because all information from source images is effectively processed in the waterfat decomposition. Compared with fatsaturated acquisition with three averages, there is no time or SNR penalty ( 20 ). In contrast, a drawback of IDEAL imaging is the need for three images, which increases imaging time threefold ( 6 ). This study had several limitations. First, we conducted the MR imaging study with only titanium-based pedicle screws, which are widely used in most institutions. Future studies should examine other metallic devices made of various materials to verify the general applicability of the IDEAL sequence for reducing metallic artifacts. Second, the image quality results showed differences in k values between the IDEAL and conventional image sets for evaluation of the dural sac and uniformity of fat saturation. However, as the difference between the two sets was highly significant, such differences are unlikely to affect the outcome of our study. Third, this study included only a small number of patients, which could weaken the statistical power. Fourth, our study was not focused on the evaluation of the visibility of pathologic lesions, including infectious, tumorous, and traumatic conditions. Finally, only one radiologist measured the metallic artifacts in the quantitative analysis. In conclusion, IDEAL contrastenhanced T1- and T2-weighted imaging effectively reduces the degree of tissueobscuring artifacts produced by spinal fixation hardware and improves image quality compared with conventional sequences. Acknowledgments: The authors thank Young Soo Kim, Sung Jin Kang, Tae Hyun Park, Myung Sik Joo, and Jae Hwan Cho for assistance during MR imaging. Disclosures of Potential Conflicts of Interest: J.G.C. No potential conflicts of interest to disclose. W.J. No potential conflicts of interest to disclose. M.H.L. No potential conflicts of interest to disclose. D.H.K. No potential conflicts of interest to disclose. J.S.P. Financial activities related to the present article: as chief of the radiology department of Soonchunhyang University Hospital, was main beneficiary for the IDEAL pulse sequence from GE Healthcare. Financial activities not related to the present article: none to disclose. Other relationships: none to disclose. W.H.S. No potential conflicts of interest to disclose. B.H.Y. No potential conflicts of interest to disclose. References 1. Rutherford EE, Tarplett LJ, Davies EM, Harley JM, King LJ. Lumbar spine fusion and stabilization: hardware, techniques, and imaging appearances. RadioGraphics 2007 ; 27 ( 6 ): Petersilge CA, Lewin JS, Duerk JL, Yoo JU, Ghaneyem AJ. Optimizing imaging parameters for MR evaluation of the spine with titanium pedicle screws. AJR Am J Roentgenol 1996 ; 166 ( 5 ): Gerdes CM, Kijowski R, Reeder SB. IDEAL imaging of the musculoskeletal system: robust water fat separation for uniform fat suppression, marrow evaluation, and cartilage imaging. AJR Am J Roentgenol 2007 ; 189 ( 5 ): W284 W Tartaglino LM, Flanders AE, Vinitski S, Friedman DP. Metallic artifacts on MR images of the postoperative spine: reduction with fast spin-echo techniques. Radiology 1994 ; 190 ( 2 ): Chang SD, Lee MJ, Munk PL, Janzen DL, MacKay A, Xiang QS. MRI of spinal hardware: comparison of conventional T1-weighted sequence with a new metal artifact reduction sequence. Skeletal Radiol 2001 ; 30 ( 4 ): Reeder SB, Pineda AR, Wen Z, et al. Iterative decomposition of water and fat with echo asymmetry and least-squares estimation (IDEAL): application with fast spinecho imaging. Magn Reson Med 2005 ; 54 ( 3 ): Reeder SB, Wen Z, Yu H, et al. Multicoil Dixon chemical species separation with an iterative least-squares estimation method. Magn Reson Med 2004 ; 51 ( 1 ): Siepmann DB, McGovern J, Brittain JH, Reeder SB. High-resolution 3D cartilage imaging with IDEAL SPGR at 3 T. AJR Am J Roentgenol 2007 ; 189 ( 6 ): Dietrich O, Raya JG, Reeder SB, Reiser MF, Schoenberg SO. Measurement of signalto-noise ratios in MR images: influence of multichannel coils, parallel imaging, and reconstruction filters. J Magn Reson Imaging 2007 ; 26 ( 2 ): Meyer JM, Buecker A, Schuermann K, Ruebben A, Guenther RW. MR evaluation of stent patency: in vitro test of 22 metallic stents and the possibility of determining their patency by MR angiography. Invest Radiol 2000 ; 35 ( 12 ): Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977 ; 33 ( 1 ): Lee MJ, Kim S, Lee SA, et al. Overcoming artifacts from metallic orthopedic implants at high-field-strength MR imaging and multi-detector CT. RadioGraphics 2007 ; 27 ( 3 ): Eggers G, Rieker M, Kress B, Fiebach J, Dickhaus H, Hassfeld S. Artefacts in magnetic resonance imaging caused by dental material. MAGMA 2005 ; 18 ( 2 ): Guermazi A, Miaux Y, Zaim S, Peterfy CG, White D, Genant HK. Metallic artefacts in MR imaging: effects of main field orientation and strength. Clin Radiol 2003 ; 58 ( 4 ): Viano AM, Gronemeyer SA, Haliloglu M, Hoffer FA. Improved MR imaging for patients with metallic implants. Magn Reson Imaging 2000 ; 18 ( 3 ): Bydder GM, Steiner RE, Blumgart LH, Khenia S, Young IR. MR imaging of the liver using short TI inversion recovery sequences. J Comput Assist Tomogr 1985 ; 9 ( 6 ): Lee MJ, Janzen DL, Munk PL, MacKay A, Xiang QS, McGowen A. Quantitative assessment of an MR technique for reducing metal artifact: application to spin-echo imaging in a phantom. Skeletal Radiol 2001 ; 30 ( 7 ): White LM, Buckwalter KA. Technical considerations: CT and MR imaging in the postoperative orthopedic patient. Semin Musculoskelet Radiol 2002 ; 6 ( 1 ): Bobman SA, Atlas SW, Listerud J, Grossman RI. Postoperative lumbar spine: contrastenhanced chemical shift MR imaging. Radiology 1991 ; 179 ( 2 ): Reeder SB, Yu H, Johnson JW, et al. T1- and T2-weighted fast spin-echo imaging of the brachial plexus and cervical spine with IDEAL water-fat separation. J Magn Reson Imaging 2006 ; 24 ( 4 ): Radiology: Volume 259: Number 3 June 2011 n radiology.rsna.org 893