Comparison of T1-weighted fast spin-echo and T1-weighted fluid-attenuated inversion recovery images of the lumbar spine at 3.

Transcription

1 Acta Radiologica ISSN: (Print) (Online) Journal homepage: Comparison of T1-weighted fast spin-echo and T1-weighted fluid-attenuated inversion recovery images of the lumbar spine at 3.0 Tesla Eleftherios Lavdas, Marianna Vlychou, Nikos Arikidis, Eftychia Kapsalaki, Violetta Roka & Ioannis V. Fezoulidis To cite this article: Eleftherios Lavdas, Marianna Vlychou, Nikos Arikidis, Eftychia Kapsalaki, Violetta Roka & Ioannis V. Fezoulidis (2010) Comparison of T1-weighted fast spin-echo and T1-weighted fluid-attenuated inversion recovery images of the lumbar spine at 3.0 Tesla, Acta Radiologica, 51:3, To link to this article: Published online: 19 Feb Submit your article to this journal Article views: 747 Full Terms & Conditions of access and use can be found at

2 ORIGINAL ARTICLE Acta Radiologica Comparison of T1-weighted fast spin-echo and T1-weighted fluid-attenuated inversion recovery images of the lumbar spine at 3.0 Tesla El e f t h e r i o s La v d a s, Ma r i a n n a Vly c h o u, Ni k o s Arikidis, Ef t y c h i a Ka p s a l a k i, Vi o l e t ta Ro k a & Io a n n i s V. Fe z o u l i d i s Department of Radiology, University Hospital of Larissa, Medical School of Thessaly, Mezourlo, Greece Background: T1-weighted fluid-attenuated inversion recovery (FLAIR) sequence has been reported to provide improved contrast between lesions and normal anatomical structures compared to T1-weighted fast spin-echo (FSE) imaging at 1.5T regarding imaging of the lumbar spine. Purpose: To compare T1-weighted FSE and fast T1-weighted FLAIR imaging in normal anatomic structures and degenerative and metastatic lesions of the lumbar spine at 3.0T. Material and Methods: Thirty-two consecutive patients (19 females, 13 males; mean age 44 years, range years) with lesions of the lumbar spine were prospectively evaluated. Sagittal images of the lumbar spine were obtained using T1-weighted FSE and fast T1-weighted FLAIR sequences. Both qualitative and quantitative analyses measuring the signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and relative contrast (ReCon) between degenerative and metastatic lesions and normal anatomic structures were conducted, comparing these sequences. Results: On quantitative evaluation, SNRs of cerebrospinal fluid (CSF), nerve root, and fat around the root of fast T1-weighted FLAIR imaging were significantly lower than those of T1-weighted FSE images (P,0.001). CNRs of normal spinal cord/csf and disc herniation/ CSF for fast T1-weighted FLAIR images were significantly higher than those for T1-weighted FSE images (P,0.001). ReCon of normal spinal cord/csf, disc herniation/csf, and vertebral lesions/csf for fast T1-weighted FLAIR images were significantly higher than those for T1-weighted FSE images (P,0.001). On qualitative evaluation, it was found that CSF nulling and contrast at the spinal cord (cauda equina)/csf interface for T1-weighted FLAIR images were significantly superior compared to those for T1-weighted FSE images (P,0.001), and the disc/spinal cord (cauda equina) interface was better for T1-weighted FLAIR images (P,0.05). Conclusion: The T1-weighted FLAIR sequence may be considered as the preferred lumbar spine imaging sequence compared to T1-weighted FSE, as it has demonstrated superior CSF nulling, better conspicuousness of normal anatomic structures and degenerative and metastatic lesions, and improved image contrast. Key words: 3.0T MRI; fast spin echo (FSE); fluid-attenuated inversion recovery (FLAIR) technique; lumbar spine Marianna Vlychou, University Hospital of Larissa, Mezourlo 41110, Greece (tel , . mvlychou@med.uth.gr) Submitted July 9, 2009; accepted for publication November 20, 2009 Three-Tesla (3.0T) magnetic resonance (MR) imaging gives the potential of improved image quality and decreased scan times (1, 2). However, field inhomogeneities, susceptibility artifacts, and specific absorption rate (SAR) level limitations are some serious disadvantages of 3.0T MR imaging. A common approach to overcome SAR limitations in conventional spin-echo (SE) and fast spin-echo (FSE) sequences is to prolong the repetition time (TR) without increasing the number of the sections (2, 3). At 3.0T, T1 relaxation times are increased, leading to a reduced T1 contrast, which is further reduced when repetition time (TR) is prolonged (4). Additionally, a long echo train length (ETL) is not feasible for decreasing scan time, because the fluids are imaged with increased signal (5). Fluid-attenuated inversion recovery (FLAIR) magnetic resonance (MR) imaging techniques have been well described (6 8), providing improved contrast between lesions and normal anatomical structures at acquisition times comparable to those DOI / Informa UK Ltd. (Informa Healthcare, Taylor & Francis AS)

3 T1W FSE and T1W FLAIR of the lumbar spine at 3.0T 291 of T1-weighted FSE imaging (9, 10). The T1-weighted FLAIR technique can be used to image the spine (4, 10, 11). However, the clinical application of this sequence has not been extensively evaluated (4, 10 13). The lumbar spine region is among the most commonly involved sites of severe primary spinal degenerative changes (14), and the T1-weighted sequence is always included in the MR protocol (15 17). Our purpose was to compare the relative diagnostic efficacy of T1-weighted FSE and T1-weighted FLAIR images with respect to normal anatomic structures and degenerative and metastatic lesions of the lumbar spine. Material and Methods Thirty-two consecutive patients (19 females, 13 males; mean age 44 years, range years) were studied, and measurements were depicted at normal anatomic structures (disc, cauda equina, cerebrospinal fluid [CSF], roots), degenerated lesions (disc herniation), and metastatic lesions of the vertebral body. There were six patients with a total of 16 metastatic vertebral lesions, and 26 patients with 26 degenerative discs. We assessed a total of 32 normal discs, 32 normal spinal cords at the l0evel of the cauda equina, and 92 normal spinal nerve roots in all the patients of our study group. In addition, we assessed 16 metastatic lesions of the vertebral bodies, and 26 degenerated lesions of vertebral discs (Table 1). The MR imaging protocol included T1-weighted FSE (TR/echo time [TE] 680/17 ms, ETL 2, number of signal averages [NSA] 2, matrix size , scan time 2.36 min) and fast T1-weighted FLAIR (TR/inversion time [TI]/TE 3200/1150/29 ms, ETL 7, NSA 2, matrix size , scan time 3.13 min) sequences in the sagittal plane during the same imaging session on a 3.0T GE Signa HDxt whole-body MR scanner (General Electric Healthcare, Milwaukee, Wisc., USA) using a synergy body phase-array surface coil. The remaining parameters of the fast T1-weighted FLAIR sequence were identical to the sagittal T1-weighted FSE lumbar spine imaging protocol used at our institution: flip angle 90, slice thickness 3 mm no. of slices 13, slice gap 1 mm field of view (FOV) 320 mm. Twenty patients were initially tested for parameter optimization of fast T1-weighted FLAIR sequences before the onset of the study. Table 1. Summary of normal and pathologic findings Lumbar spinal structures No. of cases Normal anatomic structures Disc 32 Spinal cord (cauda equina) 32 Roots 92 Metastatic lesions of the vertebral body 16 Degenerated lesions of vertebral discs (herniation) 26 Quantitative analysis Quantitative analyses were performed for all patients. In the quantitative analysis, the following items were analyzed: a) the signal-to-noise ratio (SNR) of the spinal cord, normal trabecular bone marrow, disc, root, fat, cortical bone of the vertebral body, CSF, and metastatic lesions at the vertebral body; b) the contrast-tonoise ratio (CNR); and c) the relative contrast (ReCon) between the CSF and spinal cord, normal bone marrow and disc, root and its surrounding fat, metastatic lesions at the vertebral body and CSF, and disc herniation and CSF. For calculating these values, the signal intensities (SI) of the spinal cord, CSF, normal bone marrow, disc, and standard deviation (SD) of the background noise were measured by placing regions of interest (ROIs). The ROIs for each patient were basically placed at an identical position and size on T1-weighted FSE and fast T1-weighted FLAIR images. When the positions of ROIs were moved in some cases because of patient motion, ROIs were selected from the relative position to adjacent tissues. The SNR was calculated as: SNR = SI A A Noise where A represents the tissue of interest, SIA the signal intensity of A measured by an elliptical ROI on the system console, and Noise the background noise, which was defined as the standard deviation of a measurement made by placing an elliptical 2-cm ROI anterior to the abdominal wall (air). The CNR was calculated as: SI A SI B CNRAB Noise where SI A and SI B define the SI of the corresponding tissues. The ReCon was calculated as: SI A SI B ReConAB 3100% SI 1SI A Qualitative analysis T1-weighted FSE and fast T1-weighted FLAIR images were compared qualitatively in all of the 32 patients. All MR images from the T1-weighted FSE and fast T1-weighted FLAIR sequences were evaluated independently at two separate settings with 3 weeks interval by two radiologists, and the results of the blinded evaluation were used for analyses. The images from two sequences were filmed at optimal window and level settings. The radiologists graded on a five-point scale (0, nonvisualization; 1, poor; 2, average; 3, good; 4, excellent) for each of the following image characteristics: 1) overall image quality, 2) CSF nulling, 3) conspicuousness of morphologic abnormalities in the discovertebral junction, 4) conspicuousness of the nerve roots in the B

4 292 Eleftherios Lavdas et al. neural foramen, 5) contrast at the disc CSF interface, 6) contrast at the disc spinal cord (cauda equina) interface, 7) contrast at the metastatic lesion of the vertebral body CSF, and 8) contrast at the spinal cord (cauda equina) CSF interface. The readers also evaluated the presence of image artifacts using a separate score (0, minimum; 1, slight; 2, moderate; 3, severe; 4, maximum). Statistical methods Statistical analysis was carried out in MATLAB. The statistical significance of quantitative and qualitative data was determined by using a Wilcoxon s signed-rank test for paired samples, and P values,0.05 were considered to be statistically significant. The kappa test (Cohen s coefficient of agreement) was performed on observer ratings for each sequence in order to ensure that there was statistically significant interobserver agreement. Results Both T1-weighted FSE and fast T1-weighted FLAIR imaging were effective for all patients in demonstrating both the normal anatomical structures and degenerative and metastatic changes of the lumbar spine. Quantitative results The results of the quantitative analysis obtained from all the patients are presented in Table 2. Specifically, SNRs of CSF, neural roots, and fat around roots obtained with fast T1-weighted FLAIR imaging were significantly lower than those for T1-weighted FSE images (P,0.001). CNRs for fast T1-weighted FLAIR images were significantly higher than those for T1-weighted TSE images in cases of spinal cord/csf and disc herniation/csf (P,0.001). ReCons for fast T1-weighted FLAIR images were significantly higher than those for T1-weighted TSE images in cases of spinal cord/csf, disc herniation/csf, and metastatic lesions of the vertebral body/csf (P,0.001). Qualitative results Effective nulling of CSF signal using fast T1-weighted FLAIR imaging was achieved in all patients. The conspicuousness of morphologic abnormalities in the discovertebral junction and nerve roots in the neural foramina of fast T1-weighted FLAIR was non-significantly superior compared to those for T1-weighted FSE images (P.0.05). Parameters in fast T1-weighted FLAIR images that were found Table 2. Quantitative results comparing T1-weighted FSE and T1-weighted FLAIR sequences FSE T1 FLAIR Avg. SD Avg. SD P value SNR CNR Disc Bone marrow Vertebral body cortical bone Vertebral body trabecular bone CSF Root Fat around root Disc herniation Lesions of the vertebral body Disc bone marrow Disc vertebral body cortical bone Spinal cord CSF Root-fat Disc herniation CSF Metastatic lesions of the vertebral body CSF Relative contrast (ReCon) Disc bone marrow Disc vertebral body cortical bone Spinal cord CSF Root fat Disc herniation CSF Metastatic lesions of the vertebral body CSF

5 T1W FSE and T1W FLAIR of the lumbar spine at 3.0T 293 to be significantly superior compared to those for T1-weighted FSE images (P,0.001) included CSF nulling and contrast at the disc CSF interface, disc spinal cord (cauda equina) interface, spinal cord (cauda equina) CSF interface, and metastatic lesions (vertebral body) CSF interface (Figs. 1 3). The difference in the overall image quality and image artifact scores between the two techniques was not statistically significant (P.0.05). On qualitative evaluation, both radiologists found CSF nulling for T1-weighted FLAIR images to be significantly superior to T1-weighted FSE images (P,0.001). The interobserver agreement was almost perfect for T1-weighted FLAIR imaging (kappa 0.833) and perfect for T1-weighted FSE imaging (kappa 0.975). Contrast at the spinal cord (cauda equina) CSF interface for T1-weighted FLAIR images was also found to be significantly superior to that for T1-weighted FSE images (P,0.001). The interobserver agreement was almost perfect for T1-weighted FLAIR imaging (kappa 0.833), while moderate for T1-weighted FSE imaging (kappa 0.467). One of the observers found the contrast at the disc spinal cord (cauda equina) interface better for T1-weighted FLAIR images (P,0.05), demonstrating substantial interobserver agreement (kappa 0.674). Fig. 1. Sagittal MR images of a patient with degenerative changes in the lumbar spine. A. T1-weighted FSE image shows degenerative discs and vertebrae at various levels. B. T1-weighted FLAIR image shows improved conspicuousness of intraspinal structures, superior CSF nulling (asterisk) as well as improved contrast at the disc CSF interface (arrowhead) and cauda equina CSF interface (arrow). Discussion The FLAIR technique is an inversion recovery (IR) sequence in which the inversion time (TI) is chosen to null the signal of CSF (4, 10, 18). IR provides superior T1-weighted image contrast to that of SE imaging (19 21). FSE techniques can be implemented to T1-weighted FLAIR to allow an acceptable acquisition time (10). It has also been suggested that a fast T1-weighted IR sequence provides superior gray-to-white matter contrast and improved lesion-to-background contrast in the brain as compared with conventional T1-weighted SE imaging (11, 13). The T1-weighted FLAIR technique can be used to image the spine (8, 10, 12), but there are only a limited number of studies assessing the clinical application of the T1-weighted FLAIR technique (4, 8, 10 13). Also, all of the above studies included a small number of patients and heterogeneous diseases such as various brain (3, 5, 6) and spinal diseases (10). In the present study, the value of fast T1-weighted FLAIR technique was presented in a homogeneous and relatively sufficient patient group for the first time. The value of fast T1-weighted FLAIR technique at 3.0T is greater, because the T1 relaxation times are increased, leading to a reduced T1 contrast. The longer T1 times cause the signal intensity of CSF to change from dark at low field strength to gray (4, 8). Using an ETL of six, the suppression of fluids of T1-weighted FLAIR becomes comparable to that of the conventional T1-weighted SE sequence and limits image artifacts (4, 5, 10). CSF-nulled images have been reported to improve the conspicuity of syringes, spinal cord cysts, intraspinal tumors, and multiple sclerosis, the detection of edema and metastatic lesions in the fatty bone marrow, and increased contrast at the CSF cord and CSF disc herniation interfaces (10). According to the qualitative assessment, CSF nulling, contrast at the disc CSF interface, disc spinal cord (cauda equina) interface, spinal cord (cauda equina) CSF interface, and degenerative or metastatic lesions of the vertebral body CSF of fast T1-weighted FLAIR images were significantly superior to those for T1-weighted FSE images. Quantitatively, the SNR of CSF was significantly lower in the case of T1-weighted FLAIR images, indicating CSF suppression. Additionally, T1-weighted FLAIR images demonstrated significantly higher CNR and ReCon over T1-weighted FSE, in cases of disc herniation/csf (P,0.001), spinal cord/csf (P,0.001), and degenerative or metastatic lesions of the vertebral body/csf (P,0.05). Those results are similar to data found by Melhem et al. (10). Our study indicated in particular an increased relative contrast of T1-weighted

6 294 Eleftherios Lavdas et al. Fig. 2. Sagittal MR images of a patient with advanced disc herniation at the L4 L5 level. A. T1-weighted FSE image does not clearly show distinction between herniated disc nerve roots and CSF at the L4 L5 level (circle). B. T1-weighted FLAIR image clearly shows the margin between herniated disc CSF and nerve roots CSF. Also, CSF nulling is superior compared to FSE image. FLAIR images between the metastatic lesions of the vertebral body CSF (P,0.001), facilitating the evaluation of the expansion of the lesion within the spinal canal, and therefore characterizing the nature of the lesion. Sh a p i r o et al. (4) have reported that T1-weighted FLAIR is better than the T1-weighted FSE technique in delineating all normal tissue interfaces between soft tissue/csf bone or disc/csf as well as abnormal/ normal tissue interface. The T1-weighted FLAIR technique is also superior for mitigating susceptibility artifacts because of its longer ETL. In our study, there were 16 cases with metastatic lesions in the vertebral body, and CNR was calculated between the lesion and surrounding tissues. The increased CNR and ReCon of lesions of the vertebral body CSF provided by T1-weighted FLAIR over T1-weighted FSE images were helpful in the evaluation of the expansion of the lesion within the spinal canal. Qualitative analysis indicated good to excellent contrast performance provided by T1-weighted FLAIR images, significantly higher than T1-weighted FSE images (P,0.05). In conclusion, the results of our study show that fast T1-weighted FLAIR sequence at 3.0T demonstrates clear advantages compared to T1-weighted FSE sequence. Therefore, T1-weighted FLAIR sequence is proposed as an optional sequence that should be included in the protocol for lumbar spine in 3.0T MR imaging. Further investigations with a larger patient group, and subgroups according to lesion and postcontrast studies, are needed in order to assure the incorporation of T1-weighted FLAIR sequence into the standard lumbar spine protocol. Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper. References Fig. 3. Sagittal MR images of a patient with metastatic disease of the lumbar spine. A. T1-weighted FSE image shows multiple levels of involved vertebral bodies with adequate distinction of osseous infiltration/ disc at the T12, L1, and L2 levels. B. T1-weighted FLAIR image shows improved conspicuousness between metastasis/disc at the upper endplate of L5 (asterisk). All metastatic vertebrae show improved contrast compared to CSF and conus medullaris. The posterior part of the TH10 vertebra is better delineated, and there is no evidence of intraspinal invasion (arrows). 1. Phalke VV, Gujar S, Quint DJ. Comparison of 3.0 T versus 1.5 T MR: imaging of the spine. Neuroimaging Clin N Am 2006;16:241 8, ix. 2. Ramnath RR. 3.0T MR imaging of the musculoskeletal system (part I): considerations, coils, and challenges. Magn Reson Imaging Clin N Am 2006;14: Tetzlaff RH, Mader I, Kuker W, Weber J, Ziyeh S, Schulze- Bonhage A, et al. Hyperecho-turbo spin-echo sequences at 3.0T: clinical application in neuroradiology. Am J Neuroradiol 2008;29: Shapiro MD. MR imaging of the spine at 3.0T. Magn Reson Imaging Clin N Am 2006;14: Stehling C, Niederstadt T, Kramer S, Kugel H, Schwindt W, Heindel W, et al. Comparison of a T1-weighted inversionrecovery-, gradient-echo- and spin-echo sequence for imaging of the brain at 3.0 Tesla. Rofo 2005;177: Hajnal JV, Bryant DJ, Kasuboski L, Pattany PM, De Coene B, Lewis PD, et al. Use of fluid attenuated inversion recovery (FLAIR) pulse sequences in MRI of the brain. J Comput Assist Tomogr 1992;16: White SJ, Hajnal JV, Young IR, Bydder GM. Use of fluidattenuated inversion-recovery pulse sequences for imaging the spinal cord. Magn Reson Med 1992;28: Erdem LO, Erdem CZ, Acikgoz B, Gundogdu S. Degenerative disc disease of the lumbar spine: a prospective

7 T1W FSE and T1W FLAIR of the lumbar spine at 3.0T 295 comparison of fast T1-weighted fluid-attenuated inversion recovery and T1-weighted turbo spin echo MR imaging. Eur J Radiol 2005;55: Rydberg JN, Hammond CA, Grimm RC, Erickson BJ, Jack CR Jr, Huston J 3rd, et al. Initial clinical experience in MR imaging of the brain with a fast fluid-attenuated inversion-recovery pulse sequence. Radiology 1994;193: Melhem ER, Israel DA, Eustace S, Jara H. MR of the spine with a fast T1-weighted fluid-attenuated inversion recovery sequence. Am J Neuroradiol 1997;18: Hori M, Okubo T, Uozumi K, Ishigame K, Kumagai H, Araki T. T1-weighted fluid-attenuated inversion recovery at low field strength: a viable alternative for T1-weighted intracranial imaging. Am J Neuroradiol 2003; 24: White ML. Cervical spine: MR imaging techniques and anatomy. Magn Reson Imaging Clin N Am 2000;8: Rydberg JN, Hammond CA, Huston J 3rd, Jack CR Jr, Grimm RC, Riederer SJ. T1-weighted MR imaging of the brain using a fast inversion recovery pulse sequence. J Magn Reson Imaging 1996;6: Leonardi M, Simonetti L, Agati R. Neuroradiology of spine degenerative diseases. Best Pract Res Clin Rheumatol 2002;16: Demaerel P, Sunaert S, Wilms G. Sequences and techniques in spinal MR imaging. JBR-BTR 2003;86: Ruggieri PM. Pulse sequences in lumbar spine imaging. Magn Reson Imaging Clin N Am 1999;7:425 37, vii. 17. Gillams AR, Soto JA, Carter AP. Fast spin echo vs conventional spin echo in cervical spine imaging. Eur Radiol 1997;7: De Coene B, Hajnal JV, Gatehouse P, Longmore DB, White SJ, Oatridge A, et al. MR of the brain using fluid-attenuated inversion recovery (FLAIR) pulse sequences. Am J Neuroradiol 1992;13: Fries P, Runge VM, Kirchin MA, Watkins DM, Buecker A, Schneider G. Magnetic resonance imaging of the spine at 3 Tesla. Semin Musculoskelet Radiol 2008;12: Oshio K, Jolesz FA, Melki PS, Mulkern RV. T2-weighted thin-section imaging with the multislab three-dimensional RARE technique. J Magn Reson Imaging 1991;1: Tanenbaum LN. Clinical 3.0T MR imaging: mastering the challenges. Magn Reson Imaging Clin North Am 2006;14:1 15.