Magnetic Resonance Imaging 33 (2015) 194–200
Contents lists available at ScienceDirect
Magnetic Resonance Imaging journal homepage: www.mrijournal.com
Reduction of motion, truncation and flow artifacts using BLADE sequences in cervical spine MR imaging Eleftherios Lavdas a, Panayiotis Mavroidis b, c,⁎, Spiros Kostopoulos d, Constantin Ninos d, Aspasia-Dimitra Strikou a, Dimitrios Glotsos d, Anna Vlachopoulou a, Georgia Oikonomou a, Nikolaos Economopoulos e, Violeta Roka f, Georgios K. Sakkas g, Antonios Tsagkalis h, Sotirios Stathakis b, Nikos Papanikolaou b, Georgios Batsikas i a
Department of Medical Radiological Technologists, Technological Education Institute of Athens, Greece Department of Radiological Sciences, University of Texas Health Sciences Center at San Antonio, San Antonio, TX, USA c Department of Medical Physics, Karolinska Institutet & Stockholm University, Stockholm, Sweden d Department of Medical Instruments Technology, Technological Education Institute of Athens, Greece e 2nd Department of Radiology, General University Hospital ATTIKON, National and Kapodistrian University of Athens, Greece f Health Center of Farkadona, Trikala, Greece g Center for Research and Technology Thessaly Trikala h Department of Orthopaedic Surgery, IASO Thessalias Hospital, Larissa, Greece i Department of Medical Imaging, IASO Thessalias Hospital, Larissa, Greece b
a r t i c l e
i n f o
Article history: Received 2 January 2014 Revised 21 October 2014 Accepted 31 October 2014 Keywords: Cervical spine MR imaging BLADE sequences Image quality Artifact reduction
a b s t r a c t Purpose: To assess the efficacy of the BLADE technique (MR imaging with ‘rotating blade-like k-space covering’) to significantly reduce motion, truncation, flow and other artifacts in cervical spine compared to the conventional technique. Materials and methods: In eighty consecutive subjects, who had been routinely scanned for cervical spine examination, the following pairs of sequences were compared: a) T2 TSE SAG vs. T2 TSE SAG BLADE and b) T2 TIRM SAG vs. T2 TIRM SAG BLADE. A quantitative analysis was performed using the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) measures. A qualitative analysis was also performed by two radiologists, who graded seven image characteristics on a 5-point scale (0: non-visualization; 1: poor; 2: average; 3: good; 4: excellent). The observers also evaluated the presence of image artifacts (motion, truncation, flow, indentation). Results: In quantitative analysis, the CNR values of the CSF/SC between TIRM SAG and TIRM SAG BLADE were found to present statistically significant differences (p b 0.001). Regarding motion and truncation artifacts, the T2 TSE BLADE SAG was superior compared to the T2 TSE SAG, and the T2 TIRM BLADE SAG was superior compared to the T2 TIRM SAG. Regarding flow artifacts, T2 TIRM BLADE SAG eliminated more artifacts than T2 TIRM SAG. Conclusions: In cervical spine MRI, BLADE sequences appear to significantly reduce motion, truncation and flow artifacts and improve image quality. BLADE sequences are proposed to be used for uncooperative subjects. Nevertheless, more research needs to be done by testing additional specific pathologies. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Magnetic resonance imaging has become an established tool for the assessment of the cervical spine. cervical spine MRI examinations are used to assess soft disc herniations, suspicion of disc hernia
⁎ Corresponding author at: Division of Medical Physics, Department of Radiological Sciences, Cancer Therapy and Research Center, University of Texas Health Sciences Center San Antonio, 7979 Wurzbach Rd, MC 7889, San Antonio TX 78229-4427, USA. Tel.: +1 210 450 1027, +1 210 478 9703. E-mail address: [email protected] (P. Mavroidis). http://dx.doi.org/10.1016/j.mri.2014.10.014 0730-725X/© 2015 Elsevier Inc. All rights reserved.
recurrence after operation, cervical spondylosis, osteophytes, joint arthrosis, spinal canal lesions (tumor, multiple sclerosis etc.), bone diseases and paravertebral spaces [1]. Pulsation and motion can lead to modulation of the MRI k-space data resulting in severe artifacts. The most common artifacts are: flow artifacts which are depicted as flow voids in MR images, pulsation artifacts which are coherent and localized ghosting, motion artifacts which are incoherent and distributed ghosting, truncation artifacts (they stem from swallowing and respiration during the acquisition of the sequence), which appear as intensity ripples following high contrast edges and a relatively new artifact is
E. Lavdas et al. / Magnetic Resonance Imaging 33 (2015) 194–200
the indentation artifact, which composes signal voids not consistent with flow voids (see Fig. 6 in the study by Fellner et al. [2]). A number of techniques have been developed in an attempt to eliminate pulsation artifacts. The most widely used of these techniques include: ordered phase encoding [3], gradient moment nulling [4,5], spatial pre-saturation [6] and multiple averaging [7]. MR imaging with ‘rotating blade-like k-space covering’ (BLADE) and ‘Periodically Rotated Overlapping Parallel Lines with Enhanced Reconstruction’ (PROPELLER) have been shown to effectively reduce motion and pulsatile flow artifacts [8–11]. The term BLADE is the product name (used by Siemens Medical System, Erlagen, Germany) for a turbo spin echo (TSE) sequence that uses the PROPELLER k-space trajectory [12]. The BLADE method acquires a number of blades that are rotated around the center of the k-space (successive blades are acquired at different angles). Each blade consists of a number of lowest phase encoding lines of a conventional rectilinear k-space trajectory that are acquired after a single radiofrequency excitation. This technique can potentially eliminate pulsation in MR images when that pulsation is caused by the unwanted modulation of k-space data [12–14]. In cervical spine images, BLADE sequences have shown to be capable of eliminating motion, truncation and flow artifacts and improve the image quality [2,13]. The most frequently used and useful sequences in cervical spine MRI are T2-weighted sagittal TSE and turbo inversion recovery (TIRM), which is mainly used to remove the fat signal. Therefore, the purpose of this study is to evaluate the efficiency of BLADE in combination with the TIRM sequence in reducing motion, pulsatile and flow artifacts and improve image quality in cervical spine MR images, due to the fact that TIRM sequences shows high sensitivity to demyelinating lesions [15,16]. Additionally, this study aims at investigating the ability of these sequences (T2 TSE BLADE, T2 TIRM BLADE) to reduce impact of inherent artifacts such as indentation and wrap (or string) artifacts, which were presented by Finkenzeller et al. [17]. The originality of the present study is due to the fact that TIRM sequences have not been implemented in any MRI examination protocol for cervical spine. The present study is among the first to implement T2 TIRM SAG BLADE in MR imaging at a different however clinical case. Due to the very promising results that this sequence previously showed it was also implemented in cervical spine MRI as well. In summary, the present work aims at providing a complete picture of the impact of the BLADE technique to the image quality by examining a series of artifacts both quantitatively as well as qualitatively. Its objective is to give more specific answers about what artifacts are eliminated, significantly reduced or just reduced and based on that what clinical cases will mostly benefit by the application of this technique and the respective change of the local protocols. In this study, it is the first time that TIRM sequences are examined quantitatively and qualitatively in cervical spine.
195
2.2. MR imaging techniques All the cervical spine examinations were performed using a 1.5 T scanner (Magnetiom Avanto, Siemens Medical Systems, Erlangen, Germany) with multichannel cervical coil. The parameters of the different sequences are presented in Table 1. For the conventional sequences, the number of excitations (NEX) was two (NEX = 2), whereas for the BLADE sequences NEX was reduced to 1. The T2 TSE SAG BLADE sequence was implemented according to the standardized protocol of the manufacturer with the only difference being that in our case the REST (Regional Saturation Technique) slabs were applied parallel to the FOV exactly as in the conventional T2 TSE SAG sequence. Furthermore, in the tests that we initially performed we realized that by using parallel REST slabs we got better image quality (e.g. foldover artifacts were eliminated). The values of ETL (echo train length) and rBw (received bandwidth) parameters were increased as part of the intrinsic nature of BLADE technique, which has been found to reduce or eliminate various artifacts [18–22]. The reduction of T2 decay blurring, susceptibility and chemical shift artifacts with a change in rBw are well established as well as the relation of ETL with ghosting and truncation artifacts. The other parameters (e.g. TR, TE) were automatically adjusted by the system. In all four sequences 100% phase oversampling was used. In summary, for the rectilinear sequences and the T2 TIRM SAG the parameters, provided by the manufacturer, were used without any change [13,23]. The protocol parameters of the T2 TIRM BLADE SAG sequence were selected based on the grounds that they produced the best image quality. 2.3. Quantitative analysis A quantitative analysis was performed for all the subjects and sequences. In the quantitative analysis, the following items were analyzed: the signal-to-noise ratio (SNR) of the spinal cord, normal trabecular bone marrow, vertebral disc, neural root, fatty tissue, as well as the contrast-to-noise ratio (CNR); in order to calculate these values, the signal intensities (SI) of the spinal cord, CSF, normal bone marrow, vertebral disc, neural root, fatty tissue and standard deviation (SD) of the background noise were measured by using elliptical regions, which were placed on the images using the system console. The regions of interest (ROIs) for each subject were placed at identical positions and had the same sizes in all four sequences. In the cases that the tissue of interest was moving, the ROIs were manually repositioned based on their relative location to adjacent tissues. The SNR was calculated as: SNRA ¼ 0:655
SIA N
ð1Þ
Table 1 Summary of the sequences that were applied in the cervical spine MR examinations.
2. Materials and methods 2.1. Subjects From May 2012 to August 2013, eighty consecutive subjects (44 females, 36 males; mean age 44 years, range 16–58 years), who had been routinely scanned for cervical examination using four different image acquisition techniques, participated in the study. More specifically, the following pairs of sequences were compared: a) T2 TSE SAG vs. T2 TSE SAG BLADE and b) T2 TIRM SAG vs. T2 TIRM SAG BLADE. This study was approved by the local institutional review board, and a written informed consent was obtained from all the subjects participating in the study protocol.
Sequences
T2 TSE Sag.
T2 TSE Sag. Blade
T2 TIRM Sag.
T2 TIRM Sag. Blade
TR (ms) TE (ms) TI (ms) Matrix (phase/freq) BW (Hz/pix) Acquisition time (min) ET/TF Thickness (mm) Slice gap (mm) Oversampling Average FOV (mm)
3500 89 – 384/269 191 3:35 30/18 3 mm 10% 100% 2 220/220
6000 107 – 320/320 252 3:38 35/29 3 mm 10% 100% 1 220/220
4000 31 160 256/192 181 3:18 24/8 3 mm 10% 100% 2 220/220
4490 48 160 256/256 310 4:17 11/58 3 mm 10% 100% 1 220/220
196
E. Lavdas et al. / Magnetic Resonance Imaging 33 (2015) 194–200
where A represents the tissue of interest, SIA is the signal intensity of tissue A measured by an elliptical ROI on the system console, and N is the background noise, which was defined as the standard deviation of a measurement made by placing an elliptical ROI (1.0 cm 2) anterior to the neck (air). The CNR was calculated by the following expression: CNRAB ¼
SI A −SI B N
ð2Þ
where SIA and SIB define the SI of tissues A and B. In order to compare the SNR or CNR values of different sequences, the resolution was accounted for by normalizing the SNR and CNR values of the examined sequences by their voxel sizes, TR, TE and rBw. The Kolmogorov–Smirnov non parametric test was used to perform the statistical quantitative analysis. 2.4. Qualitative analysis All the images of the four corresponding MR sequences with and without BLADE were visually evaluated and compared independently at two separate settings with 3 weeks interval by two evaluators (two radiologists with at least 10-year experience on MR imaging), both were blinded to residual image information when they made their assessments. The images from the two corresponding sequences were imprinted–depicted at optimal window and level settings. The radiologists graded on a 5-point scale (0: non-visualization; 1: poor; 2: average; 3: good; 4: excellent) for each of the following image characteristics: overall image quality, CSF brightness, conspicuousness of morphologic abnormalities in the discovertebral junction, conspicuousness of the nerve roots in the neural foramen, contrast at the vertebral disc–CSF interface, contrast at the vertebral disc–spinal cord (cauda equina) interface, contrast at the metastatic lesion of the vertebral body– CSF interface, and contrast at the spinal cord–CSF interface. The evaluators also evaluated the presence of image artifacts (motion, truncation, flow, indentation) using a separate score (4: minimum; 3: slight; 2: moderate; 1: severe; 0: maximum). For the scoring of the truncation artifacts, a separate scale was used (5: no pseudolaminar appearance of the spinal cord; 3: the pseudolaminar appearance of the spinal body could be readily recognized; and 1: the pseudolaminar appearance was extremely conspicuous) [24,25]. The statistical significance (p b 0.01) of the qualitative data was determined by the Kruskal–Wallis nonparametric test.
Table 2 Summary of the quantitative comparison between the BLADE and conventional sequences. SNR
T2 TSE SAG
T2 TSE SAG BLADE
T2 TIRM SAG
BM DS RT SC CSF FT
83.3 35.9 61.9 58.1 175.6 143.9
99.1 43.3 70.3 66.4 190.4 176.8
50.6 51.8 52.1 63.8 128.9 19.9
CNR
T2 TSE SAG
T2 TSE SAG BLADE
BM/DS CSF/SC RT/FT CSF/BM CSF/DS DS/RT DS/FT
49.3 117.5 82.0 92.3 139.7 27.6 108.0
58.6 124.0 106.6 91.3 147.1 30.7 133.5
± ± ± ± ± ±
± ± ± ± ± ± ±
52.2 25.6 26.7 26.4 85.6 67.1
37.4 60.4 48.8 45.1 67.1 19.6 52.1
± ± ± ± ± ±
± ± ± ± ± ± ±
58.7 24.1 27.7 33.0 82.4 71.9
48.5 51.5 50.5 35.2 71.4 22.8 61.1
± ± ± ± ± ±
25.1 31.4 42.6 21.9 48.6 14.6
T2 TIRM SAG 21.4 65.2 32.2 78.3 77.1 32.3 32.9
± ± ± ± ± ± ±
22.1 28.7 32.2 34.6 35.9 23.4 28.1
T2 TIRM SAG BLADE 70.9 67.3 70.4 84.0 ⁎ 217.2 26.2
± ± ± ± ± ±
55.8 42.7 68.2 50.6 151.3 21.8
T2 TIRM SAG BLADE 26.2 ⁎⁎ 133.2 49.4 ⁎ 146.3 ⁎ 149.9 41.8 45.6
± ± ± ± ± ± ±
28.2 103.7 54.2 99.4 119.4 33.5 35.1
BM: bone marrow, DS: vertebral disc, RT: neural root, SC: spinal cord, NS: noise, CSF: cerebrospinal fluid, FT: fatty tissue, NS: noise. The analysis was performed using the Kolmogorov–Smirnov non parametric test. ⁎ p b 0.01. ⁎⁎ p b 0.001.
BLADE sequences showed that the BLADE sequence was greater in all the cases with the differences being statistically significant between CSF/SC (p b 0.001), CSF/BM (p b 0.01) and CSF/DS (p b 0.01). 3.2. Qualitative results In T2 TSE SAG sequence, there were 6 uncooperative subjects, 3 of which had images that were graded as non-diagnostic (Fig. 1). These
3. Results 3.1. Quantitative results The results of the quantitative analysis including all the subjects are presented in the Table 2. It is observed that in the BLADE sequences the SNR and CNR values are larger than the conventional ones in most of the cases examined. More specifically, based on results of the SNR of the T2 TSE SAG and T2 TSE SAG BLADE sequences it was found that the BLADE sequence was greater than the conventional one in all the examined tissues. The same pattern was observed for the T2 TIRM SAG and T2 TIRM SAG BLADE sequences, where statistically significant differences were found for the CSF (p b 0.01). Regarding the results of the CNR of the T2 TSE SAG and T2 TSE SAG BLADE sequences it was found that the BLADE sequence was greater than the conventional one in all the cases except one, where the T2 TSE was greater but with no statistical significance. Respectively, the results of the T2 TIRM SAG and T2 TIRM SAG
Fig. 1. Uncooperative subjects. Left: T2 TSE sagittal (A and C) and right: T2 TSE sagittal BLADE (B and D) images of the cervical spine. It is shown that the motion artifacts that are seen in the T2 TSE image are eliminated in the T2 TSE BLADE improving the overall image quality.
E. Lavdas et al. / Magnetic Resonance Imaging 33 (2015) 194–200
subjects were scanned with the corresponding BLADE sequence, and those images collected in the same patient with the BLADE approaches did not show any noticeable artifacts. In T2 TIRM SAG sequence, there were 7 uncooperative subjects, 5 of which had images that were graded as non-diagnostic (Fig. 2). These subjects were scanned with corresponding BLADE sequence, and those images were graded as diagnostic. Regarding the truncation artifacts shown in the conventional sequences, T2 TSE SAG BLADE reduced them in 8 cases and T2 TIRM SAG BLADE in 12 cases. The results of the qualitative analysis including all the subjects are presented in Table 3. The BLADE sequences were superior to the corresponding conventional ones regarding most of the image characteristics. Based on the qualitative results of T2 TSE SAG and T2 TSE SAG BLADE sequences it is apparent that BLADE is significantly superior to the conventional one with statistically significant differences in terms of: 1) overall image quality (OIQ) (p b 0.01), 2) contrast of the spinal cord-CSF interface (CSCCSF) (p b 0.01) (Figs. 3 and 4, A and B), 3) contrast at the vertebral disc– CSF interface (CDCSF) (p b 0.01) (Fig. 4, A and B), 4) motion artifacts (MotArt) (p b 0.001) (Fig. 2, C and D), 5) truncation artifacts (TruArt) (p b 0.001) (Fig. 4, A and B), 6) flow artifacts (FlwArt) (p b 0.01) (Fig. 4, A and B). From all the comparisons of T2 TIRM SAG and T2 TIRM SAG BLADE, it appears that BLADE was better than the conventional one with statistically significant differences in 1) CSF (p b 0.01) (Fig. 3, C, D, E and F, Fig. 4, C and D), 2) CSCCSF (p b 0.01) (Fig. 3, C, D, E and F, Fig. 4, C and D), 3) TruArt (p b 0.001), 4) FlwArt (p b 0.01) (Fig. 4, C and D).
Fig. 2. Uncooperative subjects. Left: T2 TIRM sagittal (A and C) and right: T2 TIRM sagittal BLADE (B and D) images of the cervical spine. It is shown that motion artifacts that are shown in T2 TIRM are eliminated in the T2 TIRM BLADE sequence, which results in an improvement of the image quality. Especially, in the lower case, where the motion artifacts were very extensive in the conventional sequence, the BLADE sequence made the image diagnosable revealing the metastasis at the vertebral body.
197
Table 3 Summary of the results of the qualitative comparison between the different sequences.
OIQ CSF CMADJ CSCCSF CDCSF MotArt TruArt FlwArt IndArt
T2 TSE SAG
T2 TSE SAG BLADE
T2 TIRM SAG
T2 TIRM SAG BLADE
1.8 3.3 3.0 2.2 2.3 2.2 2.2 2.3 2.4
⁎ 2.8 3.7 3.2 ⁎ 2.9 ⁎ 3.0 ⁎⁎ 3.6 ⁎⁎ 2.9 3.1 ⁎ 3.1
1.0 2.5 2.8 1.7 1.9 2.1 1.8 1.7 2.5
1.6 3.0 3.2 ⁎ 2.5 2.4 3.0 ⁎⁎ 2.8 ⁎ 2.5 2.7
± ± ± ± ± ± ± ± ±
1.5 0.9 0.7 1.3 1.3 1.4 1.0 1.5 1.3
± ± ± ± ± ± ± ± ±
1.2 0.5 0.6 1.1 0.9 0.5 0.8 0.6 0.6
± ± ± ± ± ± ± ± ±
1.0 0.6 0.8 1.1 1.1 1.3 1.1 1.0 1.3
± ± ± ± ± ± ± ± ±
1.1 0.7 0.7 0.9 0.8 0.9 0.8 1.0 1.2
OIQ: overall image quality, CSF: cerebrospinal fluid bright, CMADJ: conspicuousness of morphologic abnormalities in the discovertebral junction, CSCCSF: contrast at the spinal cord (cauda equina)–CSF interface, CDCSF: contrast at the disc–CSF interface, MotArt: motion artifacts, TruArt: truncation artifacts, FlwArt: flow artifacts, IndArt: indentation artifacts. The statistical significance of the qualitative data was determined by the Kruskal–Wallis non-parametric test. ⁎ p b 0.01. ⁎⁎ p b 0.001.
4. Discussion In studies employing T2 TSE SAG sequences, the BLADE technique has been reported to eliminate motion, truncation and flow artifacts and improve the image quality [2,13]. In the present study, the T2 TSE BLADE SAG sequence was applied as indicated by its standardized protocol (manufacturer's default) with the only difference being that the REST slabs were applied parallel to the FOV exactly like they are in the conventional T2 TSE SAG sequence. This amendment was made in order to compatibly compare the conventional and BLADE techniques, but also in order to eliminate the artifacts of the anterior cervical region, which were mainly due to swallowing and respiration movements. In trial tests that we performed it was observed that when the REST slabs were perpendicular to FOV, a lot of artifacts appeared in the spinal cord, which might be due to swallowing. In most implementations the readout covers the same extent of k-space; this provides a more uniform extent of k-space coverage in all in-plane directions, whereas the phase encoding direction of conventional Cartesian sampling is often limited relative to the readout direction, and so it is more prone to truncation artifacts. The downside of BLADE is that if the same number of k-space lines is acquired in total, there are some wedge shaped gaps in the outer areas that may lead to artifacts in their own right. To limit this factor it is typical to perform an oversampling which generally lengthens scan time. However, when we increased the scan time of the conventional sequences, we did not see any significant improvement of image quality. Indentation artifacts are related to wrap artifacts. According to other authors, the reduction of indentation artifacts is achieved when REST slabs are applied perpendicularly to the FOV or when the FOV is increased (this way a concomitant resolution loss is observed) [2,13,26,27]. In this study, we have used 100% oversampling, and consequently we did not have wrap artifacts. This would have an impact on the indentation artifacts too. Furthermore, the REST slabs were applied parallel to FOV, which remained constant. In this way, in two cases where indentation artifacts were visible, the diagnostic value of the image was not downgraded at all, and wrap artifacts were significantly reduced. According to our qualitative results, the T2 TSE SAG BLADE sequence was always superior to the conventional T2 TSE SAG with the differences being statistically significant in almost all cases. More specifically, the BLADE technique achieved better overall image quality (p b 0.01), better contrast between SC/CSF and vertebral disc/CSF interface (p b 0.01). All the image artifacts were reduced remarkably. Motion and truncation artifacts, which are very often in
198
E. Lavdas et al. / Magnetic Resonance Imaging 33 (2015) 194–200
Fig. 4. Cooperative subjects. Upper: In T2 TSE sagittal (A) sequence, the flow artifacts in the posterior of the vertebrae of the 4th and 5th vertebrae (the arrows point out those positions) results in an image that indicates that there is no CSF between the vertebra and spinal cord, whereas in the T2 TSE sagittal BLADE (B) sequence, CSF is clearly shown. A better contrast is observed between spinal cord and CSF in the posterior part of spinal cord. Also, the truncations artifacts have been eliminated too, which results in better visualization of CSF in the 4o cervical vertebral body. Lower: The flow artifacts in CSF posteriorly of the 6th and 7th cervical vertebral body (the arrows point out those positions) that are seen in T2 TIRM sagittal (C) are reduced by the T2 TIRM sagittal BLADE (D) sequence. Also, the BLADE sequence eliminates the truncation artifacts in spinal cord. The final result is a better visualization of spinal cord and better contrast between CSF and spinal cord as well as between CSF and cervical vertebral body.
Fig. 3. Cooperative subjects. Upper: The motion and swallowing artifacts that are seen in the T2 TSE image (A) are eliminated in the T2 TSE BLADE (B) along the spinal cord. Also, the BLADE sequence reduces the truncation artifacts, which results in a better visualization of spinal cord. Middle: In T2 TIRM Sagittal BLADE (D), a better contrast between the spinal cord and the CSF as well as between disc and CSF is observed compared to the conventional image (C). Also in this case, the BLADE sequence reduces the truncation artifacts (the arrow point out this artifact). Lower: In T2 TIRM sagittal BLADE (F), a higher image quality and better CSF visualization are shown compared to the conventional sequence (E). The reduction of the truncation artifacts is also evident in this case.
conventional sequences, were not only reduced by the BLADE technique, but the differences between the corresponding sequences were found to be statistically significant (p b 0.001) (Fig. 1, A and B). The flow artifacts were also reduced with the differences being again statistically significant (p b 0.01) (Fig. 4, A and B). Our results are in line with those of other authors [2,23,28]. More specifically, Fellner et al. found that BLADE sequence is superior to the conventional one with statistically significant differences in image sharpness (p b 0.001), image motion, truncation and flow artifacts (p b 0.001), contrast between vertebral body and vertebral disc, spinal cord and CSF (p b 0.001) and in the diagnostic reliability of
spinal cord (p b 0.001) [2]. Fellner et al. found that the BLADE sequence is superior to the conventional one with statistically significant difference in overall image quality [2]. It is known that T2 TIRM sequences suppress fat revealing the pathology, including regions of high inhomogeneity This is a very important property especially in cervical spine MRI, which is prone to artifacts and where spatial pre-saturation cannot be achieved [29]. The present study is among the first to implement T2 TIRM SAG BLADE in MR imaging at a different however clinical case, and due to the very promising results received this sequence was also implemented in cervical spine MRI as well. It is noticeable that the conspicuity and demarcation of the lesions are similar in the T2 TIRM SAG and T2 TSE SAG sequences. Here, the parameters of the T2 TIRM SAG BLADE sequence were kept the same with the conventional sequence with the only difference being that the values of the ETL and the rBw parameters were increased. An increase of the rBw leads to the reduction of several artifacts such as the truncation artifacts, motion artifacts, chemical shift artifacts, susceptibility artifacts, T2 blurring in TSE and ghosting artifacts in EPI [18,19], while an increase of the ETL leads to the reduction of truncation and motion artifacts. The T2 TIRM SAG sequence shows more artifacts than the T2 TSE SAG, and one of the reasons is the larger TR that is used in the TIRM
E. Lavdas et al. / Magnetic Resonance Imaging 33 (2015) 194–200
sequences. Furthermore, due to fat saturation the artifacts are more visible because they are shown as parallel lines of increased signal. Since the conventional sequences show more artifacts, BLADE sequences have a larger margin of artifact elimination and in this way to show a clear improvement in image quality. As mentioned above, the application of BLADE sequences is related to increased values of ETL, rBw and TR. Consequently, these parameters are by necessity different from conventional sequences. Due to the fact that the values of these parameters are related to the intrinsic characteristics of the BLADE sequences the presented results are mostly sequence related. In a previous case report on cervical spine [13], for two cases with multiple sclerosis, image quality was significantly improved by the use of T2 TIRM BLADE and T2 TSE BLADE sequences compared to the conventional sequences. Although TIRM achieves fat suppression, due to the high TR [27], this is accomplished at the expense of having more motion, truncation and flow artifacts, which however get satisfactorily reduced by the T2 TIRM SAG BLADE sequence (Fig. 2). Moreover, in a previous study it was found that when sequences have fat suppression the pulsation artifacts are more prominent in dark regions [28]. Knowing the great usefulness of the TIRM sequences we believe that the use of TIRM BLADE sequences will be very helpful and beneficial not only in the imaging of uncooperative subjects but also for subjects with multiple sclerosis due to the fact that they visualize very clearly spinal cord (Fig. 3, C, D, E and F). Based on our results in the imaging of cooperative subjects using the T2 TIRM SAG and T2 TIRM SAG BLADE sequences, it is clear that the BLADE sequence is significantly superior to the conventional one due to the reasons mentioned above. More specifically, statistically significant differences between the TIRM and the TIRM BLADE sequences were found in terms of contrast of the spinal cord- CSF interface (p b 0.01) and truncation artifacts (p b 0.01). The truncation artifacts known as Gibbs ringing are shown as bright or dark lines, that are seen parallel and next to borders of abrupt intensity change, and they may emulate a syrinx on sagittal images of spinal cord. These artifacts are related to the finite number of encoding steps used by the Fourier transform. It was observed that the T2 TSE SAG BLADE and T2 TIRM SAG BLADE sequences eliminated the truncation artifacts especially in two cases where syringomyelia was observed. The reason BLADE images had less truncation artifact is two-fold: 1) there were more phase encoding steps and 2) rotating through k-space with BLADE helps reduce truncation artifacts in the phase-encoding direction. Regarding the visualization of CSF in the T2 TIRM SAG and T2 TIRM SAG BLADE sequences, in the latter CSF has better SNR because this sequence reduces flow artifacts which reduce the SIs and therefore the SNR of CSF (Fig. 4, C and D). BLADE sequences restrict also the truncation artifacts, which may be the second reason for the redaction of SIs and therefore the SNR of CSF in the conventional TIRM sequence. One of the floating structures which are known for causing ghosting artifacts is the CSF. The signal of CSF splits when artifacts are created in the normal position and the region of overlaid artifacts. When it does not generate artifacts, the T2 TIRM BLADE sequence has higher signal. This is also observed in the qualitative results of the T2 TSE SAG BLADE and T2 TIRM SAG BLADE images, and it is clearly visible in cases where there were stenoses in spinal canal. This is due to the fact that in conventional sequences, flow artifacts are more intensive in borderline cases (such as stenosis), where it is observed that although the conventional sequences had a weak signal in the CSF, the signal was higher in the BLADE sequences. This phenomenon was observed here in the qualitative measurements of both sets of sequences, and it occurs mostly in the conventional sequences, which are more prone to flow artifacts. It is known that human eye can distinguish differences in borderline cases, such as a weak SNR of CSF in conventional sequences
199
against a very high SNR of CSF in BLADE sequences (Fig. 4, C and D). In these cases (stenosis with flow artifacts), the SNR measurements of CSF with BLADE were roughly double those of the T2 TSE Sag sequence. However, in the quantitative measurements (Table 2) the CSF was measured, and statistically significant difference was found only in the cervical foramen, where there were no flow artifacts and partial volume effect. Also, in 8 cases (4 in T2 TSE SAG and 4 in T2 TIRM SAG), the conventional sequences showed that there was no CSF space between the disk herniation and spinal cord, whereas in the corresponding BLADE sequences the presence of CSF was seen. CSF brightness plays a very significant role for the radiological evaluation. The above observation may seem as a minor detail, but it has great clinical importance for the subject and his treatment. It is very important to be able to assess whether cerebrospinal fluid exists between the vertebral body and spinal cord. This issue is related to the limitation of the subarachnoid space, which is the cause of the lower movement of the CSF and the disturbance of the metabolism of nervous tissue in the various syndromes. As indicated by the presented quantitative and qualitative results the proposed methodology appears to be objective and reliable for evaluating CSF brightness. BLADE sequences manage to deal with the aforementioned artifacts mainly through the repeated sampling of the centre of k-space and the greater extent of k-space in all in-plane directions. The weaknesses of BLADE are the potential manifestation of indentation artifacts and the larger acquisition time. In this study, the slice thickness and the FOV were identical for the examined pairs of sequences. Due to the fact that matrix size was different, normalization to the voxel size was performed so that the results of the different sequences refer to the same voxel size. It is known that TR and TE parameters influence directly the SNR results and therefore image quality. More specifically, Fellner et al. [2] state that a short TR and a short TE increase SNR and reduce acquisition time while maintaining sufficient T2 contrast. Additionally, Malamateniou et al. [27] point out that a high TR increases the derivation possibility of motion artifacts. In this study, the analysis was performed with increased values for the TR and TE parameters in the T2 TIRM BLADE SAG sequence even though this was a disadvantage for the BLADE sequences. Despite that, the BLADE sequences always demonstrated higher SNR and CNR values than the conventional sequences. These findings are in-line with the purpose of this study, which is to show that BLADE sequences are useful. In conclusion, the use of BLADE sequences in cervical spine MR examinations appears to be capable of potentially eliminating motion, pulsatile flow and truncation artifacts. Furthermore, BLADE sequences are proposed to be used in the standard examination protocols based on the fact that a significantly improved image quality could be achieved. Finally, we suggest that the clinical use of T2 TIRM SAG BLADE due to its capacity to reduce image artifacts and achieve better contrast. References [1] Cohen WA, Maravilla KR. MRI atlas of the spine. London: Martin Dunitz Ltd; 1991. [2] Fellner C, Menzel C, Fellner FA, Ginthoer C, Zorger N, Schreyer A, et al. BLADE in sagittal T2-weighted MR imaging of the cervical spine. AJNR Am J Neuroradiol 2010;31:674–81. [3] Bailes DR, Gilderdale DJ, Bydder GM. Respiratory ordering of phase encoding (ROPE): a method for reducing respiratory motion artifacts in MR imaging. J Comput Assist Tomogr 1985;9:835–8. [4] Pattany PM, Phillips JJ, Chiu LC. Motion artifact suppression technique (MAST) for MR imaging. J Comput Assist Tomogr 1987;11:369–77. [5] Haacke EM, Lenz GW. Improving MR image quality in the presence of motion by using rephrasing gradients. AJR 1987;148:1251–8. [6] Felmlee JP, Ehman RL. Spatial presaturation: a method for suppressing flow artifacts and improving depiction of vascular anatomy in MR imaging. Radiology 1987;164:559–64. [7] Dixon WT, Brummer ME, Malko JA. Acquisition order and motional artifact reduction in spin warp images. Magn Reson Med 1988;6:74–83.
200
E. Lavdas et al. / Magnetic Resonance Imaging 33 (2015) 194–200
[8] Runge VM, Wood ML, Kaufman DM, Traill MR, Nelson KL. The straight and narrow path to good head and spine MRI. Radiographics 1988;8:507–31. [9] Kallmes DF, Hui FK, Mugler III JP. Suppression of cerebrospinal fluid and blood flow artifacts in FLAIR MR imaging with a single-slab three-dimensional pulse sequence: initial experience. Radiographics 2001;221:251–5. [10] Naganawa S, Satake H, Iwano S, Kawai H, Kubota S, Komada T, et al. Contrastenhanced MR imaging of the brain using T1-weighted FLAIR with BLADE compared with a conventional spin-echo sequence. Eur Radiol 2008;18:337–42. [11] Alibek S, Adamietz B, Cavallaro A, Stemmer A, Anders K, Kramer M, et al. Contrast enhanced T1-weighted fluid-attenuated inversion-recovery BLADE magnetic resonance imaging of the brain: an alternative to spin-echo technique for detection of brain lesions in the unsedated pediatric patient? Acad Radiol 2008; 15:986–95. [12] Pipe JG. Motion correction with PROPELLER MRI: application to head motion and free-breathing cardiac imaging. Magn Reson Med 1999;42:963–9. [13] Ragoschke-Schumm A, Schmidt P, Schumm J, Reimann G, Mentzel HJ, Kaiser WA, et al. Decreased CSF-flow artefacts in T2 imaging of the cervical spine with periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER/BLADE). Neuroradiology 2011;53:13–8. [14] Pipe JG. An optimized center-out k-space trajectory for multishot MRI: comparison with spiral and projection reconstruction. Magn Reson Med 1999; 42:714–20. [15] Campi A, Pontesill S, Gerevini S, Scotti G. Comparison of MRI pulse sequences for investigations of lesions of cervical spinal cord. Neuroradiology 2000;42:669–75. [16] Wesselly M, Cervical MRI. PART 1 a basic overview. USA: Northwestern Health Sciences University; 2003. [17] Finkenzeller T, Zorger N, Kühnel T, Paetzel C, Schuierer G, Stroszczynski C, et al. Novel application of T1-weighted BLADE sequences with fat suppression compared to TSE in contrast-enhanced T1-weighted imaging of the neck: cutting-edge images? J Magn Reson Imaging 2013;37:660–8. [18] Celik A, Elmaoglu M. MRI Handbook. MR physics, patient positioning and protocols. New York: Springer; 2012.
[19] Lavdas E, Mavroidis P, Vassiou K, Roka V, Fezoulidis IV, Vlychou M. Elimination of chemical shift artifacts of thoracic spine with contrast enhanced Flair imaging with fat suppression at 3.0 Tesla. Magn Reson Imaging 2010;28:1535–40. [20] Shapiro MD. MR imaging of the spine at 3 T. Magn Reson Imaging Clin N Am 2006;14:97–108. [21] Tanenbaum LN. Clinical 3 T MR Imaging: mastering the challenges. Magn Reson Imaging Clin N Am 2006;14:1–15. [22] Tamhane AA. Motion correction in periodically-rotated overlapping parallel lines with enhanced reconstruction (PROPELLER) and turboprop MRI. Magn Reson Med 2009;62:174–82. [23] Lavdas E, Mavroidis P, Topaltzikis T, Slatinopoulos V, Roka V, Vlachopoulou A, et al. Comparison of BLADE and conventional T2-TSE sequences for the sagittal visualization of the cervical spinal cord in multiple sclerosis patients - a case report. Magn Reson Imaging 2013;31:1766–70. [24] Yoshioka H, Stevens K, Genovese M, Dillingham MF, Lang P. Articular cartilage of knee: normal patterns at MR imaging that mimic disease in healthy subjects and patients with osteoarthritis. Radiology 2004;231:31–8. [25] Philpott C, Brotchie P. Comparison of MRI sequences for evaluation of multiple sclerosis of the cervical spinal cord at 3 T. Eur J Radiol 2011;80:780–5. [26] Rocca MA, Mastronardo G, Horsfield MA, Pereira C, Iannucci G, Colombo B, et al. Comparison of three MR sequences for the detection of cervical cord lesions in patients with multiple sclerosis. AJNR Am J Neuroradiol 1999;20:1710–6. [27] Malamateniou C, Malik SJ, Counsell SJ, Allsop JM, McGuinness AK, Hayat T, et al. Motion-compensation techniques in neonatal and fetal MR imaging. 2013;34: 1124–36. [28] Lavdas E, Mavroidis P, Hatzigeorgiou V, Roka V, Arikidis N, Oikonomou G, et al. Elimination of motion and pulsation artifacts using Blade sequences in knee MR imaging. Magn Reson Imaging 2012;30:1099–110. [29] Mahnken AH, Wildberger JE, Adam G, Stanzel S, Schmitz-Rode T, Günther RW, et al. Is there a need for contrast-enhanced T1-weighted MRI of the spine after inconspicuous short tau inversion recovery imaging? Eur Radiol 2005;15: 1387–92.
Contents lists available at ScienceDirect
Magnetic Resonance Imaging journal homepage: www.mrijournal.com
Reduction of motion, truncation and flow artifacts using BLADE sequences in cervical spine MR imaging Eleftherios Lavdas a, Panayiotis Mavroidis b, c,⁎, Spiros Kostopoulos d, Constantin Ninos d, Aspasia-Dimitra Strikou a, Dimitrios Glotsos d, Anna Vlachopoulou a, Georgia Oikonomou a, Nikolaos Economopoulos e, Violeta Roka f, Georgios K. Sakkas g, Antonios Tsagkalis h, Sotirios Stathakis b, Nikos Papanikolaou b, Georgios Batsikas i a
Department of Medical Radiological Technologists, Technological Education Institute of Athens, Greece Department of Radiological Sciences, University of Texas Health Sciences Center at San Antonio, San Antonio, TX, USA c Department of Medical Physics, Karolinska Institutet & Stockholm University, Stockholm, Sweden d Department of Medical Instruments Technology, Technological Education Institute of Athens, Greece e 2nd Department of Radiology, General University Hospital ATTIKON, National and Kapodistrian University of Athens, Greece f Health Center of Farkadona, Trikala, Greece g Center for Research and Technology Thessaly Trikala h Department of Orthopaedic Surgery, IASO Thessalias Hospital, Larissa, Greece i Department of Medical Imaging, IASO Thessalias Hospital, Larissa, Greece b
a r t i c l e
i n f o
Article history: Received 2 January 2014 Revised 21 October 2014 Accepted 31 October 2014 Keywords: Cervical spine MR imaging BLADE sequences Image quality Artifact reduction
a b s t r a c t Purpose: To assess the efficacy of the BLADE technique (MR imaging with ‘rotating blade-like k-space covering’) to significantly reduce motion, truncation, flow and other artifacts in cervical spine compared to the conventional technique. Materials and methods: In eighty consecutive subjects, who had been routinely scanned for cervical spine examination, the following pairs of sequences were compared: a) T2 TSE SAG vs. T2 TSE SAG BLADE and b) T2 TIRM SAG vs. T2 TIRM SAG BLADE. A quantitative analysis was performed using the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) measures. A qualitative analysis was also performed by two radiologists, who graded seven image characteristics on a 5-point scale (0: non-visualization; 1: poor; 2: average; 3: good; 4: excellent). The observers also evaluated the presence of image artifacts (motion, truncation, flow, indentation). Results: In quantitative analysis, the CNR values of the CSF/SC between TIRM SAG and TIRM SAG BLADE were found to present statistically significant differences (p b 0.001). Regarding motion and truncation artifacts, the T2 TSE BLADE SAG was superior compared to the T2 TSE SAG, and the T2 TIRM BLADE SAG was superior compared to the T2 TIRM SAG. Regarding flow artifacts, T2 TIRM BLADE SAG eliminated more artifacts than T2 TIRM SAG. Conclusions: In cervical spine MRI, BLADE sequences appear to significantly reduce motion, truncation and flow artifacts and improve image quality. BLADE sequences are proposed to be used for uncooperative subjects. Nevertheless, more research needs to be done by testing additional specific pathologies. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Magnetic resonance imaging has become an established tool for the assessment of the cervical spine. cervical spine MRI examinations are used to assess soft disc herniations, suspicion of disc hernia
⁎ Corresponding author at: Division of Medical Physics, Department of Radiological Sciences, Cancer Therapy and Research Center, University of Texas Health Sciences Center San Antonio, 7979 Wurzbach Rd, MC 7889, San Antonio TX 78229-4427, USA. Tel.: +1 210 450 1027, +1 210 478 9703. E-mail address: [email protected] (P. Mavroidis). http://dx.doi.org/10.1016/j.mri.2014.10.014 0730-725X/© 2015 Elsevier Inc. All rights reserved.
recurrence after operation, cervical spondylosis, osteophytes, joint arthrosis, spinal canal lesions (tumor, multiple sclerosis etc.), bone diseases and paravertebral spaces [1]. Pulsation and motion can lead to modulation of the MRI k-space data resulting in severe artifacts. The most common artifacts are: flow artifacts which are depicted as flow voids in MR images, pulsation artifacts which are coherent and localized ghosting, motion artifacts which are incoherent and distributed ghosting, truncation artifacts (they stem from swallowing and respiration during the acquisition of the sequence), which appear as intensity ripples following high contrast edges and a relatively new artifact is
E. Lavdas et al. / Magnetic Resonance Imaging 33 (2015) 194–200
the indentation artifact, which composes signal voids not consistent with flow voids (see Fig. 6 in the study by Fellner et al. [2]). A number of techniques have been developed in an attempt to eliminate pulsation artifacts. The most widely used of these techniques include: ordered phase encoding [3], gradient moment nulling [4,5], spatial pre-saturation [6] and multiple averaging [7]. MR imaging with ‘rotating blade-like k-space covering’ (BLADE) and ‘Periodically Rotated Overlapping Parallel Lines with Enhanced Reconstruction’ (PROPELLER) have been shown to effectively reduce motion and pulsatile flow artifacts [8–11]. The term BLADE is the product name (used by Siemens Medical System, Erlagen, Germany) for a turbo spin echo (TSE) sequence that uses the PROPELLER k-space trajectory [12]. The BLADE method acquires a number of blades that are rotated around the center of the k-space (successive blades are acquired at different angles). Each blade consists of a number of lowest phase encoding lines of a conventional rectilinear k-space trajectory that are acquired after a single radiofrequency excitation. This technique can potentially eliminate pulsation in MR images when that pulsation is caused by the unwanted modulation of k-space data [12–14]. In cervical spine images, BLADE sequences have shown to be capable of eliminating motion, truncation and flow artifacts and improve the image quality [2,13]. The most frequently used and useful sequences in cervical spine MRI are T2-weighted sagittal TSE and turbo inversion recovery (TIRM), which is mainly used to remove the fat signal. Therefore, the purpose of this study is to evaluate the efficiency of BLADE in combination with the TIRM sequence in reducing motion, pulsatile and flow artifacts and improve image quality in cervical spine MR images, due to the fact that TIRM sequences shows high sensitivity to demyelinating lesions [15,16]. Additionally, this study aims at investigating the ability of these sequences (T2 TSE BLADE, T2 TIRM BLADE) to reduce impact of inherent artifacts such as indentation and wrap (or string) artifacts, which were presented by Finkenzeller et al. [17]. The originality of the present study is due to the fact that TIRM sequences have not been implemented in any MRI examination protocol for cervical spine. The present study is among the first to implement T2 TIRM SAG BLADE in MR imaging at a different however clinical case. Due to the very promising results that this sequence previously showed it was also implemented in cervical spine MRI as well. In summary, the present work aims at providing a complete picture of the impact of the BLADE technique to the image quality by examining a series of artifacts both quantitatively as well as qualitatively. Its objective is to give more specific answers about what artifacts are eliminated, significantly reduced or just reduced and based on that what clinical cases will mostly benefit by the application of this technique and the respective change of the local protocols. In this study, it is the first time that TIRM sequences are examined quantitatively and qualitatively in cervical spine.
195
2.2. MR imaging techniques All the cervical spine examinations were performed using a 1.5 T scanner (Magnetiom Avanto, Siemens Medical Systems, Erlangen, Germany) with multichannel cervical coil. The parameters of the different sequences are presented in Table 1. For the conventional sequences, the number of excitations (NEX) was two (NEX = 2), whereas for the BLADE sequences NEX was reduced to 1. The T2 TSE SAG BLADE sequence was implemented according to the standardized protocol of the manufacturer with the only difference being that in our case the REST (Regional Saturation Technique) slabs were applied parallel to the FOV exactly as in the conventional T2 TSE SAG sequence. Furthermore, in the tests that we initially performed we realized that by using parallel REST slabs we got better image quality (e.g. foldover artifacts were eliminated). The values of ETL (echo train length) and rBw (received bandwidth) parameters were increased as part of the intrinsic nature of BLADE technique, which has been found to reduce or eliminate various artifacts [18–22]. The reduction of T2 decay blurring, susceptibility and chemical shift artifacts with a change in rBw are well established as well as the relation of ETL with ghosting and truncation artifacts. The other parameters (e.g. TR, TE) were automatically adjusted by the system. In all four sequences 100% phase oversampling was used. In summary, for the rectilinear sequences and the T2 TIRM SAG the parameters, provided by the manufacturer, were used without any change [13,23]. The protocol parameters of the T2 TIRM BLADE SAG sequence were selected based on the grounds that they produced the best image quality. 2.3. Quantitative analysis A quantitative analysis was performed for all the subjects and sequences. In the quantitative analysis, the following items were analyzed: the signal-to-noise ratio (SNR) of the spinal cord, normal trabecular bone marrow, vertebral disc, neural root, fatty tissue, as well as the contrast-to-noise ratio (CNR); in order to calculate these values, the signal intensities (SI) of the spinal cord, CSF, normal bone marrow, vertebral disc, neural root, fatty tissue and standard deviation (SD) of the background noise were measured by using elliptical regions, which were placed on the images using the system console. The regions of interest (ROIs) for each subject were placed at identical positions and had the same sizes in all four sequences. In the cases that the tissue of interest was moving, the ROIs were manually repositioned based on their relative location to adjacent tissues. The SNR was calculated as: SNRA ¼ 0:655
SIA N
ð1Þ
Table 1 Summary of the sequences that were applied in the cervical spine MR examinations.
2. Materials and methods 2.1. Subjects From May 2012 to August 2013, eighty consecutive subjects (44 females, 36 males; mean age 44 years, range 16–58 years), who had been routinely scanned for cervical examination using four different image acquisition techniques, participated in the study. More specifically, the following pairs of sequences were compared: a) T2 TSE SAG vs. T2 TSE SAG BLADE and b) T2 TIRM SAG vs. T2 TIRM SAG BLADE. This study was approved by the local institutional review board, and a written informed consent was obtained from all the subjects participating in the study protocol.
Sequences
T2 TSE Sag.
T2 TSE Sag. Blade
T2 TIRM Sag.
T2 TIRM Sag. Blade
TR (ms) TE (ms) TI (ms) Matrix (phase/freq) BW (Hz/pix) Acquisition time (min) ET/TF Thickness (mm) Slice gap (mm) Oversampling Average FOV (mm)
3500 89 – 384/269 191 3:35 30/18 3 mm 10% 100% 2 220/220
6000 107 – 320/320 252 3:38 35/29 3 mm 10% 100% 1 220/220
4000 31 160 256/192 181 3:18 24/8 3 mm 10% 100% 2 220/220
4490 48 160 256/256 310 4:17 11/58 3 mm 10% 100% 1 220/220
196
E. Lavdas et al. / Magnetic Resonance Imaging 33 (2015) 194–200
where A represents the tissue of interest, SIA is the signal intensity of tissue A measured by an elliptical ROI on the system console, and N is the background noise, which was defined as the standard deviation of a measurement made by placing an elliptical ROI (1.0 cm 2) anterior to the neck (air). The CNR was calculated by the following expression: CNRAB ¼
SI A −SI B N
ð2Þ
where SIA and SIB define the SI of tissues A and B. In order to compare the SNR or CNR values of different sequences, the resolution was accounted for by normalizing the SNR and CNR values of the examined sequences by their voxel sizes, TR, TE and rBw. The Kolmogorov–Smirnov non parametric test was used to perform the statistical quantitative analysis. 2.4. Qualitative analysis All the images of the four corresponding MR sequences with and without BLADE were visually evaluated and compared independently at two separate settings with 3 weeks interval by two evaluators (two radiologists with at least 10-year experience on MR imaging), both were blinded to residual image information when they made their assessments. The images from the two corresponding sequences were imprinted–depicted at optimal window and level settings. The radiologists graded on a 5-point scale (0: non-visualization; 1: poor; 2: average; 3: good; 4: excellent) for each of the following image characteristics: overall image quality, CSF brightness, conspicuousness of morphologic abnormalities in the discovertebral junction, conspicuousness of the nerve roots in the neural foramen, contrast at the vertebral disc–CSF interface, contrast at the vertebral disc–spinal cord (cauda equina) interface, contrast at the metastatic lesion of the vertebral body– CSF interface, and contrast at the spinal cord–CSF interface. The evaluators also evaluated the presence of image artifacts (motion, truncation, flow, indentation) using a separate score (4: minimum; 3: slight; 2: moderate; 1: severe; 0: maximum). For the scoring of the truncation artifacts, a separate scale was used (5: no pseudolaminar appearance of the spinal cord; 3: the pseudolaminar appearance of the spinal body could be readily recognized; and 1: the pseudolaminar appearance was extremely conspicuous) [24,25]. The statistical significance (p b 0.01) of the qualitative data was determined by the Kruskal–Wallis nonparametric test.
Table 2 Summary of the quantitative comparison between the BLADE and conventional sequences. SNR
T2 TSE SAG
T2 TSE SAG BLADE
T2 TIRM SAG
BM DS RT SC CSF FT
83.3 35.9 61.9 58.1 175.6 143.9
99.1 43.3 70.3 66.4 190.4 176.8
50.6 51.8 52.1 63.8 128.9 19.9
CNR
T2 TSE SAG
T2 TSE SAG BLADE
BM/DS CSF/SC RT/FT CSF/BM CSF/DS DS/RT DS/FT
49.3 117.5 82.0 92.3 139.7 27.6 108.0
58.6 124.0 106.6 91.3 147.1 30.7 133.5
± ± ± ± ± ±
± ± ± ± ± ± ±
52.2 25.6 26.7 26.4 85.6 67.1
37.4 60.4 48.8 45.1 67.1 19.6 52.1
± ± ± ± ± ±
± ± ± ± ± ± ±
58.7 24.1 27.7 33.0 82.4 71.9
48.5 51.5 50.5 35.2 71.4 22.8 61.1
± ± ± ± ± ±
25.1 31.4 42.6 21.9 48.6 14.6
T2 TIRM SAG 21.4 65.2 32.2 78.3 77.1 32.3 32.9
± ± ± ± ± ± ±
22.1 28.7 32.2 34.6 35.9 23.4 28.1
T2 TIRM SAG BLADE 70.9 67.3 70.4 84.0 ⁎ 217.2 26.2
± ± ± ± ± ±
55.8 42.7 68.2 50.6 151.3 21.8
T2 TIRM SAG BLADE 26.2 ⁎⁎ 133.2 49.4 ⁎ 146.3 ⁎ 149.9 41.8 45.6
± ± ± ± ± ± ±
28.2 103.7 54.2 99.4 119.4 33.5 35.1
BM: bone marrow, DS: vertebral disc, RT: neural root, SC: spinal cord, NS: noise, CSF: cerebrospinal fluid, FT: fatty tissue, NS: noise. The analysis was performed using the Kolmogorov–Smirnov non parametric test. ⁎ p b 0.01. ⁎⁎ p b 0.001.
BLADE sequences showed that the BLADE sequence was greater in all the cases with the differences being statistically significant between CSF/SC (p b 0.001), CSF/BM (p b 0.01) and CSF/DS (p b 0.01). 3.2. Qualitative results In T2 TSE SAG sequence, there were 6 uncooperative subjects, 3 of which had images that were graded as non-diagnostic (Fig. 1). These
3. Results 3.1. Quantitative results The results of the quantitative analysis including all the subjects are presented in the Table 2. It is observed that in the BLADE sequences the SNR and CNR values are larger than the conventional ones in most of the cases examined. More specifically, based on results of the SNR of the T2 TSE SAG and T2 TSE SAG BLADE sequences it was found that the BLADE sequence was greater than the conventional one in all the examined tissues. The same pattern was observed for the T2 TIRM SAG and T2 TIRM SAG BLADE sequences, where statistically significant differences were found for the CSF (p b 0.01). Regarding the results of the CNR of the T2 TSE SAG and T2 TSE SAG BLADE sequences it was found that the BLADE sequence was greater than the conventional one in all the cases except one, where the T2 TSE was greater but with no statistical significance. Respectively, the results of the T2 TIRM SAG and T2 TIRM SAG
Fig. 1. Uncooperative subjects. Left: T2 TSE sagittal (A and C) and right: T2 TSE sagittal BLADE (B and D) images of the cervical spine. It is shown that the motion artifacts that are seen in the T2 TSE image are eliminated in the T2 TSE BLADE improving the overall image quality.
E. Lavdas et al. / Magnetic Resonance Imaging 33 (2015) 194–200
subjects were scanned with the corresponding BLADE sequence, and those images collected in the same patient with the BLADE approaches did not show any noticeable artifacts. In T2 TIRM SAG sequence, there were 7 uncooperative subjects, 5 of which had images that were graded as non-diagnostic (Fig. 2). These subjects were scanned with corresponding BLADE sequence, and those images were graded as diagnostic. Regarding the truncation artifacts shown in the conventional sequences, T2 TSE SAG BLADE reduced them in 8 cases and T2 TIRM SAG BLADE in 12 cases. The results of the qualitative analysis including all the subjects are presented in Table 3. The BLADE sequences were superior to the corresponding conventional ones regarding most of the image characteristics. Based on the qualitative results of T2 TSE SAG and T2 TSE SAG BLADE sequences it is apparent that BLADE is significantly superior to the conventional one with statistically significant differences in terms of: 1) overall image quality (OIQ) (p b 0.01), 2) contrast of the spinal cord-CSF interface (CSCCSF) (p b 0.01) (Figs. 3 and 4, A and B), 3) contrast at the vertebral disc– CSF interface (CDCSF) (p b 0.01) (Fig. 4, A and B), 4) motion artifacts (MotArt) (p b 0.001) (Fig. 2, C and D), 5) truncation artifacts (TruArt) (p b 0.001) (Fig. 4, A and B), 6) flow artifacts (FlwArt) (p b 0.01) (Fig. 4, A and B). From all the comparisons of T2 TIRM SAG and T2 TIRM SAG BLADE, it appears that BLADE was better than the conventional one with statistically significant differences in 1) CSF (p b 0.01) (Fig. 3, C, D, E and F, Fig. 4, C and D), 2) CSCCSF (p b 0.01) (Fig. 3, C, D, E and F, Fig. 4, C and D), 3) TruArt (p b 0.001), 4) FlwArt (p b 0.01) (Fig. 4, C and D).
Fig. 2. Uncooperative subjects. Left: T2 TIRM sagittal (A and C) and right: T2 TIRM sagittal BLADE (B and D) images of the cervical spine. It is shown that motion artifacts that are shown in T2 TIRM are eliminated in the T2 TIRM BLADE sequence, which results in an improvement of the image quality. Especially, in the lower case, where the motion artifacts were very extensive in the conventional sequence, the BLADE sequence made the image diagnosable revealing the metastasis at the vertebral body.
197
Table 3 Summary of the results of the qualitative comparison between the different sequences.
OIQ CSF CMADJ CSCCSF CDCSF MotArt TruArt FlwArt IndArt
T2 TSE SAG
T2 TSE SAG BLADE
T2 TIRM SAG
T2 TIRM SAG BLADE
1.8 3.3 3.0 2.2 2.3 2.2 2.2 2.3 2.4
⁎ 2.8 3.7 3.2 ⁎ 2.9 ⁎ 3.0 ⁎⁎ 3.6 ⁎⁎ 2.9 3.1 ⁎ 3.1
1.0 2.5 2.8 1.7 1.9 2.1 1.8 1.7 2.5
1.6 3.0 3.2 ⁎ 2.5 2.4 3.0 ⁎⁎ 2.8 ⁎ 2.5 2.7
± ± ± ± ± ± ± ± ±
1.5 0.9 0.7 1.3 1.3 1.4 1.0 1.5 1.3
± ± ± ± ± ± ± ± ±
1.2 0.5 0.6 1.1 0.9 0.5 0.8 0.6 0.6
± ± ± ± ± ± ± ± ±
1.0 0.6 0.8 1.1 1.1 1.3 1.1 1.0 1.3
± ± ± ± ± ± ± ± ±
1.1 0.7 0.7 0.9 0.8 0.9 0.8 1.0 1.2
OIQ: overall image quality, CSF: cerebrospinal fluid bright, CMADJ: conspicuousness of morphologic abnormalities in the discovertebral junction, CSCCSF: contrast at the spinal cord (cauda equina)–CSF interface, CDCSF: contrast at the disc–CSF interface, MotArt: motion artifacts, TruArt: truncation artifacts, FlwArt: flow artifacts, IndArt: indentation artifacts. The statistical significance of the qualitative data was determined by the Kruskal–Wallis non-parametric test. ⁎ p b 0.01. ⁎⁎ p b 0.001.
4. Discussion In studies employing T2 TSE SAG sequences, the BLADE technique has been reported to eliminate motion, truncation and flow artifacts and improve the image quality [2,13]. In the present study, the T2 TSE BLADE SAG sequence was applied as indicated by its standardized protocol (manufacturer's default) with the only difference being that the REST slabs were applied parallel to the FOV exactly like they are in the conventional T2 TSE SAG sequence. This amendment was made in order to compatibly compare the conventional and BLADE techniques, but also in order to eliminate the artifacts of the anterior cervical region, which were mainly due to swallowing and respiration movements. In trial tests that we performed it was observed that when the REST slabs were perpendicular to FOV, a lot of artifacts appeared in the spinal cord, which might be due to swallowing. In most implementations the readout covers the same extent of k-space; this provides a more uniform extent of k-space coverage in all in-plane directions, whereas the phase encoding direction of conventional Cartesian sampling is often limited relative to the readout direction, and so it is more prone to truncation artifacts. The downside of BLADE is that if the same number of k-space lines is acquired in total, there are some wedge shaped gaps in the outer areas that may lead to artifacts in their own right. To limit this factor it is typical to perform an oversampling which generally lengthens scan time. However, when we increased the scan time of the conventional sequences, we did not see any significant improvement of image quality. Indentation artifacts are related to wrap artifacts. According to other authors, the reduction of indentation artifacts is achieved when REST slabs are applied perpendicularly to the FOV or when the FOV is increased (this way a concomitant resolution loss is observed) [2,13,26,27]. In this study, we have used 100% oversampling, and consequently we did not have wrap artifacts. This would have an impact on the indentation artifacts too. Furthermore, the REST slabs were applied parallel to FOV, which remained constant. In this way, in two cases where indentation artifacts were visible, the diagnostic value of the image was not downgraded at all, and wrap artifacts were significantly reduced. According to our qualitative results, the T2 TSE SAG BLADE sequence was always superior to the conventional T2 TSE SAG with the differences being statistically significant in almost all cases. More specifically, the BLADE technique achieved better overall image quality (p b 0.01), better contrast between SC/CSF and vertebral disc/CSF interface (p b 0.01). All the image artifacts were reduced remarkably. Motion and truncation artifacts, which are very often in
198
E. Lavdas et al. / Magnetic Resonance Imaging 33 (2015) 194–200
Fig. 4. Cooperative subjects. Upper: In T2 TSE sagittal (A) sequence, the flow artifacts in the posterior of the vertebrae of the 4th and 5th vertebrae (the arrows point out those positions) results in an image that indicates that there is no CSF between the vertebra and spinal cord, whereas in the T2 TSE sagittal BLADE (B) sequence, CSF is clearly shown. A better contrast is observed between spinal cord and CSF in the posterior part of spinal cord. Also, the truncations artifacts have been eliminated too, which results in better visualization of CSF in the 4o cervical vertebral body. Lower: The flow artifacts in CSF posteriorly of the 6th and 7th cervical vertebral body (the arrows point out those positions) that are seen in T2 TIRM sagittal (C) are reduced by the T2 TIRM sagittal BLADE (D) sequence. Also, the BLADE sequence eliminates the truncation artifacts in spinal cord. The final result is a better visualization of spinal cord and better contrast between CSF and spinal cord as well as between CSF and cervical vertebral body.
Fig. 3. Cooperative subjects. Upper: The motion and swallowing artifacts that are seen in the T2 TSE image (A) are eliminated in the T2 TSE BLADE (B) along the spinal cord. Also, the BLADE sequence reduces the truncation artifacts, which results in a better visualization of spinal cord. Middle: In T2 TIRM Sagittal BLADE (D), a better contrast between the spinal cord and the CSF as well as between disc and CSF is observed compared to the conventional image (C). Also in this case, the BLADE sequence reduces the truncation artifacts (the arrow point out this artifact). Lower: In T2 TIRM sagittal BLADE (F), a higher image quality and better CSF visualization are shown compared to the conventional sequence (E). The reduction of the truncation artifacts is also evident in this case.
conventional sequences, were not only reduced by the BLADE technique, but the differences between the corresponding sequences were found to be statistically significant (p b 0.001) (Fig. 1, A and B). The flow artifacts were also reduced with the differences being again statistically significant (p b 0.01) (Fig. 4, A and B). Our results are in line with those of other authors [2,23,28]. More specifically, Fellner et al. found that BLADE sequence is superior to the conventional one with statistically significant differences in image sharpness (p b 0.001), image motion, truncation and flow artifacts (p b 0.001), contrast between vertebral body and vertebral disc, spinal cord and CSF (p b 0.001) and in the diagnostic reliability of
spinal cord (p b 0.001) [2]. Fellner et al. found that the BLADE sequence is superior to the conventional one with statistically significant difference in overall image quality [2]. It is known that T2 TIRM sequences suppress fat revealing the pathology, including regions of high inhomogeneity This is a very important property especially in cervical spine MRI, which is prone to artifacts and where spatial pre-saturation cannot be achieved [29]. The present study is among the first to implement T2 TIRM SAG BLADE in MR imaging at a different however clinical case, and due to the very promising results received this sequence was also implemented in cervical spine MRI as well. It is noticeable that the conspicuity and demarcation of the lesions are similar in the T2 TIRM SAG and T2 TSE SAG sequences. Here, the parameters of the T2 TIRM SAG BLADE sequence were kept the same with the conventional sequence with the only difference being that the values of the ETL and the rBw parameters were increased. An increase of the rBw leads to the reduction of several artifacts such as the truncation artifacts, motion artifacts, chemical shift artifacts, susceptibility artifacts, T2 blurring in TSE and ghosting artifacts in EPI [18,19], while an increase of the ETL leads to the reduction of truncation and motion artifacts. The T2 TIRM SAG sequence shows more artifacts than the T2 TSE SAG, and one of the reasons is the larger TR that is used in the TIRM
E. Lavdas et al. / Magnetic Resonance Imaging 33 (2015) 194–200
sequences. Furthermore, due to fat saturation the artifacts are more visible because they are shown as parallel lines of increased signal. Since the conventional sequences show more artifacts, BLADE sequences have a larger margin of artifact elimination and in this way to show a clear improvement in image quality. As mentioned above, the application of BLADE sequences is related to increased values of ETL, rBw and TR. Consequently, these parameters are by necessity different from conventional sequences. Due to the fact that the values of these parameters are related to the intrinsic characteristics of the BLADE sequences the presented results are mostly sequence related. In a previous case report on cervical spine [13], for two cases with multiple sclerosis, image quality was significantly improved by the use of T2 TIRM BLADE and T2 TSE BLADE sequences compared to the conventional sequences. Although TIRM achieves fat suppression, due to the high TR [27], this is accomplished at the expense of having more motion, truncation and flow artifacts, which however get satisfactorily reduced by the T2 TIRM SAG BLADE sequence (Fig. 2). Moreover, in a previous study it was found that when sequences have fat suppression the pulsation artifacts are more prominent in dark regions [28]. Knowing the great usefulness of the TIRM sequences we believe that the use of TIRM BLADE sequences will be very helpful and beneficial not only in the imaging of uncooperative subjects but also for subjects with multiple sclerosis due to the fact that they visualize very clearly spinal cord (Fig. 3, C, D, E and F). Based on our results in the imaging of cooperative subjects using the T2 TIRM SAG and T2 TIRM SAG BLADE sequences, it is clear that the BLADE sequence is significantly superior to the conventional one due to the reasons mentioned above. More specifically, statistically significant differences between the TIRM and the TIRM BLADE sequences were found in terms of contrast of the spinal cord- CSF interface (p b 0.01) and truncation artifacts (p b 0.01). The truncation artifacts known as Gibbs ringing are shown as bright or dark lines, that are seen parallel and next to borders of abrupt intensity change, and they may emulate a syrinx on sagittal images of spinal cord. These artifacts are related to the finite number of encoding steps used by the Fourier transform. It was observed that the T2 TSE SAG BLADE and T2 TIRM SAG BLADE sequences eliminated the truncation artifacts especially in two cases where syringomyelia was observed. The reason BLADE images had less truncation artifact is two-fold: 1) there were more phase encoding steps and 2) rotating through k-space with BLADE helps reduce truncation artifacts in the phase-encoding direction. Regarding the visualization of CSF in the T2 TIRM SAG and T2 TIRM SAG BLADE sequences, in the latter CSF has better SNR because this sequence reduces flow artifacts which reduce the SIs and therefore the SNR of CSF (Fig. 4, C and D). BLADE sequences restrict also the truncation artifacts, which may be the second reason for the redaction of SIs and therefore the SNR of CSF in the conventional TIRM sequence. One of the floating structures which are known for causing ghosting artifacts is the CSF. The signal of CSF splits when artifacts are created in the normal position and the region of overlaid artifacts. When it does not generate artifacts, the T2 TIRM BLADE sequence has higher signal. This is also observed in the qualitative results of the T2 TSE SAG BLADE and T2 TIRM SAG BLADE images, and it is clearly visible in cases where there were stenoses in spinal canal. This is due to the fact that in conventional sequences, flow artifacts are more intensive in borderline cases (such as stenosis), where it is observed that although the conventional sequences had a weak signal in the CSF, the signal was higher in the BLADE sequences. This phenomenon was observed here in the qualitative measurements of both sets of sequences, and it occurs mostly in the conventional sequences, which are more prone to flow artifacts. It is known that human eye can distinguish differences in borderline cases, such as a weak SNR of CSF in conventional sequences
199
against a very high SNR of CSF in BLADE sequences (Fig. 4, C and D). In these cases (stenosis with flow artifacts), the SNR measurements of CSF with BLADE were roughly double those of the T2 TSE Sag sequence. However, in the quantitative measurements (Table 2) the CSF was measured, and statistically significant difference was found only in the cervical foramen, where there were no flow artifacts and partial volume effect. Also, in 8 cases (4 in T2 TSE SAG and 4 in T2 TIRM SAG), the conventional sequences showed that there was no CSF space between the disk herniation and spinal cord, whereas in the corresponding BLADE sequences the presence of CSF was seen. CSF brightness plays a very significant role for the radiological evaluation. The above observation may seem as a minor detail, but it has great clinical importance for the subject and his treatment. It is very important to be able to assess whether cerebrospinal fluid exists between the vertebral body and spinal cord. This issue is related to the limitation of the subarachnoid space, which is the cause of the lower movement of the CSF and the disturbance of the metabolism of nervous tissue in the various syndromes. As indicated by the presented quantitative and qualitative results the proposed methodology appears to be objective and reliable for evaluating CSF brightness. BLADE sequences manage to deal with the aforementioned artifacts mainly through the repeated sampling of the centre of k-space and the greater extent of k-space in all in-plane directions. The weaknesses of BLADE are the potential manifestation of indentation artifacts and the larger acquisition time. In this study, the slice thickness and the FOV were identical for the examined pairs of sequences. Due to the fact that matrix size was different, normalization to the voxel size was performed so that the results of the different sequences refer to the same voxel size. It is known that TR and TE parameters influence directly the SNR results and therefore image quality. More specifically, Fellner et al. [2] state that a short TR and a short TE increase SNR and reduce acquisition time while maintaining sufficient T2 contrast. Additionally, Malamateniou et al. [27] point out that a high TR increases the derivation possibility of motion artifacts. In this study, the analysis was performed with increased values for the TR and TE parameters in the T2 TIRM BLADE SAG sequence even though this was a disadvantage for the BLADE sequences. Despite that, the BLADE sequences always demonstrated higher SNR and CNR values than the conventional sequences. These findings are in-line with the purpose of this study, which is to show that BLADE sequences are useful. In conclusion, the use of BLADE sequences in cervical spine MR examinations appears to be capable of potentially eliminating motion, pulsatile flow and truncation artifacts. Furthermore, BLADE sequences are proposed to be used in the standard examination protocols based on the fact that a significantly improved image quality could be achieved. Finally, we suggest that the clinical use of T2 TIRM SAG BLADE due to its capacity to reduce image artifacts and achieve better contrast. References [1] Cohen WA, Maravilla KR. MRI atlas of the spine. London: Martin Dunitz Ltd; 1991. [2] Fellner C, Menzel C, Fellner FA, Ginthoer C, Zorger N, Schreyer A, et al. BLADE in sagittal T2-weighted MR imaging of the cervical spine. AJNR Am J Neuroradiol 2010;31:674–81. [3] Bailes DR, Gilderdale DJ, Bydder GM. Respiratory ordering of phase encoding (ROPE): a method for reducing respiratory motion artifacts in MR imaging. J Comput Assist Tomogr 1985;9:835–8. [4] Pattany PM, Phillips JJ, Chiu LC. Motion artifact suppression technique (MAST) for MR imaging. J Comput Assist Tomogr 1987;11:369–77. [5] Haacke EM, Lenz GW. Improving MR image quality in the presence of motion by using rephrasing gradients. AJR 1987;148:1251–8. [6] Felmlee JP, Ehman RL. Spatial presaturation: a method for suppressing flow artifacts and improving depiction of vascular anatomy in MR imaging. Radiology 1987;164:559–64. [7] Dixon WT, Brummer ME, Malko JA. Acquisition order and motional artifact reduction in spin warp images. Magn Reson Med 1988;6:74–83.
200
E. Lavdas et al. / Magnetic Resonance Imaging 33 (2015) 194–200
[8] Runge VM, Wood ML, Kaufman DM, Traill MR, Nelson KL. The straight and narrow path to good head and spine MRI. Radiographics 1988;8:507–31. [9] Kallmes DF, Hui FK, Mugler III JP. Suppression of cerebrospinal fluid and blood flow artifacts in FLAIR MR imaging with a single-slab three-dimensional pulse sequence: initial experience. Radiographics 2001;221:251–5. [10] Naganawa S, Satake H, Iwano S, Kawai H, Kubota S, Komada T, et al. Contrastenhanced MR imaging of the brain using T1-weighted FLAIR with BLADE compared with a conventional spin-echo sequence. Eur Radiol 2008;18:337–42. [11] Alibek S, Adamietz B, Cavallaro A, Stemmer A, Anders K, Kramer M, et al. Contrast enhanced T1-weighted fluid-attenuated inversion-recovery BLADE magnetic resonance imaging of the brain: an alternative to spin-echo technique for detection of brain lesions in the unsedated pediatric patient? Acad Radiol 2008; 15:986–95. [12] Pipe JG. Motion correction with PROPELLER MRI: application to head motion and free-breathing cardiac imaging. Magn Reson Med 1999;42:963–9. [13] Ragoschke-Schumm A, Schmidt P, Schumm J, Reimann G, Mentzel HJ, Kaiser WA, et al. Decreased CSF-flow artefacts in T2 imaging of the cervical spine with periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER/BLADE). Neuroradiology 2011;53:13–8. [14] Pipe JG. An optimized center-out k-space trajectory for multishot MRI: comparison with spiral and projection reconstruction. Magn Reson Med 1999; 42:714–20. [15] Campi A, Pontesill S, Gerevini S, Scotti G. Comparison of MRI pulse sequences for investigations of lesions of cervical spinal cord. Neuroradiology 2000;42:669–75. [16] Wesselly M, Cervical MRI. PART 1 a basic overview. USA: Northwestern Health Sciences University; 2003. [17] Finkenzeller T, Zorger N, Kühnel T, Paetzel C, Schuierer G, Stroszczynski C, et al. Novel application of T1-weighted BLADE sequences with fat suppression compared to TSE in contrast-enhanced T1-weighted imaging of the neck: cutting-edge images? J Magn Reson Imaging 2013;37:660–8. [18] Celik A, Elmaoglu M. MRI Handbook. MR physics, patient positioning and protocols. New York: Springer; 2012.
[19] Lavdas E, Mavroidis P, Vassiou K, Roka V, Fezoulidis IV, Vlychou M. Elimination of chemical shift artifacts of thoracic spine with contrast enhanced Flair imaging with fat suppression at 3.0 Tesla. Magn Reson Imaging 2010;28:1535–40. [20] Shapiro MD. MR imaging of the spine at 3 T. Magn Reson Imaging Clin N Am 2006;14:97–108. [21] Tanenbaum LN. Clinical 3 T MR Imaging: mastering the challenges. Magn Reson Imaging Clin N Am 2006;14:1–15. [22] Tamhane AA. Motion correction in periodically-rotated overlapping parallel lines with enhanced reconstruction (PROPELLER) and turboprop MRI. Magn Reson Med 2009;62:174–82. [23] Lavdas E, Mavroidis P, Topaltzikis T, Slatinopoulos V, Roka V, Vlachopoulou A, et al. Comparison of BLADE and conventional T2-TSE sequences for the sagittal visualization of the cervical spinal cord in multiple sclerosis patients - a case report. Magn Reson Imaging 2013;31:1766–70. [24] Yoshioka H, Stevens K, Genovese M, Dillingham MF, Lang P. Articular cartilage of knee: normal patterns at MR imaging that mimic disease in healthy subjects and patients with osteoarthritis. Radiology 2004;231:31–8. [25] Philpott C, Brotchie P. Comparison of MRI sequences for evaluation of multiple sclerosis of the cervical spinal cord at 3 T. Eur J Radiol 2011;80:780–5. [26] Rocca MA, Mastronardo G, Horsfield MA, Pereira C, Iannucci G, Colombo B, et al. Comparison of three MR sequences for the detection of cervical cord lesions in patients with multiple sclerosis. AJNR Am J Neuroradiol 1999;20:1710–6. [27] Malamateniou C, Malik SJ, Counsell SJ, Allsop JM, McGuinness AK, Hayat T, et al. Motion-compensation techniques in neonatal and fetal MR imaging. 2013;34: 1124–36. [28] Lavdas E, Mavroidis P, Hatzigeorgiou V, Roka V, Arikidis N, Oikonomou G, et al. Elimination of motion and pulsation artifacts using Blade sequences in knee MR imaging. Magn Reson Imaging 2012;30:1099–110. [29] Mahnken AH, Wildberger JE, Adam G, Stanzel S, Schmitz-Rode T, Günther RW, et al. Is there a need for contrast-enhanced T1-weighted MRI of the spine after inconspicuous short tau inversion recovery imaging? Eur Radiol 2005;15: 1387–92.
