| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Neuroradiology |
1 From the Department of Radiology, University of Bonn, Sigmund-Freud-Strasse 25, D-53105 Bonn, Germany (W.A.W., M.v.F., B.N., H.H.S., C.K.K.); and Philips Medical Systems, Hamburg, Germany (J.G.). From the 2001 RSNA scientific assembly. Received March 18, 2002; revision requested June 4; final revision received December 13; accepted January 10, 2003. Address correspondence to W.A.W. (e-mail: willinek@uni-bonn.de).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: This intraindividual comparative study included 62 patients suspected of having cerebral ischemia. Patients underwent T2-weighted fluid-attenuated inversion-recovery (FLAIR) (n = 62), T2-weighted turbo spin-echo (TSE) (n = 48), and single-shot echo-planar diffusion-weighted imaging (n = 27) with standard sequential and SENSE MR acquisitions with a 1.5-T magnet and phased-array coil. With SENSE, acquisition time was reduced from 1 minute 12 seconds to 35 seconds for FLAIR and from 1 minute 18 seconds to 39 seconds for T2-weighted TSE imaging. For diffusion-weighted imaging, echo train length was shortened (78 vs 71 msec) to reduce susceptibility effects while acquisition time was maintained. Two radiologists scored quality of standard and SENSE images with a five-point scale and assessed presence of artifacts (motion, susceptibility) and lesion conspicuity. To assess statistical significance, Wilcoxon signed rank and 
2 tests were used.
RESULTS: Statistical analysis revealed no significant difference in terms of image quality and presence of artifacts between standard and SENSE T2-weighted TSE (image quality, P = .724; presence of artifacts, P = .378) and FLAIR (image quality, P = .127; presence of artifacts, P = .275) images. Image quality at SENSE diffusion-weighted imaging was scored significantly higher compared with that at standard diffusion-weighted imaging (P = .002). Susceptibility artifacts were significantly reduced at SENSE diffusion-weighted imaging when compared with those at standard diffusion-weighted imaging (P < .001). Conspicuity of 84 lesions was rated equivalent with both standard and SENSE protocols.
CONCLUSION: SENSE allowed acquisition of T2-weighted TSE and FLAIR images with image quality and lesion conspicuity that did not differ from those of standard acquisition techniques but in only half the acquisition time. Use of SENSE with diffusion-weighted imaging significantly reduces susceptibility artifacts while lesion conspicuity is maintained.
© RSNA, 2003
Index terms: Brain, ischemia, 10.781 • Brain, MR, 10.121411, 10.121413, 10.12144 • Magnetic resonance (MR), technology
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
As magnetic resonance (MR) imaging becomes more and more available in stroke centers, nowadays more than 40% of patients who are admitted to the hospital because they are suspected of having ischemic stroke undergo MR imaging (3).
In the setting of ischemic stroke, MR imaging is a demanding task. Dedicated MR imaging protocols need to be used to ensure an optimum diagnostic yield, including structural, functional, and angiographic information (4), to remain within the given tight time constraints and to assist in imaging of a usually restless and uncooperative patient.
MR imaging protocols for assessment of acute stroke include sets of sequences that vary widely among institutions, and there is no commonly accepted standard. Yet, there is broad agreement that diffusion-weighted MR imaging plays a pivotal role in the early detection of stroke (5–9). In addition, a variety of pulse sequences (ie, usually fast sequences) are in use to delineate structural damage, edema, or preexisting cerebral disease.
Recently, parallel imaging techniques such as simultaneous acquisition of spatial harmonics, or SMASH, and sensitivity encoding (SENSE) have been proposed to markedly reduce image acquisition time (10–12) or to prevent susceptibility artifacts with reduction of echo train length in diffusion-weighted imaging (11,13). Because short acquisition times are vital in the setting of acute stroke, and because diffusion-weighted imaging plays such a pivotal role in early stroke assessment (14–16), parallel imaging seems ideally suited to enhance dedicated MR imaging protocols for patients with stroke. The objective of this study was to evaluate the SENSE technique in a clinical setting for imaging of patients who are suspected of having a stroke.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
|
All patients underwent transverse T2-weighted FLAIR (n = 62), transverse T2-weighted TSE (n = 48), and transverse single-shot echo-planar diffusion-weighted (n = 27) imaging by using both standard sequential and SENSE acquisitions. Patients were randomized for the order in which the standard and SENSE acquisitions were performed. A reduction factor (R) of two for phase encoding was used in all SENSE sequences. After a fast scout image was obtained, a reference pulse sequence (sensitivity mapping) was acquired (acquisition time, 56 seconds). Concurrently, the actual imaging sequences were planned. To compensate for signal inhomogeneity attributable to the flexible surface coils, a homogeneity correction (constant level appearance, or CLEAR) with sensitivity mapping that is automatically implemented in all SENSE protocols was also performed with the standard sequences.
Acquisition parameters for T2-weighted FLAIR were as follows: TR/TE/inversion time msec, 6,000/100/2,000; rectangular field of view, 80%; image matrix, 256 x 256 on a 230-mm field of view; sections, 22; and section thickness, 5 mm. With SENSE, acquisition time was reduced from 1 minute 12 seconds to 35 seconds. Acquisition parameters for T2-weighted TSE were as follows: 5,336/120; rectangular field of view, 80%; image matrix, 372 x 496 on a 230-mm field of view; sections, 22; and section thickness, 5 mm. With SENSE, acquisition time was reduced from 1 minute 18 seconds to 39 seconds.
Acquisition parameters of diffusion-weighted imaging were as follows: rectangular field of view, 90%; image matrix, 108 x 128 on a 240-mm field of view; sections, 22; and section thickness, 6 mm. For echo-planar diffusion-weighted imaging, echo train length was shortened with SENSE (TR, 5,311 vs 4,977 msec; TE, 78 vs 71 msec) to reduce susceptibility effects while the acquisition time (45 vs 47 seconds) was maintained. Echo planar imaging factors were 93 and 49, respectively.
Acquisition parameters of the pulse sequences are listed in Table 1.
All patients underwent a transverse T1-weighted gradient-echo sequence with a very short acquisition time of 29 seconds (185/2.2; sections, 22; section thickness, 6 mm). Because there was no improvement expected with SENSE for this already very short acquisition time, this pulse sequence was not included in the actual study protocol.
SENSE Technique
For a detailed explanation of the SENSE technique, the reader is referred to Pruessmann et al (11). In brief, SENSE makes use of the spatially varying sensitivity of receiver coils for encoding of spatial information. By doing so, the technique enables one to reduce the number of Fourier encoding steps by increasing the distance between sampling lines in cartesian k space (11). As a consequence of the undersampled k space, a Fourier transformation of the data will first create an aliased (ie, folded) image for each receiver coil element. For SENSE image reconstruction, "sensitivity maps" for the receiver coil elements are generated with the built-in body coil of the system.
SENSE reconstruction then uses the sensitivity encoding of the array coils to unfold the aliased signal components (11). The reduction of the number of phase-encoding steps for SENSE in comparison with full Fourier encoding is indicated with the SENSE reduction factor R. The signal-to-noise ratio (SNR) is inversely proportional to the square root of R (11). Noise enhancement occurs when the geometric relationship of the coil sensitivities is not optimal; in our setup, this applies for the image center (ie, interhemispheric cleft on transverse images). This SENSE-specific noise enhancement is indicated with the local geometry factor g. Thus, in any given SENSE image, SNR compared with that on a standard Fourier image will be as follows: SNRSENSE/SNRstandard = 1/g
R (11). Results from simulations show that at R = 2, g is close to one (11,17).
Image Evaluation
Two radiologists (C.K.K., W.A.W.) were blinded to the image acquisition technique and to results of other diagnostic tests, but they were aware of each patient’s clinical history. Evaluation criteria for image quality were as follows: Score 1 was assigned for excellent diagnostic image quality if no ghosting or motion artifacts were present, SNR appeared high, fine anatomic details (eg, cranial nerves) were clearly visualized, and gray matter–to–white matter contrast was high. Score 2 was assigned for more than adequate diagnostic image quality in cases with minor ghosting or motion artifacts or in cases in which SNR appeared somewhat reduced but otherwise comparable to that of score 1. Score 3 was assigned for adequate diagnostic image quality in cases with minor ghosting or motion artifacts and in which SNR appeared somewhat reduced but otherwise comparable to that of scores 1 and 2. Score 4 was assigned for poor diagnostic image quality in cases in which image quality was impaired by ghosting and motion artifacts and in which SNR appeared reduced and fine anatomic details were poorly visualized or gray matter–to–white matter contrast was poor so that the diagnostic value of the images for diagnosis was questionable. Score 5 was assigned for nondiagnostic image quality if image quality was heavily impaired by ghosting and motion artifacts and low SNR appearance and in cases in which fine anatomic details were not visualized and gray matter–to–white matter contrast was low so that no final diagnosis could be determined.
For quantitative evaluation, the SNR was calculated from all images for each patient by using the largest possible region of interest in the right periventricular white matter. SNR was determined as the mean value of signal intensity in the cerebral white matter divided by the SD of the signal intensity of noise divided by ghosting-free background. To ensure consistency, regions of interest were placed by the same author (B.N.) and copied onto the images.
Presence of artifacts was evaluated. Artifacts were categorized as being attributable to regular MR image degradation secondary to ghosting, pulsation, motion, or susceptibility effects or as representing SENSE-related artifacts. SENSE-related artifacts were defined as artifacts arising from errors in sensitivity values. The most probable typical systematic error in sensitivity values is regional over- or underestimation (11). At unfolding of the raw data, these sensitivity errors cause periodic artifacts in the reconstructed images. Artifacts were rated to be present (grade 2, pronounced; grade 1, minimal) or absent (grade 0).
Lesion conspicuity was rated for both the standard and SENSE images according to a four-point scale: grade 1, excellent lesion conspicuity; grade 2, moderate lesion conspicuity; grade 3, fair lesion conspicuity; and grade 4, lesion not visible. For this purpose, the lesion conspicuity in the sequentially acquired standard image served as reference for comparison. The readers were asked to indicate a diagnosis twice per case with consensus: once on the basis of findings obtained with the SENSE pulse sequences and once on the basis of findings obtained with the standard sequences ("final diagnosis"). Validation of the final diagnosis was performed on the basis of results from other diagnostic tests, such as computed tomography and digital subtraction angiography, short-term MR imaging follow-up (within 1 month), and the clinical course of the patient.
Statistical Analysis
Statistical analyses were performed by using the Wilcoxon signed rank and 2 tests with software (SPSS, version 10.0.7; SPSS, Chicago, Ill). A paired Wilcoxon signed rank test was used to test results of grading image quality and lesion conspicuity of images obtained with the SENSE and standard acquisitions. To analyze results of scoring the presence or absence of artifacts, a 2 test was performed. A difference with P < .05 indicated statistical significance.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Image Quality
Quality of images obtained with the standard T2-weighted FLAIR sequence was assigned a score of 1 (excellent) or 2 (more than adequate for diagnosis) in 49 (79%) of 62 patients, whereas the quality of images obtained with the SENSE T2-weighted FLAIR sequence was assigned this score in 44 (71%) patients. The median score was 2 in both cases (P = .127, Table 2). Quality of images obtained with the standard T2-weighted TSE sequence was assigned a score of excellent or more than adequate for diagnosis in 39 (81%) of 48 patients, and quality of images obtained with the SENSE sequence was assigned this score in 44 (92%) patients (Fig 1). The median scores were 1 and 2 (P = .724), respectively. Quality of diffusion-weighted images was assigned a score of excellent or more than adequate for diagnosis for images obtained in 96% (26 of 27) of the standard examinations (median score, 2) in comparison with the quality of images obtained in 100% (27 of 27) of the SENSE examinations (median score, 2; P = .002).
|
|
|
|
|
|
SNR Results
For the standard and the SENSE protocols, respectively, mean SNRs were calculated as follows: T2-weighted FLAIR, 28.10 ± 7.23 (SD) and 19.43 ± 5.69; T2-weighted TSE, 36.85 ± 6.15 and 30.65 ± 4.01; and diffusion-weighted imaging, 54.04 ± 4.44 and 37.37 ± 3.27 (Table 4). Mean region-of-interest size used for calculation was as follows: 334.47 mm2 ± 47.19 (range, 232.97–455.63 mm2). The lower SNR with the SENSE protocol of all pulse sequences as compared with that with the standard protocol did not perceivably affect the image interpretation.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
In our study, SENSE allowed acquisition of T2-weighted TSE and FLAIR images in only half the standard acquisition time but with diagnostic image quality and lesion conspicuity that did not differ from those of standard acquisition techniques. Although degradation of image quality caused by head movements is a major problem in patients who are suspected of having an acute stroke, reduction of acquisition time with SENSE may help to prevent motion artifacts. With SENSE, the prevalence of motion artifacts was reduced from 47% (29 of 62) to 37% (23 of 62) with the T2-weighted FLAIR sequence and from 27% (17 of 62) to 21% (13 of 62) with the T2-weighted TSE sequence as compared with the standard protocol.
However, in three patients, periodic artifacts of superficial fat tissue were seen with both the standard and SENSE T2-weighted FLAIR and TSE images. These patients were particularly restless and moved their heads during the image acquisition. Because the artifacts were prominent on both the SENSE and standard images, we suppose that these artifacts were caused by movement and displacement of the two flexible phased-array coils. In the future, further efforts should be made to evaluate the tolerable amount of coil movement and to design specific rigid phased-array coils for SENSE imaging of the brain.
Single-shot diffusion-weighted imaging is generally more robust against motion artifacts in comparison with other pulse sequences (eg, T2-weighted FLAIR and TSE sequences), and motion artifacts were absent at diffusion-weighted imaging in our study. Instead of motion, inhomogeneity (susceptibility) typically impairs image quality at echo-planar diffusion-weighted imaging (5). Shortening of the echo train length has been demonstrated to reduce such inhomogeneity artifacts (11,13). With a reduced echo train length at SENSE diffusion-weighted imaging as compared with standard diffusion-weighted imaging (TE, 78 msec instead of 71 msec), prevalence of susceptibility artifacts was significantly reduced in our study while total acquisition time was maintained. As a result, overall image quality at diffusion-weighted imaging was significantly improved in combination with SENSE. The Wilcoxon signed rank test that was used to test for statistical significance, however, has some limitations. Owing to the relatively large number of ties in the data, the effective sample size was smaller than the original study population (T2-weighted FLAIR: grading of image quality, n = 21; T2-weighted TSE: grading of image quality, n = 16; diffusion-weighted imaging: grading of image quality, n = 10).
Decrease in SNR inherent to SENSE is well known (11) and was also observed in our study (Table 4). According to the experimental and first clinical data (11,17,18), the reduction in SNR is characterized by the square root of the reduction factor R (ie, R = 2). Any reduction of SNR in cerebral imaging is in principle associated with the risk of deterioration in the detection of subtle lesions. Yet, it is important to notice that the lower SNR with the SENSE acquisitions did not perceivably interfere with the diagnostic image interpretation and did not impair conspicuity of ischemic lesions in the obtained SENSE images even in cases with very small and subtle lesions in our study cohort.
In all 36 patients, acute cerebral ischemia was correctly identified by using the fast SENSE protocol and was confirmed by using standard imaging. In addition, five of 62 patients who had acute or metachronous hemorrhage at presentation were correctly identified by using both the standard and SENSE sequence protocols. Although the number of study subjects was relatively small, our findings suggest that a reduced imaging protocol and faster brain imaging with SENSE can provide sufficient information for the diagnosis in patients with stroke.
To date, the use of parallel imaging for fast brain MR imaging in patients with stroke has not been evaluated. It was our aim to evaluate the feasibility of implementation of the SENSE technique for brain imaging in patients who are suspected of having a cerebral infarction. The results of our study are promising; in patients with stroke, SENSE allowed structural MR imaging of the brain with image quality and lesion conspicuity that did not differ from those of standard techniques but in half the acquisition time, and it allowed diffusion-weighted imaging with significantly reduced susceptibility artifacts. We conclude that SENSE imaging may well replace standard structural and diffusion brain imaging, particularly in restless patients who are suspected of having an acute stroke.
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, W.A.W., H.H.S., C.K.K.; study concepts and design, W.A.W., J.G., C.K.K.; literature research, W.A.W., M.v.F., B.N.; clinical studies, W.A.W., M.v.F., B.N., C.K.K.; data acquisition and analysis/interpretation, W.A.W., M.v.F., B.N., C.K.K.; statistical analysis, W.A.W.; manuscript preparation, W.A.W., J.G., M.v.F.; manuscript definition of intellectual content, W.A.W., J.G., C.K.K.; manuscript editing, W.A.W., M.v.F., B.N.; manuscript revision/review, W.A.W., M.v.F., J.G., C.K.K., H.H.S.; manuscript final version approval, all authors
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  Y.A. Bhagat, D.J. Emery, S. Naik, T. Yeo, and C. Beaulieu Comparison of Generalized Autocalibrating Partially Parallel Acquisitions and Modified Sensitivity Encoding for Diffusion Tensor Imaging AJNR Am. J. Neuroradiol., February 1, 2007; 28(2): 293 - 298. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Yoshikawa, H. Kawamitsu, D. G. Mitchell, Y. Ohno, Y. Ku, Y. Seo, M. Fujii, and K. Sugimura ADC measurement of abdominal organs and lesions using parallel imaging technique. Am. J. Roentgenol., December 1, 2006; 187(6): 1521 - 1530. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J M U-King-Im, R A Trivedi, M J Graves, K Harkness, H Eales, I Joubert, B Koo, N Antoun, E A Warburton, J H Gillard, et al. Utility of an ultrafast magnetic resonance imaging protocol in recent and semi-recent strokes J. Neurol. Neurosurg. Psychiatry, July 1, 2005; 76(7): 1002 - 1005. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M Ryan, P Cunningham, C Cantwell, D Brennan, and S Eustace A comparison of fast MRI of hips with and without parallel imaging using SENSE Br. J. Radiol., April 1, 2005; 78(928): 299 - 302. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Taouli, A. J. Martin, A. Qayyum, R. B. Merriman, D. Vigneron, B. M. Yeh, and F. V. Coakley Parallel Imaging and Diffusion Tensor Imaging for Diffusion-Weighted MRI of the Liver: Preliminary Experience in Healthy Volunteers Am. J. Roentgenol., September 1, 2004; 183(3): 677 - 680. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
