Clinical Imaging 39 (2015) 765–769
Contents lists available at ScienceDirect
Clinical Imaging journal homepage: http://www.clinicalimaging.org
Original Articles
Readout-segmented echo-planar imaging in diffusion-weighted MR imaging of acute infarction of the brainstem and posterior fossa: comparison of single-shot echo-planar diffusion-weighted sequences Jonghyun Byeon a, Jee Young Kim b,⁎, A-Hyun Cho c a b c
Department of Radiology, St. Vincent’s Hospital, College of Medicine, The Catholic University of Korea, Suwon, Korea Department of Radiology, Yeouido St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul, Korea Department of Neurology, Yeouido St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul, Korea
a r t i c l e
i n f o
Article history: Received 13 February 2015 Received in revised form 2 May 2015 Accepted 1 June 2015 Keywords: Infarction Diffusion-weighted MRI Magnetic resonance imaging Bran stem
a b s t r a c t Objective: The purpose of this study was to evaluate the clinical usefulness of readout-segmented echo-planar imaging (rs-EPI) in the diagnosis of acute infarction of the brainstem and posterior fossa, by comparing its results with those of single-shot echo-planar imaging (ss-EPI) at 3-T magnetic resonance imaging. Materials and methods: Twenty-nine patients with acute infarctions of the brainstem and posterior fossa underwent both ss-EPI and rs-EPI. Two readers independently assessed two sets of diffusion-weighted (DW) images for the qualitative comparison of image quality. Signal-to-noise ratio (SNR), lesion contrast, and contrast-tonoise ratio (CNR) were calculated for the assessment of image parameters. Results: There were no significant differences in the conspicuity of acute infarction upon qualitative comparison; however, distinctions of anatomical structures, susceptibility artifact, the presence of uncertain high signal intensity in the brain parenchyma, and overall image quality were significantly better in rs-EPI DW images. There were no significant differences in SNR, lesion contrast, CNR, and apparent diffusion coefficient values of acute infarction and normal thalamus between rs-EPI and ss-EPI. Conclusion: rs-EPI DWI is a clinically useful technique for evaluating lesions in the brainstem and posterior fossa by producing high-resolution DW images with reduced susceptibility artifact. However, there are no significant differences in the conspicuity of acute infarctions in the brainstem and posterior fossa between rs-EPI and ss-EPI. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Diffusion-weighted imaging (DWI) is well established as a valuable technique for the diagnosis of acute infarction that provides high sensitivity and specificity [1]. Single-shot echo-planar imaging (ss-EPI) DWI is the current clinical standard due to its resistance to phase errors resulting from patient motion. However, susceptibility artifacts that manifest as geometric distortion, signal drop, and image blurring can be severe, particularly at 3 T, due to the long echo-planar imaging readout and corresponding low bandwidth per pixel in the phase-encoding direction [2]. These artifacts can be diminished by accelerating the kspace traversal along the phase-encoding direction [3]. Multishot or readout-segmented echo-planar imaging (rs-EPI) is used to sample a subset of k-space points in the readout direction at each shot, which allows a substantial reduction in echo spacing and an associated reduction in the time taken to traverse the k-space in the phase-encoding direction [4]. Thus, this multishot sequence reduces the susceptibility ⁎ Corresponding author. Department of Radiology, Yeouido St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, 10 63-ro Yeongdeunpo-gu, Seoul 150–713, Republic of Korea. Tel.: +82-2-3779-2017; fax: +82-2-783-5288. E-mail address: [email protected] (J.Y. Kim). http://dx.doi.org/10.1016/j.clinimag.2015.06.001 0899-7071/© 2015 Elsevier Inc. All rights reserved.
artifact caused by T2* decay during readout. The readout-segmented sampling scheme has been known for several years, but recent technical improvements have integrated the method with two-dimensional navigator correction, to allow robust correction of shot-to-shot phase errors that occur in DWI [4–6]. A number of studies have assessed the image quality of rs-EPI in the brain or in comparison with ss-EPI of 1.5-T magnetic resonance (MR) [7,8], but there are few reports concerning clinical applications of rs-EPI in acute infarctions. The purpose of the present study is to evaluate the clinical usefulness of rs-EPI in the diagnosis of acute infarction of the brainstem and posterior fossa, by comparing its results with those of ss-EPI at 3-T magnetic resonance imaging (MRI). 2. Materials and methods 2.1. Patients This retrospective study was approved by our hospital institutional review board, and informed consent was obtained from all subjects. From December 2012 to December 2013, 902 patients underwent diffusion MRI for the diagnosis of acute infarction. Of these patients, 447 patients with vague symptoms, alert mentality, and stable vital
766
J. Byeon et al. / Clinical Imaging 39 (2015) 765–769
sign underwent both ss-EPI and rs-EPI sequences. We retrospectively reviewed radiologic and clinical reports, and we identified 83 patients who were diagnosed with acute infarction of the brainstem and posterior fossa. Among these, 42 patients underwent both ss-EPI and rs-EPI sequences and 13 patients were excluded due to coexistence of chronic infarction or hemorrhage and severe motion artifacts. Finally, 29 patients (10 women and 19 men) were enrolled in the study. The mean age of the patients was 65.5 years (range, 32–89 years) (Table 1). 2.2. Image acquisition Imaging was performed with a 3.0-T MAGNETOM Skyra MR system (Siemens Healthcare, Erlangen, Germany). MR images were acquired using the following sequences: (1) ss-EPI DWI sequence; b values, 0 and 1000 s/mm 2; repetition time (TR)/echo time (TE), 6800/72 ms; flap angle, 80; field of view (FOV), 220×220 mm; matrix size, 158×158; slice thickness, 4 mm; and acquisition time, 50 s; (2) rs-EPI DWI sequence; b values, 0 and 1000 s/mm2; repetition time (TR)/echo time (TE), 8400/61 ms; field of view (FOV), 230×230 mm; matrix size, 154×154; slice thickness, 4 mm; and acquisition time, 4 min 21 s; 7 readout segments; and (3) apparent diffusion coefficient (ADC) maps that were calculated automatically using MRI software from the diffusion-weighted (DW) images.
Table 2 Criteria for qualitative comparison of image parameters in DWI with acute infarction Distinction of anatomical structure 1–5: Number of distinguishable structures (cerebral gray-white matter, basal ganglia, cerebellar folia and deep white matter, substantia nigra-red nucleus) Susceptibility artifact 1: Severe — obvious parenchymal distortion of brainstem and posterior fossa 2: Moderate — slight distortion with arc or linear shaped hyperintensity in the brain parenchyma 3: Mild — arc or linear shaped hyperintensity in the peripheral portion of brain parenchyma 4: None Lesion conspicuity 1: Nondiagnostic 2: Poor 3: Acceptable 4: Detectable with good quality Presence of uncertain high signal intensity 1: Multiple 2: Several 3: A few 4: None Overall image quality 1: Poor 2: Fair 3: Good 4: Excellent
2.3. Qualitative comparisons of image quality Two radiologists (JY Kim and JH Byun, with 11 years of experience in interpreting neuroimages and second-year radiology resident, respectively) independently reviewed and visually assessed the two sets of DW images and ADC maps for the assessment of DW image quality. The two readers were blinded to patient history, final diagnosis, and type of image sequence. The intervals between reviews of two sets of sequences were at least 2 weeks long. The readers scored images based on five criteria: distinction of anatomical structures, susceptibility artifact, conspicuity of acute infarction, presence of uncertain high signal intensity in the brain parenchyma, and overall image quality (Table 2). Table 1 Dermography of patients No.
Sex
Age (years)
Symptom
Location of infarction
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
F M F F F M F M M M M M F M M F M M F
78 77 82 66 77 83 74 70 89 32 68 45 67 64 72 80 48 40 34
Pons Pons, left cerebellum Pons Pons Pons, right cerebellum Pons Pons Pons Pons, right cerebellum Pons Pons Pons, right cerebellum Midbrain, pons Right lateral medulla Left lateral medulla Right lateral medulla Right lateral medulla Light lateral medulla Right cerebellum
20 21 22 23 24 25 26 27 28 29
M M M M M M M M F F
71 56 47 56 70 62 77 70 67 77
Left-sided weakness Headache Left-sided weakness Right-sided weakness Left arm weakness Left extremity numbness Right-sided weakness Dysarthria Right-sided weakness Dizziness Dysarthria, left-sided weakness Dizziness Stuporous mentality Left side hypesthesia Dizziness Right veering tendency Vertigo Right hypesthesia Dizziness Dysarthria Dysarthria Lower extremity astrexia Dizziness Dizziness, vomiting Right-sided weakness Left facial palsy Headache Dysarthria Dizziness Mental change
Left cerebellum Left cerebellum Right cerebellum Pons, left cerebellum Right cerebellum Left cerebellum Right cerebellum Left cerebellum Left cerebellum Both cerebellum
2.4. Quantitative comparisons of image quality For the assessment of image parameters, regions of interest (ROIs) were manually drawn on the same areas of both DW images in lesions and in contralateral normal brain parenchyma. Lesion ROIs were drawn in the centers of acute infarctions with high signal intensity in DW images and then copied to the ADC map. Circular ROIs with a minimum size of 10 mm2 were obtained (Fig. 1). If the lesion was not measurable in DW images, the case was excluded from quantitative analysis. ROIs were drawn manually in the thalamus to compare the absolute ADC values of normal brain tissue between DW images. Mean signal intensity and standard deviation were determined for each ROI. Signal-to-noise ratio (SNR) was calculated using the ratio between the mean signal intensities of infarction area ROIs (SILesion) and the standard deviation of the background noise (SDBackground) (SNR=SILesion/SDBackground). Contrast was calculated using the ratio between the mean signal intensity of the lesion (SILesion) and that of the normal brain parenchyma (SIContralateral Brain) (Contrast=SILesion/SIContralateral Brain). Contrast-to-noise ratio (CNR) was also calculated using the difference between SILesion and SIContralateral Brain divided by the standard deviation in the lesion ROI (SDLesion) and normal contralateral brain parenchyma ROI (SDContralateral Brain), using the following equation ｜SI Lesion −SI Contrlateral Brain ｜ CNR ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi SDLesionÇþ SDContralateral BrainÇ
2.5. Statistical analysis All statistical analyses were performed with Medcalc Version 13.3.1.0 (MedcalcSoftware, Mariakerke, Belgium). The interobserver agreement between the two radiologists was assessed using a linearweighted kappa test with a calculation of 95% confidence interval (CI). Kappa values greater than 0.75, from 0.40 to 0.75, and less than 0.4 were regarded as excellent, fair to good, and poor, respectively. The Wilcoxon signed rank test was used to determine significant differences between the qualitative scales of two DW images, and the paired t test was used in calculations of quantitative average scores. A P value less than b .05 was considered to indicate a statistically significant difference.
J. Byeon et al. / Clinical Imaging 39 (2015) 765–769
767
Fig. 1. Representative images of susceptibility artifact. Images in the upper row (a and b) represent ss-EPI, and images in the lower row (c and d) represent rs-EPI. The ss-EPI images demonstrate increased prominence of high signal intensity artifact at the interface of the frontal lobe and orbit and of the temporal lobe and petrous bone (arrowheads). Obvious parenchymal distortion is seen on the anterior surface of the pons (arrow).
3. Results A total of 35 acute infarction lesions were identified in 29 patients. Six patients had two infarction lesions and all others had one lesion each. The locations of acute infarctions were 1 in the midbrain (2.9%), 14 in the pons (40.0%), 15 in the cerebellum (42.9%), and 5 in the medulla (14.3%). 3.1. Qualitative comparisons of image quality The interobserver agreement between radiologists was excellent for susceptibility artifact (k=0.794) and lesion conspicuity (k=0.867) and was good for distinction of anatomic structure (k=0.643), uncertain high signal intensity foci (k=0.725), and overall image quality (k= 0.64). Overall, the scores for both readers were in excellent agreement (k=0.79, 95% CI 0.72–0.85). The average scores given by the two readers for distinction of anatomic structures were 4.0 for ss-EPI and 4.7 for rs-EPI. For susceptibility artifact, the mean scores were 3.2 (mid) for ss-EPI and 3.9 (none) for rs-EPI. For conspicuity of acute infarction, the mean scores were 3.5 (acceptable to detectable with good quality) for ss-EPI and 3.6 for rs-EPI (Fig. 2). For uncertain high signal intensity, the mean scores were 3.4 (a few to none) for ss-EPI and 3.8 for rs-EPI. For overall image quality, the mean scores were 3.0 (good) for ssEPI and 3.8 (excellent) for rs-EPI. rs-EPI was superior to ss-EPI for most criteria regarding qualitative comparisons. Distinction of anatomic structure (P=.001), reducing susceptibility artifact (P=.001), elimination of uncertain high signal foci (Pb.001), and overall image quality
(Pb .001) were significantly better in rs-EPI (Fig. 1). However, there was no significant difference between rs-EPI and ss-EPI in the conspicuity of acute infarction (P=.195) (Fig. 2 and Table 3). 3.2. Quantitative comparisons of image quality Twelve lesions in 10 patients were excluded from quantitative analysis because they were not measurable. Image parameters for ss-EPI and rs-EPI are listed in Table 4. SNR, contrast, and CNR were 235.7, 2.0, and 6.6 for ss-EPI, respectively, and 223.1, 1.8, and 5.7 for rs-EPI, respectively. There were no significant differences between ss-EPI and rs-EPI in SNR, contrast, CNR, or ADC value of lesions (P=.41, .071, and .616, respectively). There were also no significant differences in ROI size or ADC of normal thalamus between ss-EPI and rs-EPI (P=.582 and .074, respectively) (Table 4). 4. Discussion The quality of DWI can be significantly improved by using rs-EPI sequences. However, there are no differences in conspicuity of acute infarctions of the brainstem and posterior fossa between ss-EPI and rsEPI. rs-EPI sequence doses typically increase acquisition time and did so in our study (~5 min) due to the application of 7 readout segments. Acquisition time is an important factor in stroke patient evaluation. If there is no difference in the conspicuity of acute infarctions between two sequences, the rs-EPI sequence is not useful for highly suspicious stroke patients because of the long scan time. However, some patients
768
J. Byeon et al. / Clinical Imaging 39 (2015) 765–769
Fig. 2. Representative images for the conspicuity of acute infarctions. Images in the upper row (a and b) represent ss-EPI, and images in the lower row (c and d) represent rs-EPI. There are no differences in conspicuity of acute infarction at the right lateral medulla (arrow). However, rs-EPI shows improved image quality due to the decreased susceptibility artifacts overlying the medulla.
with vague symptoms such as headache, dizziness, or vertigo require high-resolution diffusion images to exclude the possibility of acute infarction and to identify other diseases. rs-EPI scan time is longer than that of ss-EPI by an amount that is approximately proportional to the number of readout segments used [7–9]. In practice, rs-EPI images with reasonable SNR can be obtained within a clinically acceptable scan time by the modulation of image parameters, including readout segments. Therefore, the difference in scan time can be much smaller than in the present study. In this study, we observed that uncertain high signal foci on DW images are significantly decreased in rs-EPI sequences (Fig. 3). Uncertain high signal foci are subtle high signal intensity lesions in brain parenchyma on DW images without significantly decreased ADC value. These lesions may confound the diagnosis of acute infarctions with vague symptoms. rs-EPI also exhibited significant improvements of image quality over ss-EPI due to decreased susceptibility artifact, particularly on the midbrain and pons [2,10]. Therefore, rs-EPI sequences are useful and exhibit high specificity to diagnose acute infarctions in patients with nonspecific symptoms even though the rs-EPI scan time is longer than that of ss-EPI.
Table 3 Comparison of qualitative scores between ss-EPI and rs-EPI in patients with acute infarction Parameter
ss-EPI
rs-EPI
P
kappa
Anatomic structure distinction Susceptibility artifact Lesion conspicuity Uncertain high signal intensity Overall image quality
3.95±0.9 3.22±0.6 3.50±0.7 3.36±0.6 2.97±0.5
4.74±0.5 3.93±0.2 3.62±0.6 3.79±0.4 3.79±0.4
.0001 .0001 .1953 .0001 b.0001
0.643 0.794 0.867 0.725 0.641
Note: Numbers show mean value±standard deviation except for P and kappa value.
In quantitative comparisons, there were no differences in SNR or contrast-noise ratio (CNR) between rs-EPI and ss-EPI. Our results are consistent with those of previous studies [2]. However, Morelli et al. [11] compared ss-EPI and rs-EPI in brain MRI and found that the SNR was lower in rs-EPI than in ss-EPI. In a study of breast cancer patients, no difference was found in CNR between the two types of DW images [12], but another study showed that rs-EPI was superior to ss-EPI in SNR and CNR [13]. In rs-EPI, the shorter readout time results in shorter effective echo time. At first glance, this seems to be disadvantageous because noise scales inversely with the square root of readout time. However, the shorter echo time results in increased signal level and therefore SNR, as long as all other imaging parameters remain fixed. The degree of signal increase will depend on the TE and T2 value of the tissue being measured. In this setting, the shorter TE of rsEPI resulted in reduced T2 contrast on DW images, which effectively obscured the borders between lesions and normal tissue. While increasing the echo time of rs-EPI may increase T2 contrast, this change is likely to come at a cost to SNR [14,15]. Therefore, it is difficult to conclude whether SNR and CNR are higher for ss-EPI or rs-EPI because many
Table 4 Comparison of quantitative assessment between ss-EPI and rs-EPI in patients with acute infarction Parameter
ss-EPI (mean±SD)
rs-EPI (mean±SD)
P
ROI size (mm2) SNR Contrast CNR ADC of lesion (×10−6 mm2/s) ADC of thalamus (×10−6 mm2/s)
10.96±3.0 235.71±87.7 2.01±0.5 6.60±5.7 514.76±87.8 854.59±80.2
11.39±1.03 223.08±59.8 1.75±0.4 5.74±3.9 511.94±115.4 822.88±80.3
.582 .410 .071 .616 .896 .074
Note: SD=standard deviation.
J. Byeon et al. / Clinical Imaging 39 (2015) 765–769
769
We demonstrated that rs-EPI showed superior image quality compared to ss-EPI in distinction of anatomic structures, reducing susceptibility artifact, eliminating uncertain high signal foci, and overall image quality. rs-EPI has benefits for pathological evaluation of the brainstem and posterior fossa. In conclusion, no differences in the conspicuity of acute infarctions in the brainstem and posterior fossa are apparent between ss-EPI and rsEPI. However, rs-EPI is useful for producing high-quality DW images of the brainstem and posterior fossa. References
Fig. 3. Representative images of uncertain high signal intensity. Axial ss-EPI (a) and rs-EPI (b). A tiny area of high signal intensity observed at the left paracentral portion of the pons on ss-EPI (arrowhead) is not present on rs-EPI. A small acute infarction also observed at the anterior portion of the left paracentral portion of the pons on ss-EPI and rs-EPI (arrow).
factors affect SNR and CNR and results will depend on the exact protocol being used. Our study has a number of limitations. First, this study was retrospective. Therefore, some acquisition parameters were not matched for the two DW sequences. Second, we calculated SNR and CNR using small areas of ROIs that were manually drawn and there was the possibility of sampling errors. Third, our study subjects included only patients with acute infarctions of the brainstem and posterior fossa. Further large-scale studies are required to examine the diagnostic utility of rs-EPI DW images.
[1] Schaefer PW. Applications of DWI in clinical neurology. J Neurol Sci 2001;186: S25–35. [2] Yeom KW, Holdsworth SJ, Van AT, Iv M, Skare S, Lober RM, et al. Comparison of readout-segmented echo-planar imaging (EPI) and single-shot EPI in clinical application of diffusion-weighted imaging of the pediatric brain. AJR Am J Roentgenol 2013;200:W437–43. [3] Farzaneh F, Riederer SJ, Pelc NJ. Analysis of T2 limitations and off-resonance effects on spatial resolution and artifacts in echo-planar imaging. Magn Reson Med 1990;14:123–39. [4] Porter DA, Heidemann RM. High resolution diffusion-weighted imaging using readout-segmented echo-planar imaging, parallel imaging and a two-dimensional navigator-based reacquisition. Magn Reson Med 2009;62:468–75. [5] Robson MD, Anderson AW, Gore JC. Diffusion-weighted multiple shot echo planar imaging of humans without navigation. Magn Reson Med 1997;38:82–8. [6] Miller KL, Pauly JM. Nonlinear phase correction for navigated diffusion imaging. Magn Reson Med 2003;50:343–53. [7] Holdsworth SJ, Skare S, Newbould RD, Guzmann R, Blevins NH, Bammer R. Readoutsegmented EPI for rapid high resolution diffusion imaging at 3 T. Eur J Radiol 2008; 65:36–46. [8] Fries P, Runge VM, Kirchin MA, Stemmer A, Naul LG, Wiliams KD, et al. Diffusionweighted imaging in patients with acute brain ischemia at 3 T: current possibilities and future perspectives comparing conventional echo-planar diffusion-weighted imaging and fast spin echo diffusion-weighted imaging sequences using BLADE (PROPELLER). Invest Radiol 2009;44:351–9. [9] Attenberger UI, Runge VM, Stemmer A, Williams KD, Naul LG, Michaely HJ, et al. Diffusion weighted imaging: a comprehensive evaluation of a fast spin echo DWI sequence with BLADE (PROPELLER) k-space sampling at 3 T, using a 32-channel head coil in acute brain ischemia. Invest Radiol 2009;44:656–61. [10] Holdsworth SJ, Yeom K, Skare S, Gentles AJ, Barnes PD, Bammer R. Clinical application of readout-segmented echo-planar imaging for diffusion-weighted imaging in pediatric brain. AJNR Am J Neuroradiol 2011;32:1274–9. [11] Morelli J, Porter D, Ai F, Gerdes C, Saettele M, Feiweier T, et al. Clinical evaluation of single-shot and readout-segmented diffusion-weighted imaging in stroke patients at 3 T. Acta Radiol 2013;54:299–306. [12] Bogner W, Pinker-Domenig K, Bickel H, Chmelik M, Weber M, Helbich TH, et al. Readout-segmented echo-planar imaging improves the diagnostic performance of diffusion-weighted MR breast examinations at 3.0 T. Radiology 2012;263:64–76. [13] Kim YJ, Kim SH, Kang BJ, Park CS, Kim HS, Son YH, et al. Readout-segmented echo-planar imaging in diffusion-weighted mr imaging in breast cancer: comparison with singleshot echo-planar imaging in image quality. Korean J Radiol 2014;15:403–10. [14] Holdsworth SJ, Skare S, Newbould RD, Bammer R. Robust GRAPPA-accelerated diffusion-weighted readout-segmented (RS)-EPI. Magn Reson Med 2009;62: 1629–40. [15] Morelli JN, Runge VM, Feiweier T, Kirsch JE, Williams KW, Attenberger UI. Evaluation of a modified Stejskal-Tanner diffusion encoding scheme, permitting a marked reduction in TE, in diffusion-weighted imaging of stroke patients at 3 T. Invest Radiol 2010;45:29–35.
Contents lists available at ScienceDirect
Clinical Imaging journal homepage: http://www.clinicalimaging.org
Original Articles
Readout-segmented echo-planar imaging in diffusion-weighted MR imaging of acute infarction of the brainstem and posterior fossa: comparison of single-shot echo-planar diffusion-weighted sequences Jonghyun Byeon a, Jee Young Kim b,⁎, A-Hyun Cho c a b c
Department of Radiology, St. Vincent’s Hospital, College of Medicine, The Catholic University of Korea, Suwon, Korea Department of Radiology, Yeouido St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul, Korea Department of Neurology, Yeouido St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul, Korea
a r t i c l e
i n f o
Article history: Received 13 February 2015 Received in revised form 2 May 2015 Accepted 1 June 2015 Keywords: Infarction Diffusion-weighted MRI Magnetic resonance imaging Bran stem
a b s t r a c t Objective: The purpose of this study was to evaluate the clinical usefulness of readout-segmented echo-planar imaging (rs-EPI) in the diagnosis of acute infarction of the brainstem and posterior fossa, by comparing its results with those of single-shot echo-planar imaging (ss-EPI) at 3-T magnetic resonance imaging. Materials and methods: Twenty-nine patients with acute infarctions of the brainstem and posterior fossa underwent both ss-EPI and rs-EPI. Two readers independently assessed two sets of diffusion-weighted (DW) images for the qualitative comparison of image quality. Signal-to-noise ratio (SNR), lesion contrast, and contrast-tonoise ratio (CNR) were calculated for the assessment of image parameters. Results: There were no significant differences in the conspicuity of acute infarction upon qualitative comparison; however, distinctions of anatomical structures, susceptibility artifact, the presence of uncertain high signal intensity in the brain parenchyma, and overall image quality were significantly better in rs-EPI DW images. There were no significant differences in SNR, lesion contrast, CNR, and apparent diffusion coefficient values of acute infarction and normal thalamus between rs-EPI and ss-EPI. Conclusion: rs-EPI DWI is a clinically useful technique for evaluating lesions in the brainstem and posterior fossa by producing high-resolution DW images with reduced susceptibility artifact. However, there are no significant differences in the conspicuity of acute infarctions in the brainstem and posterior fossa between rs-EPI and ss-EPI. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Diffusion-weighted imaging (DWI) is well established as a valuable technique for the diagnosis of acute infarction that provides high sensitivity and specificity [1]. Single-shot echo-planar imaging (ss-EPI) DWI is the current clinical standard due to its resistance to phase errors resulting from patient motion. However, susceptibility artifacts that manifest as geometric distortion, signal drop, and image blurring can be severe, particularly at 3 T, due to the long echo-planar imaging readout and corresponding low bandwidth per pixel in the phase-encoding direction [2]. These artifacts can be diminished by accelerating the kspace traversal along the phase-encoding direction [3]. Multishot or readout-segmented echo-planar imaging (rs-EPI) is used to sample a subset of k-space points in the readout direction at each shot, which allows a substantial reduction in echo spacing and an associated reduction in the time taken to traverse the k-space in the phase-encoding direction [4]. Thus, this multishot sequence reduces the susceptibility ⁎ Corresponding author. Department of Radiology, Yeouido St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, 10 63-ro Yeongdeunpo-gu, Seoul 150–713, Republic of Korea. Tel.: +82-2-3779-2017; fax: +82-2-783-5288. E-mail address: [email protected] (J.Y. Kim). http://dx.doi.org/10.1016/j.clinimag.2015.06.001 0899-7071/© 2015 Elsevier Inc. All rights reserved.
artifact caused by T2* decay during readout. The readout-segmented sampling scheme has been known for several years, but recent technical improvements have integrated the method with two-dimensional navigator correction, to allow robust correction of shot-to-shot phase errors that occur in DWI [4–6]. A number of studies have assessed the image quality of rs-EPI in the brain or in comparison with ss-EPI of 1.5-T magnetic resonance (MR) [7,8], but there are few reports concerning clinical applications of rs-EPI in acute infarctions. The purpose of the present study is to evaluate the clinical usefulness of rs-EPI in the diagnosis of acute infarction of the brainstem and posterior fossa, by comparing its results with those of ss-EPI at 3-T magnetic resonance imaging (MRI). 2. Materials and methods 2.1. Patients This retrospective study was approved by our hospital institutional review board, and informed consent was obtained from all subjects. From December 2012 to December 2013, 902 patients underwent diffusion MRI for the diagnosis of acute infarction. Of these patients, 447 patients with vague symptoms, alert mentality, and stable vital
766
J. Byeon et al. / Clinical Imaging 39 (2015) 765–769
sign underwent both ss-EPI and rs-EPI sequences. We retrospectively reviewed radiologic and clinical reports, and we identified 83 patients who were diagnosed with acute infarction of the brainstem and posterior fossa. Among these, 42 patients underwent both ss-EPI and rs-EPI sequences and 13 patients were excluded due to coexistence of chronic infarction or hemorrhage and severe motion artifacts. Finally, 29 patients (10 women and 19 men) were enrolled in the study. The mean age of the patients was 65.5 years (range, 32–89 years) (Table 1). 2.2. Image acquisition Imaging was performed with a 3.0-T MAGNETOM Skyra MR system (Siemens Healthcare, Erlangen, Germany). MR images were acquired using the following sequences: (1) ss-EPI DWI sequence; b values, 0 and 1000 s/mm 2; repetition time (TR)/echo time (TE), 6800/72 ms; flap angle, 80; field of view (FOV), 220×220 mm; matrix size, 158×158; slice thickness, 4 mm; and acquisition time, 50 s; (2) rs-EPI DWI sequence; b values, 0 and 1000 s/mm2; repetition time (TR)/echo time (TE), 8400/61 ms; field of view (FOV), 230×230 mm; matrix size, 154×154; slice thickness, 4 mm; and acquisition time, 4 min 21 s; 7 readout segments; and (3) apparent diffusion coefficient (ADC) maps that were calculated automatically using MRI software from the diffusion-weighted (DW) images.
Table 2 Criteria for qualitative comparison of image parameters in DWI with acute infarction Distinction of anatomical structure 1–5: Number of distinguishable structures (cerebral gray-white matter, basal ganglia, cerebellar folia and deep white matter, substantia nigra-red nucleus) Susceptibility artifact 1: Severe — obvious parenchymal distortion of brainstem and posterior fossa 2: Moderate — slight distortion with arc or linear shaped hyperintensity in the brain parenchyma 3: Mild — arc or linear shaped hyperintensity in the peripheral portion of brain parenchyma 4: None Lesion conspicuity 1: Nondiagnostic 2: Poor 3: Acceptable 4: Detectable with good quality Presence of uncertain high signal intensity 1: Multiple 2: Several 3: A few 4: None Overall image quality 1: Poor 2: Fair 3: Good 4: Excellent
2.3. Qualitative comparisons of image quality Two radiologists (JY Kim and JH Byun, with 11 years of experience in interpreting neuroimages and second-year radiology resident, respectively) independently reviewed and visually assessed the two sets of DW images and ADC maps for the assessment of DW image quality. The two readers were blinded to patient history, final diagnosis, and type of image sequence. The intervals between reviews of two sets of sequences were at least 2 weeks long. The readers scored images based on five criteria: distinction of anatomical structures, susceptibility artifact, conspicuity of acute infarction, presence of uncertain high signal intensity in the brain parenchyma, and overall image quality (Table 2). Table 1 Dermography of patients No.
Sex
Age (years)
Symptom
Location of infarction
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
F M F F F M F M M M M M F M M F M M F
78 77 82 66 77 83 74 70 89 32 68 45 67 64 72 80 48 40 34
Pons Pons, left cerebellum Pons Pons Pons, right cerebellum Pons Pons Pons Pons, right cerebellum Pons Pons Pons, right cerebellum Midbrain, pons Right lateral medulla Left lateral medulla Right lateral medulla Right lateral medulla Light lateral medulla Right cerebellum
20 21 22 23 24 25 26 27 28 29
M M M M M M M M F F
71 56 47 56 70 62 77 70 67 77
Left-sided weakness Headache Left-sided weakness Right-sided weakness Left arm weakness Left extremity numbness Right-sided weakness Dysarthria Right-sided weakness Dizziness Dysarthria, left-sided weakness Dizziness Stuporous mentality Left side hypesthesia Dizziness Right veering tendency Vertigo Right hypesthesia Dizziness Dysarthria Dysarthria Lower extremity astrexia Dizziness Dizziness, vomiting Right-sided weakness Left facial palsy Headache Dysarthria Dizziness Mental change
Left cerebellum Left cerebellum Right cerebellum Pons, left cerebellum Right cerebellum Left cerebellum Right cerebellum Left cerebellum Left cerebellum Both cerebellum
2.4. Quantitative comparisons of image quality For the assessment of image parameters, regions of interest (ROIs) were manually drawn on the same areas of both DW images in lesions and in contralateral normal brain parenchyma. Lesion ROIs were drawn in the centers of acute infarctions with high signal intensity in DW images and then copied to the ADC map. Circular ROIs with a minimum size of 10 mm2 were obtained (Fig. 1). If the lesion was not measurable in DW images, the case was excluded from quantitative analysis. ROIs were drawn manually in the thalamus to compare the absolute ADC values of normal brain tissue between DW images. Mean signal intensity and standard deviation were determined for each ROI. Signal-to-noise ratio (SNR) was calculated using the ratio between the mean signal intensities of infarction area ROIs (SILesion) and the standard deviation of the background noise (SDBackground) (SNR=SILesion/SDBackground). Contrast was calculated using the ratio between the mean signal intensity of the lesion (SILesion) and that of the normal brain parenchyma (SIContralateral Brain) (Contrast=SILesion/SIContralateral Brain). Contrast-to-noise ratio (CNR) was also calculated using the difference between SILesion and SIContralateral Brain divided by the standard deviation in the lesion ROI (SDLesion) and normal contralateral brain parenchyma ROI (SDContralateral Brain), using the following equation ｜SI Lesion −SI Contrlateral Brain ｜ CNR ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi SDLesionÇþ SDContralateral BrainÇ
2.5. Statistical analysis All statistical analyses were performed with Medcalc Version 13.3.1.0 (MedcalcSoftware, Mariakerke, Belgium). The interobserver agreement between the two radiologists was assessed using a linearweighted kappa test with a calculation of 95% confidence interval (CI). Kappa values greater than 0.75, from 0.40 to 0.75, and less than 0.4 were regarded as excellent, fair to good, and poor, respectively. The Wilcoxon signed rank test was used to determine significant differences between the qualitative scales of two DW images, and the paired t test was used in calculations of quantitative average scores. A P value less than b .05 was considered to indicate a statistically significant difference.
J. Byeon et al. / Clinical Imaging 39 (2015) 765–769
767
Fig. 1. Representative images of susceptibility artifact. Images in the upper row (a and b) represent ss-EPI, and images in the lower row (c and d) represent rs-EPI. The ss-EPI images demonstrate increased prominence of high signal intensity artifact at the interface of the frontal lobe and orbit and of the temporal lobe and petrous bone (arrowheads). Obvious parenchymal distortion is seen on the anterior surface of the pons (arrow).
3. Results A total of 35 acute infarction lesions were identified in 29 patients. Six patients had two infarction lesions and all others had one lesion each. The locations of acute infarctions were 1 in the midbrain (2.9%), 14 in the pons (40.0%), 15 in the cerebellum (42.9%), and 5 in the medulla (14.3%). 3.1. Qualitative comparisons of image quality The interobserver agreement between radiologists was excellent for susceptibility artifact (k=0.794) and lesion conspicuity (k=0.867) and was good for distinction of anatomic structure (k=0.643), uncertain high signal intensity foci (k=0.725), and overall image quality (k= 0.64). Overall, the scores for both readers were in excellent agreement (k=0.79, 95% CI 0.72–0.85). The average scores given by the two readers for distinction of anatomic structures were 4.0 for ss-EPI and 4.7 for rs-EPI. For susceptibility artifact, the mean scores were 3.2 (mid) for ss-EPI and 3.9 (none) for rs-EPI. For conspicuity of acute infarction, the mean scores were 3.5 (acceptable to detectable with good quality) for ss-EPI and 3.6 for rs-EPI (Fig. 2). For uncertain high signal intensity, the mean scores were 3.4 (a few to none) for ss-EPI and 3.8 for rs-EPI. For overall image quality, the mean scores were 3.0 (good) for ssEPI and 3.8 (excellent) for rs-EPI. rs-EPI was superior to ss-EPI for most criteria regarding qualitative comparisons. Distinction of anatomic structure (P=.001), reducing susceptibility artifact (P=.001), elimination of uncertain high signal foci (Pb.001), and overall image quality
(Pb .001) were significantly better in rs-EPI (Fig. 1). However, there was no significant difference between rs-EPI and ss-EPI in the conspicuity of acute infarction (P=.195) (Fig. 2 and Table 3). 3.2. Quantitative comparisons of image quality Twelve lesions in 10 patients were excluded from quantitative analysis because they were not measurable. Image parameters for ss-EPI and rs-EPI are listed in Table 4. SNR, contrast, and CNR were 235.7, 2.0, and 6.6 for ss-EPI, respectively, and 223.1, 1.8, and 5.7 for rs-EPI, respectively. There were no significant differences between ss-EPI and rs-EPI in SNR, contrast, CNR, or ADC value of lesions (P=.41, .071, and .616, respectively). There were also no significant differences in ROI size or ADC of normal thalamus between ss-EPI and rs-EPI (P=.582 and .074, respectively) (Table 4). 4. Discussion The quality of DWI can be significantly improved by using rs-EPI sequences. However, there are no differences in conspicuity of acute infarctions of the brainstem and posterior fossa between ss-EPI and rsEPI. rs-EPI sequence doses typically increase acquisition time and did so in our study (~5 min) due to the application of 7 readout segments. Acquisition time is an important factor in stroke patient evaluation. If there is no difference in the conspicuity of acute infarctions between two sequences, the rs-EPI sequence is not useful for highly suspicious stroke patients because of the long scan time. However, some patients
768
J. Byeon et al. / Clinical Imaging 39 (2015) 765–769
Fig. 2. Representative images for the conspicuity of acute infarctions. Images in the upper row (a and b) represent ss-EPI, and images in the lower row (c and d) represent rs-EPI. There are no differences in conspicuity of acute infarction at the right lateral medulla (arrow). However, rs-EPI shows improved image quality due to the decreased susceptibility artifacts overlying the medulla.
with vague symptoms such as headache, dizziness, or vertigo require high-resolution diffusion images to exclude the possibility of acute infarction and to identify other diseases. rs-EPI scan time is longer than that of ss-EPI by an amount that is approximately proportional to the number of readout segments used [7–9]. In practice, rs-EPI images with reasonable SNR can be obtained within a clinically acceptable scan time by the modulation of image parameters, including readout segments. Therefore, the difference in scan time can be much smaller than in the present study. In this study, we observed that uncertain high signal foci on DW images are significantly decreased in rs-EPI sequences (Fig. 3). Uncertain high signal foci are subtle high signal intensity lesions in brain parenchyma on DW images without significantly decreased ADC value. These lesions may confound the diagnosis of acute infarctions with vague symptoms. rs-EPI also exhibited significant improvements of image quality over ss-EPI due to decreased susceptibility artifact, particularly on the midbrain and pons [2,10]. Therefore, rs-EPI sequences are useful and exhibit high specificity to diagnose acute infarctions in patients with nonspecific symptoms even though the rs-EPI scan time is longer than that of ss-EPI.
Table 3 Comparison of qualitative scores between ss-EPI and rs-EPI in patients with acute infarction Parameter
ss-EPI
rs-EPI
P
kappa
Anatomic structure distinction Susceptibility artifact Lesion conspicuity Uncertain high signal intensity Overall image quality
3.95±0.9 3.22±0.6 3.50±0.7 3.36±0.6 2.97±0.5
4.74±0.5 3.93±0.2 3.62±0.6 3.79±0.4 3.79±0.4
.0001 .0001 .1953 .0001 b.0001
0.643 0.794 0.867 0.725 0.641
Note: Numbers show mean value±standard deviation except for P and kappa value.
In quantitative comparisons, there were no differences in SNR or contrast-noise ratio (CNR) between rs-EPI and ss-EPI. Our results are consistent with those of previous studies [2]. However, Morelli et al. [11] compared ss-EPI and rs-EPI in brain MRI and found that the SNR was lower in rs-EPI than in ss-EPI. In a study of breast cancer patients, no difference was found in CNR between the two types of DW images [12], but another study showed that rs-EPI was superior to ss-EPI in SNR and CNR [13]. In rs-EPI, the shorter readout time results in shorter effective echo time. At first glance, this seems to be disadvantageous because noise scales inversely with the square root of readout time. However, the shorter echo time results in increased signal level and therefore SNR, as long as all other imaging parameters remain fixed. The degree of signal increase will depend on the TE and T2 value of the tissue being measured. In this setting, the shorter TE of rsEPI resulted in reduced T2 contrast on DW images, which effectively obscured the borders between lesions and normal tissue. While increasing the echo time of rs-EPI may increase T2 contrast, this change is likely to come at a cost to SNR [14,15]. Therefore, it is difficult to conclude whether SNR and CNR are higher for ss-EPI or rs-EPI because many
Table 4 Comparison of quantitative assessment between ss-EPI and rs-EPI in patients with acute infarction Parameter
ss-EPI (mean±SD)
rs-EPI (mean±SD)
P
ROI size (mm2) SNR Contrast CNR ADC of lesion (×10−6 mm2/s) ADC of thalamus (×10−6 mm2/s)
10.96±3.0 235.71±87.7 2.01±0.5 6.60±5.7 514.76±87.8 854.59±80.2
11.39±1.03 223.08±59.8 1.75±0.4 5.74±3.9 511.94±115.4 822.88±80.3
.582 .410 .071 .616 .896 .074
Note: SD=standard deviation.
J. Byeon et al. / Clinical Imaging 39 (2015) 765–769
769
We demonstrated that rs-EPI showed superior image quality compared to ss-EPI in distinction of anatomic structures, reducing susceptibility artifact, eliminating uncertain high signal foci, and overall image quality. rs-EPI has benefits for pathological evaluation of the brainstem and posterior fossa. In conclusion, no differences in the conspicuity of acute infarctions in the brainstem and posterior fossa are apparent between ss-EPI and rsEPI. However, rs-EPI is useful for producing high-quality DW images of the brainstem and posterior fossa. References
Fig. 3. Representative images of uncertain high signal intensity. Axial ss-EPI (a) and rs-EPI (b). A tiny area of high signal intensity observed at the left paracentral portion of the pons on ss-EPI (arrowhead) is not present on rs-EPI. A small acute infarction also observed at the anterior portion of the left paracentral portion of the pons on ss-EPI and rs-EPI (arrow).
factors affect SNR and CNR and results will depend on the exact protocol being used. Our study has a number of limitations. First, this study was retrospective. Therefore, some acquisition parameters were not matched for the two DW sequences. Second, we calculated SNR and CNR using small areas of ROIs that were manually drawn and there was the possibility of sampling errors. Third, our study subjects included only patients with acute infarctions of the brainstem and posterior fossa. Further large-scale studies are required to examine the diagnostic utility of rs-EPI DW images.
[1] Schaefer PW. Applications of DWI in clinical neurology. J Neurol Sci 2001;186: S25–35. [2] Yeom KW, Holdsworth SJ, Van AT, Iv M, Skare S, Lober RM, et al. Comparison of readout-segmented echo-planar imaging (EPI) and single-shot EPI in clinical application of diffusion-weighted imaging of the pediatric brain. AJR Am J Roentgenol 2013;200:W437–43. [3] Farzaneh F, Riederer SJ, Pelc NJ. Analysis of T2 limitations and off-resonance effects on spatial resolution and artifacts in echo-planar imaging. Magn Reson Med 1990;14:123–39. [4] Porter DA, Heidemann RM. High resolution diffusion-weighted imaging using readout-segmented echo-planar imaging, parallel imaging and a two-dimensional navigator-based reacquisition. Magn Reson Med 2009;62:468–75. [5] Robson MD, Anderson AW, Gore JC. Diffusion-weighted multiple shot echo planar imaging of humans without navigation. Magn Reson Med 1997;38:82–8. [6] Miller KL, Pauly JM. Nonlinear phase correction for navigated diffusion imaging. Magn Reson Med 2003;50:343–53. [7] Holdsworth SJ, Skare S, Newbould RD, Guzmann R, Blevins NH, Bammer R. Readoutsegmented EPI for rapid high resolution diffusion imaging at 3 T. Eur J Radiol 2008; 65:36–46. [8] Fries P, Runge VM, Kirchin MA, Stemmer A, Naul LG, Wiliams KD, et al. Diffusionweighted imaging in patients with acute brain ischemia at 3 T: current possibilities and future perspectives comparing conventional echo-planar diffusion-weighted imaging and fast spin echo diffusion-weighted imaging sequences using BLADE (PROPELLER). Invest Radiol 2009;44:351–9. [9] Attenberger UI, Runge VM, Stemmer A, Williams KD, Naul LG, Michaely HJ, et al. Diffusion weighted imaging: a comprehensive evaluation of a fast spin echo DWI sequence with BLADE (PROPELLER) k-space sampling at 3 T, using a 32-channel head coil in acute brain ischemia. Invest Radiol 2009;44:656–61. [10] Holdsworth SJ, Yeom K, Skare S, Gentles AJ, Barnes PD, Bammer R. Clinical application of readout-segmented echo-planar imaging for diffusion-weighted imaging in pediatric brain. AJNR Am J Neuroradiol 2011;32:1274–9. [11] Morelli J, Porter D, Ai F, Gerdes C, Saettele M, Feiweier T, et al. Clinical evaluation of single-shot and readout-segmented diffusion-weighted imaging in stroke patients at 3 T. Acta Radiol 2013;54:299–306. [12] Bogner W, Pinker-Domenig K, Bickel H, Chmelik M, Weber M, Helbich TH, et al. Readout-segmented echo-planar imaging improves the diagnostic performance of diffusion-weighted MR breast examinations at 3.0 T. Radiology 2012;263:64–76. [13] Kim YJ, Kim SH, Kang BJ, Park CS, Kim HS, Son YH, et al. Readout-segmented echo-planar imaging in diffusion-weighted mr imaging in breast cancer: comparison with singleshot echo-planar imaging in image quality. Korean J Radiol 2014;15:403–10. [14] Holdsworth SJ, Skare S, Newbould RD, Bammer R. Robust GRAPPA-accelerated diffusion-weighted readout-segmented (RS)-EPI. Magn Reson Med 2009;62: 1629–40. [15] Morelli JN, Runge VM, Feiweier T, Kirsch JE, Williams KW, Attenberger UI. Evaluation of a modified Stejskal-Tanner diffusion encoding scheme, permitting a marked reduction in TE, in diffusion-weighted imaging of stroke patients at 3 T. Invest Radiol 2010;45:29–35.
