Clinical Investigative Study Effect of Number of Acquisitions in Diffusion Tensor Imaging of the Pediatric Brain: Optimizing Scan Time and Diagnostic Experience Salil Soman, MD, Samantha J. Holdsworth, PhD, Stefan Skare, PhD, Jalal B. Andre, MD, Anh T. Van, PhD, Murat Aksoy, PhD, Roland Bammer, PhD, Jarrett Rosenberg, PhD, Patrick D. Barnes, MD, Kristen W. Yeom, MD From the Department of Radiology, Stanford University, Stanford, CA (SS, JR, PDB, KWY); Department of Radiology, Lucas Center, Stanford University, Stanford, CA (SJH, ATV, MA, RB); Clinical Neuroscience, Karolinska Institute, Stockholm, Sweden (SS); and University of Washington, Radiology, Seattle, WA (JBA).

ABSTRACT BACKGROUND AND PURPOSE

Diffusion tensor imaging (DTI) is useful for multiple clinical applications, but its routine implementation for children may be difficult due to long scan times. This study evaluates the impact of decreasing the number of DTI acquisitions (NEX) on interpretability of pediatric brain DTI. METHODS

15 children with MRI-visible neuropathologies were imaged at 3T using our motioncorrected, parallel imaging- accelerated DT-EPI technique with 3 NEX (scan time 8.25 min). Using these acquisitions, NEX = 1 (scan time 2.75 min) and NEX = 2 (scan time 5.5 min) images were simulated. Two neuroradiologists scored diffusion-weighted images (DWI), apparent diffusion coefficient (ADC), fractional anisotropy (FA), and first eigenvector color-encoded (EV) images from each NEX for perceived SNR, lesion conspicuity and clinical confidence. ROI FA/ADC and image SNR values were also compared across NEX.

Keywords: Pediatric brain, DTI, MRI acquisition time, SNR. Acceptance: Received May 7, 2013, and in revised form September 11, 2013. Accepted for publication September 15, 2013. Correspondence: Address correspondence to Salil Soman, MD, MS Stanford Medical Center, Department of Neuroradiology, 300 Pasteur Drive, Room S047, Stanford, CA 94305-5105, E-mail: [email protected]. J Neuroimaging 2014;00:1-7. DOI: 10.1111/jon.12093

RESULTS

NEX = 2 perceived SNR, lesion conspicuity, and clinical confidence were not inferior to NEX = 3 images. NEX = 1 images showed comparable lesion conspicuity and clinical confidence as NEX = 3, but inferior perceived SNR. FA and ADC ROI measurements demonstrated no significant difference across NEX. The greatest SNR increase was seen between NEX = 1 and NEX = 2. CONCLUSION

Reducing NEX to shorten imaging time may impact clinical utility in a manner that does not directly correspond with SNR changes.

Background and Purpose Diffusion tensor imaging (DTI) has been shown useful in multiple pediatric clinical neuroimaging applications, including brain maturation, injury, tumors, and others.1–17 However, routine use of DTI may be limited by long scan times in motionprone patients, particularly young children. Long magnetic resonance imaging (MRI) scan times may also require anesthesia, which carries associated risks.18–20 Our experience scanning over 3,500 pediatric patients over the last three years using a parallel imaging-accelerated and motion-corrected Echo Planar Imaging (EPI) DTI sequence21–24 with in-scanner entertainment systems25 still required an average scan time of 8.25 minutes. Given challenges of lengthy scans in pediatric population, we explored various methods that could shorten the DTI acquisition time without reducing overall diagnostic quality.

Of the multiple factors influencing DTI acquisition time, we suspected reducing the number of acquisitions (NEX) would most likely shorten imaging time without significant impact on diagnostic capacity of DTI. The number of gradient directions (GDs) imaged and the NEX used to acquire each of the GDs are among the most prominent parameters that determine the DTI acquisition time. As DTI characterizes tissue based on the water movement under differing magnetic gradients, we suspected decreasing the number of GDs could adversely impact the specific tissue information acquired. This is supported by literature that describe a minimum of six GDs required for DTI,2–4,26 with higher number of GDs improving the accuracy of diffusion parameters, such as fractional anisotropy (FA) or apparent diffusion coefficient (ADC).27,28 Multiple NEX, however, is the acquisition of each of the GDs multiple times, with subsequent averaging of each direction to create a single data set

Copyright

◦ 2014 by the American Society of Neuroimaging C

1

Table 1. Summary of Clinical Presentation Subject

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Brain Pathology

Choroid plexus papilloma Cerebral dysgenesis and agenesis of corpus callosum Cerebral contusions and hemorrhage Gray matter heterotopia Lesions with reduced diffusion in the brainstem in a child with Pellagra Acute demyelinating encephalomyelitis (ADEM) Tectal glioma Hypoxic ischemic injury Cerebral dysgenesis Cerebellar dysplasia Medulloblastoma Diffuse intrinsic pontine glioma Cerebral abscess Retinoblastoma Brain infarction associated with sickle cell disease

of GDs with increased signal-to-noise ratio (SNR).1–3,24 Any excess SNR beyond what is required to generate clinically useful images would suggest a target for shortening acquisition time. There are mixed data with regard to the impact of NEX on DTI. Prior work has suggested that increasing the number of NEX from 1 to 5 in a 24 direction DTI data set yields a 75% improvement in fiber tracking results when compared to a gold standard.29 However, others have reported employing multiple NEX to be superfluous, without significant change in FA in healthy volunteers at NEX > 1.30,31 Additionally, there are no studies evaluating the impact of NEX on radiologist’s visual interpretation of DTI when evaluating brain pathology. We hypothesized that reducing the NEX from NEX = 3 to NEX = 2 or NEX = 1 can produce images more quickly than our NEX = 3 protocol (a 5.5-minute reduction for NEX = 1 or a 2.75-minute reduction for NEX = 2), without impairing overall diagnostic quality.

Materials and Methods Patients Fifteen consecutive patients (ages 1 day to 17 years) who obtained DTI as part of routine brain MRI for brain pathology were retrospectively evaluated after IRB approval (IRB no. 4947). All patients were imaged with an 8-channel receive head coil and a 3T MR system (DVMR750, GE Healthcare, Waukesha, WI, USA). Presenting clinical features of these patients are listed in Table 1.

DTI Protocol and Postprocessing DTI data were acquired on each subject with a custom-designed in-house GeneRalized Autocalibrating Partially Parallel Acquisitions (GRAPPA)-accelerated EPI sequence32 and NEX = 3 using the following scan parameters: acceleration factor R = 3, 20-24 cm FOV, TR/TE = 5500 ms/70 ms, slice thickness/gap = 3/0 mm, 1282 acquisition matrix (resulting in a 1.6 mm × 1.6 mm × 3 mm image resolution), 2562 reconstruction matrix, twice-refocused diffusion preparation, b = 1000 s/mm2 , 5 b = 0

2

Journal of Neuroimaging Vol 00 No 0 xxxx 2014

weighted images, and 25 isotropically distributed diffusion directions. Postprocessing was performed using in-house built MATLAB code which selects the best parallel imaging and ghost calibration weights; does 3-dimensional rigid-body realignment; and which employs phase correction and complex averaging to lower Rician noise and reduce phase artifacts. This is explained in more detail in Holdsworth et al.24 For each patient’s NEX = 3 data set, 25 different GDs were acquired successively, three times in a row, resulting in 3 sets of 25 distinct GDs, as per our routine protocol. Using these three sets of GDs (acquisition 1, 2, and 3), three separate DTI data sets were created: 1. NEX = 1: one of the three acquisition sets of 25 GDs was chosen at random to be the DTI data set. 2. NEX = 2: two of the three acquisition sets of 25 GDs were chosen at random. Each GD from the first acquisition set was averaged with the corresponding GD from the second acquisition set, resulting in a new set of 25 GDs. 3. NEX = 3: our standard clinical protocol using all three of the acquisition sets of GDs. Each GD from the first set was averaged with the corresponding GD from the second and third set, to create a set of 25 GDs.

Our usual postprocessing of DTI data was then performed, resulting in three sets (NEX = 1, 2, and 3) of isotropic diffusionweighted imaging (DWI), ADC, FA, and FA eigenvector (EV) color images corresponding to scan times of 2.75, 5.50, and 8.25 minutes, respectively. All images were then sent to our hospital image database (Picture Archiving and Communications System [PACS]) for storage and review.

Analysis Region of interest (ROI)-Based Analysis of Pathologic Brain Lesions (FA and ADC) As an objective measure, ROIs were manually drawn over the pathologic brain lesion identified on FA and ADC images. The ROIs were drawn identical in size and location; this was performed for all three data sets (NEX = 1, NEX = 2, and NEX = 3) by a blinded, board-certified first-year neuroradiology fellow (SS). Proper ROI placement was subsequently confirmed by a blinded board-certified attending neuroradiologist (KWY) with certificate of added qualification (6 years of experience). The mean value and the standard deviation (SD) of each ROI were recorded. One-way ANOVA analysis was performed separately on each of these four sets of values (mean FA, FA SD, mean ADC, and ADC SD).

Quantitative SNR Analysis The quantitative SNR was calculated for NEX = 1, 2, and 3 in each patient with FSL33,34 (FMRIB Software Library) using the five B = 0 images acquired as part of the DTI acquisition. A oneway ANOVA analysis was then performed on calculated SNR values. In addition, the percent increase in calculated SNR with increasing NEX (between NEX = 1 and NEX = 2, NEX = 2 and NEX = 3, and NEX = 1 and NEX = 3) was determined for each subject. One-way ANOVA analysis was performed on the percent increase in SNR for these three groups. The MedCalc

software package (http://www.medcalc.org) was used for all ANOVA and T-test calculations.

Neuroradiologist DTI Review and Scoring An attending neuroradiologist (KWY) and a second-year boardcertified neuroradiology fellow (JA), independently reviewed the DTI images acquired on the 15 patients. Each data set (NEX = 1, NEX = 2, and NEX = 3) was reviewed blindly, and in random order, with respect to clinical information, patient number, and NEX status. No reviewer consecutively reviewed two different NEX data sets for any one subject. During each session, the reviewers were only shown DTI-derived images. The reviewers evaluated for perceived SNR, lesion conspicuity, and diagnostic confidence in characterizing the brain lesions for each NEX data set. Scoring was performed using a modified Likert numerical scale of 1 = poor, 2 = below average, 3 = average, 4 = above average, and 5 = excellent. Noninferiority testing: Using the neuroradiologists’ scores, noninferiority of NEX = 1 and NEX = 2 images with respect to NEX = 3 images was tested with paired Wilcoxon tests against a null median of –1. This method indicated that the median difference between the NEX = 1 or NEX = 2, and the NEX = 3 data sets was no more than one rating step in favor of NEX = 3. A Bonferroni-adjusted P-value of .0021 was used as the significance level. Analysis of pooled scores across all image types for each NEX levels: One-way ANOVA analysis using the number of NEX as the independent variable was performed on the reviewers’ scores for perceived SNR, lesion conspicuity, and clinical confidence for all four image types (DWI, ADC, FA, and EV). Where statistical differences existed among NEX groups (NEX = 1, 2, 3), a paired samples t-test was performed between NEX subsets (NEX = 1 and NEX = 2, NEX = 1 and NEX = 3, and NEX = 2 and NEX = 3). Similar one-way ANOVA analysis was performed on the reviewers’ scores for each image type (DWI, ADC, FA, and EV).

Results ROI-Based Analysis (FA and ADC) ANOVA analysis using NEX as the independent variable demonstrated no significant difference between NEX groups for FA and ADC values.

Quantitative SNR Analysis ANOVA analysis of calculated SNR values, using NEX as the dependent variable, demonstrated a progressive increase in calculated SNR with increasing NEX (P < .0001; Figs 1 and 2). This increase in SNR was consistent with the theoretical 2 dependence of NEX on SNR with the slight variations likely due to brain pulsatility, as might be expected from in vivo studies. ANOVA analysis of percent increase in SNR with change in NEX showed the following: % SNR Increase (NEX = 1 vs. NEX = 3) > % SNR Increase (NEX = 1 vs. NEX = 2) > % SNR Increase (NEX = 2 vs. NEX = 3; P < .0001; Fig 2).

Radiologist Qualitative Assessment of DTI Across NEX Overall score

For all NEX levels (NEX = 1, 2, 3), the scores for perceived SNR, lesion conspicuity, and clinical confidence were graded as average or higher. FA and EV images scored lower, on average, than their corresponding DWI and ADC images (see Figs 3 and 4). Perceived SNR

The scores for NEX = 2, and NEX = 3 did not differ significantly across all four image types (DWI, ADC, FA, and EV). The scores for perceived SNR for NEX = 1, NEX = 2, and NEX = 3 did not differ significantly when evaluating DWI. However, the scores were lower for NEX = 1 as compared to NEX = 2 for EV (P = .0110), FA (P = .0434), and ADC (P = .0159). The scores were also lower for NEX = 1 as compared to NEX = 3 for EV (P = .0332), FA (P = .0260), and ADC (P = .0260; see Fig 4). Lesion conspicuity and clinical confidence

For all NEX and all imaging types (DWI, ADC, FA, and EV), the scores did not significantly differ (see Fig 3). Noninferiority testing

The scores for NEX = 2 were not inferior to NEX = 3 on all image types (P < .0004). The scores for NEX = 1 were inferior to NEX = 3 in terms of perceived SNR. However, NEX = 1 was not inferior to NEX = 3 in terms of lesion conspicuity or clinical confidence. Analysis of pooled scores across all image types for each NEX levels

ANOVA evaluation of pooled scores for each query using NEX as the independent variable revealed a significant difference between NEX = 1, NEX = 2, and NEX = 3 in terms of perceived SNR. Subsequent paired t-test results for perceived SNR showed lower scores for NEX = 1 as compared to NEX = 2 (P < .0001) and NEX = 3 (P < .0001). The scores for NEX = 2 and NEX = 3 were not significantly different in terms of perceived SNR (P = .3723). ANOVA analysis showed no significant difference among the NEX levels in terms of lesion conspicuity or clinical confidence.

Discussion While DWI/DTI has been shown useful in a variety of clinical settings, diagnostic experience of the radiologist and overall qualitative impact of differing NEX, as well as quantitative effects within brain pathology has not been previously studied. Here we showed that radiologist experience with regards to perceived SNR, lesion conspicuity, and clinical confidence were comparable for DTI data sets derived from NEX = 2 and NEX = 3. We also found no significant change in quantitative FA and ADC across the NEX levels within brain pathology, and greatest increase in calculated SNR by increasing from NEX = 1 to NEX = 2. Although metrics of image quality such as SNR or the contrast to noise ratio are approximately the square root of signal sampled, knowledge of these values may not translate into

Soman et al: Effect of DTI Repetitions on Diagnostic Experience

3

Fig 1. Computed SNR for NEX = 1, 2, and 3 shows increasing SNR with increased NEX. diagnostic quality that is considered adequate for clinical decision making.32 In clinical practice, radiologists routinely process various imaging features, such as contrast, SNR, distortion, and others, and combine with internal gauge of diagnostic confidence before formulating a differential. In order to better characterize the relationship between image quality and radiology reader performance, referred as ‘‘psychophysics,’’35 we evaluated subjective measures such as perceived SNR, lesion conspicuity, and clinical confidence, in addition to quantitative measures of SNR. In general, our results suggested that NEX = 2 DTI images were equivalent to NEX = 3 DTI images (Figs 3 and 4). While the readers scored perceived SNR for NEX = 1 lower than NEX = 2 (Fig 3), the calculated SNR for each NEX varied widely (Fig 1), likely due to different brain volumes. Thus, we could not correlate the NEX = 2 data to a specific quantitative SNR threshold. We also did not find a threshold below which neuroradiologists could no longer either detect or adequately characterize lesions. Comparison of the FA maps for NEX = 2 and NEX = 3 did not reveal a statistically significant difference between ROIderived mean FA values at the level of brain pathology, consistent with reports by Widjaja et al30 and Ni et al31 in healthy volunteers. Additionally, no significant difference in mean ADC,

4

Journal of Neuroimaging Vol 00 No 0 xxxx 2014

FA SD, or ADC SD values was noted with varying NEX. This is in contradistinction to work by Wang et al36 and Farrell et al,37 in which NEX = 1 acquisitions demonstrated significantly elevated FA measurements compared to NEX = 2 acquisitions. While we found no statistically significant difference in FA and ADC ROI values with varying NEX, we did observe an overall increase in perceived SNR with increasing NEX, as previously reported.37 However, our scores of image quality suggest that a change from NEX = 2 to NEX = 3 may not impact neuroradiologists’ experience with DTI interpretation, while a change from NEX = 1 to NEX = 2 could impact their overall diagnostic experience. This may be attributed to the twofold increase in SNR when comparing the difference in SNR between NEX = 1 and NEX = 2 compared with NEX = 2 and NEX = 3 (see Fig 2). Interestingly, there were a few instances where the scores for lesion conspicuity and clinical confidence were lower at higher NEX, although not statistically significant. In these instances, motion and pulsation artifacts may have contributed more to image degradation with additional NEX, despite our efforts to control for motion by applying the same motion correction to all data sets. Overall, our observations suggest that DTI with NEX = 2 did not adversely affect the diagnostic utility

Fig 2. (A) Distribution of computed SNR for NEX = 1, 2, 3 and (B) Percent increase in computed SNR with increasing NEX show greater

percent boost in SNR when changing from NEX = 1 to NEX = 2, than changing from NEX = 2 to NEX = 3. The largest percent SNR increase is seen changing from NEX = 1 to NEX = 3.

Fig 3. Average scores of the readers with regards to perceived SNR, lesion conspicuity, and clinical confidence in detecting and characterizing pathology across the NEX levels. While the chart shows NEX = 2 DWI were scored slightly lower than NEX = 3 for lesion conspicuity and that NEX = 1 was scored slightly higher than NEX = 2 or NEX = 3 for clinical confidence, these differences were not found to be statistically significant.

of the derived images. Additionally, while NEX = 1 could have provided useful ROI values and images that could still be considered diagnostic, low scores for both perceived and calculated SNR of various DTI metrics may not be adequate in daily clinical practice. A limitation of our study lies in relatively small sample size and the absence of a standard set of scan parameters that would

allow greater generalizability for our results. While NEX = 2 data was sufficient for clinical use in evaluating DTI derived images (DWI, ADC, FA, and EV), we were unable to define an SNR threshold that could be used to discriminate between studies that clinical reviewers would find inferior. In addition, given that NEX = 2 corresponded to the sufficient scan time to use in this study, another approach that would have been

Soman et al: Effect of DTI Repetitions on Diagnostic Experience

5

Fig 4. (A) NEX = 1, (B) NEX = 2, and (C), NEX = 3 EV images with corresponding (D) Axial FLAIR, (E) Coronal T2, and (F) Axial T2-weighted images show cerebral dysgenesis, including polymicrogyria, agenesis of the corpus callosum, and asymmetric white matter (arrows) in a 1-day old male infant. While color EV images across all NEX levels show asymmetric white matter directionality, (represented as blue color) this asymmetry is better delineated in extent on NEX = 2 and NEX = 3 than NEX = 1 images in the left temporo-occipital region (arrowhead) and right temporal operculum, where blue areas extend to the right subinsular and capsular white matter. Higher scores for NEX = 2 and NEX = 3 were given compared to NEX = 1 with regards to measures of perceived SNR, lesion conspicuity, and diagnostic confidence.

interesting to explore is whether twice as many directions with NEX = 1 (that is, a 50 direction data set and NEX = 1 vs. a 25 direction data set and NEX = 2, acquired in the same scan time). While we did not perform tractography for this study, one of our future goals is to attempt to find a reasonable tradeoff between the number of directions, NEX, and scan time for fiber tractography in the context of surgical planning for pediatric patients.

Conclusion Reducing the number of DTI acquisitions (NEX) can be used to shorten scan time without compromising radiologist’s perceived diagnostic quality or quantitative DTI metrics. NEX = 2 represented an appropriate balance between scan time, image quality, and diagnostic confidence, resulting in a 33% reduction in DTI acquisition time compared to NEX = 3.

6

Journal of Neuroimaging Vol 00 No 0 xxxx 2014

References 1. Basser PJ. Inferring microstructural features and the physiological state of tissues from diffusion-weighted images. NMR Biomed 1995;8:333-344. 2. Basser PJ, Pierpaoli C. Microstructural and physiological features of tissues elucidated by quantitative-diffusion-tensor MRI. J Magn Reson B 1996;111:209-219. 3. Pierpaoli C, Basser PJ. Toward a quantitative assessment of diffusion anisotropy. Magn Reson Med 1996;36:893-906. 4. Faria AV, Zhang J, Oishi K, et al. Atlas-based analysis of neurodevelopment from infancy to adulthood using diffusion tensor imaging and applications for automated abnormality detection. Neuroimage 2010;52:415-428. 5. Kumar A, Juhasz C, Asano E, et al. Diffusion tensor imaging study of the cortical origin and course of the corticospinal tract in healthy children. AJNR Am J Neuroradiol 2009;30:1963-1970. 6. Lobel U, Sedlacik J, Gullmar D, et al. Diffusion tensor imaging: the normal evolution of ADC, RA, FA, and eigenvalues studied in multiple anatomical regions of the brain. Neuroradiology 2009;51:253263.

7. Ment LR, Hirtz D, Huppi PS. Imaging biomarkers of outcome in the developing preterm brain. Lancet Neurol 2009;8:1042-1055. 8. Nagy Z, Ashburner J, Andersson J, et al. Structural correlates of preterm birth in the adolescent brain. Pediatrics 2009;124:e964972. 9. Nomura Y, Sakuma H, Takeda K, et al. Diffusional anisotropy of the human brain assessed with diffusion-weighted MR: relation with normal brain development and aging. AJNR Am J Neuroradiol 1994;15:231-238. 10. Qiu M, Li Q, Liu G, et al. Voxel-based analysis of white matter during adolescence and young adulthood. Brain Dev 2010;32:531537. 11. Qu P, Shen GX, Wang C, et al. Tailored utilization of acquired k-space points for GRAPPA reconstruction. J Magn Reson 2005;174:60-67. 12. Rollins NK. Clinical applications of diffusion tensor imaging and tractography in children. Pediatr Radiol 2007;37:769-780. 13. Rollins NK, Glasier P, Seo Y, et al. Age-related variations in white matter anisotropy in school-age children. Pediatr Radiol 2010;40:1918-1930. 14. Mori S, Kaufmann WE, Davatzikos C, et al. Imaging cortical association tracts in the human brain using diffusion-tensor-based axonal tracking. Magn Reson Med 2002;47:215-223. 15. Hoon AH, Jr. Neuroimaging in cerebral palsy: patterns of brain dysgenesis and injury. J Child Neurol 2005;20:936–939. 16. Hoon AH, Jr., Lawrie WT, Jr., Melhem ER, et al. Diffusion tensor imaging of periventricular leukomalacia shows affected sensory cortex white matter pathways. Neurology 2002;59:752-756. 17. Ito R, Mori S, Melhem ER. Diffusion tensor brain imaging and tractography. Neuroimag Clin N Am 2002;12:1-19. 18. MacLennan AI, Smith AF. An analysis of critical incidents relevant to pediatric anesthesia reported to the UK National Reporting and Learning System, 2006–2008. Paediatr Anaesth 2011;21:841-847. 19. Paterson N, Waterhouse P. Risk in pediatric anesthesia. Paediatr Anaesth 2011;21:848-857. 20. Ragg P. Critical incidents and mortality reporting in pediatric anesthesia: the Australian experience. Paediatr Anaesth 2011;21:754-757. 21. Larson PE, Lustig MS, Nishimura DG. Anisotropic field-of-view shapes for improved PROPELLER imaging. Magn Reson Imaging 2009;27:470-479. 22. Griswold MA, Jakob PM, Heidemann RM, et al. Generalized autocalibrating partially parallel acquisitions (GRAPPA). Magn Reson Med 2002;47:1202-1210. 23. Papadakis NG, Xing D, Huang CL, et al. A comparative study of acquisition schemes for diffusion tensor imaging using MRI. J Magn Reson 1999;137:67-82.

24. O’Halloran RL, Aksoy M, Van AT, et al. 3D isotropic highresolution diffusion-weighted MRI of the whole brain with a motion-corrected steady-state free precession sequence. Magn Reson Med 2013;70:466-478. 25. Lemaire C, Moran GR, Swan H. Impact of audio/visual systems on pediatric sedation in magnetic resonance imaging. J Magn Reson Imaging 2009;30:649-655. 26. Barkovich AJ, Raybaud C. Pediatric Neuroimaging. 5th ed. Philadelphia: Wolters Kluwer Health/Lippincott Williams & Wilkins, 2011. 27. Jones DK. The effect of gradient sampling schemes on measures derived from diffusion tensor MRI: a Monte Carlo study. Magn Reson Med 2004;51:807-815. 28. Papadakis NG, Murrills CD, Hall LD, et al. Minimal gradient encoding for robust estimation of diffusion anisotropy. Magn Reson Imaging 2000;18:671-679. 29. Aksoy M, Acar B, Bammer R. SNR dependence of DTI fiber tracking protocols. In: Proceedings of the ISMRM Workshop on Diffusion MRI. Canada: Lake Louise, 2005. 30. Widjaja E, Mahmoodabadi SZ, Rea D, et al. Effects of gradient encoding and number of signal averages on fractional anisotropy and fiber density index in vivo at 1.5 tesla. Acta Radiol 2009;50:1061013. 31. Ni H, Kavcic V, Zhu T, et al. Effects of number of diffusion gradient directions on derived diffusion tensor imaging indices in human brain. AJNR Am J Neuroradiol 2006;27:1776-1781. 32. O’Halloran RL, Holdsworth S, Aksoy M, et al. Model for the correction of motion-induced phase errors in multishot diffusionweighted-MRI of the head: are cardiac-motion-induced phase errors reproducible from beat-to-beat? Magn Reson Med 2012;68:430440. 33. Woolrich MW, Jbabdi S, Patenaude B, et al. Bayesian analysis of neuroimaging data in FSL. Neuroimage 2009;45:S173-S186. 34. Smith SM, Jenkinson M, Woolrich MW, et al. Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage 2004;23(Suppl 1):S208-S219. 35. Forman C, Aksoy M, Hornegger J, et al. Self-encoded marker for optical prospective head motion correction in MRI. Med Image Anal 2011;15:708-719. 36. Wang JY, Abdi H, Bakhadirov K, et al. A comprehensive reliability assessment of quantitative diffusion tensor tractography. Neuroimage 2012;60(2):1127-1138. 37. Farrell JA, Landman BA, Jones CK, et al. Effects of signal-tonoise ratio on the accuracy and reproducibility of diffusion tensor imaging-derived fractional anisotropy, mean diffusivity, and principal eigenvector measurements at 1.5 T. J Magn Reson Imaging 2007;26:756-767.

Soman et al: Effect of DTI Repetitions on Diagnostic Experience

7

Effect of number of acquisitions in diffusion tensor imaging of the pediatric brain: optimizing scan time and diagnostic experience.

Diffusion tensor imaging (DTI) is useful for multiple clinical applications, but its routine implementation for children may be difficult due to long ...
661KB Sizes 0 Downloads 3 Views