Childs Nerv Syst DOI 10.1007/s00381-015-2767-6
Diffusion tensor imaging of the cervical spinal cord in children Gunes Orman 1 & Kevin Yuqi Wang 1 & Ximin Li 2 & Carol Thompson 2 & Thierry A. G. M. Huisman 1 & Izlem Izbudak 1
Received: 16 March 2015 / Accepted: 22 May 2015 # Springer-Verlag Berlin Heidelberg 2015
Abstract Purpose Obtaining fast, reliable, high-resolution diffusion tensor imaging (DTI) of the pediatric cervical spinal cord (CSC) is challenging, given the multitude of technical limitations involved. Overcoming these limitations may further potentiate DTI as a valuable quantitative tool in evaluating the pediatric CSC. Methods Sixteen patients (9 girls and 7 boys) with hypoxic brain injury, craniocervical junction malformations, and head trauma were included in this retrospective study. Region of interests were placed from C1–C2 through C7–T1 consecutively at the cervical intervertebral disc levels. DTI metrics were compared with a pediatric DTI database of healthy controls. Clinical background and outcomes were tabulated. Results Patients with hypoxic brain injury, Chiari I and II malformations, and head trauma demonstrated lower fractional anisotropy values than that of healthy controls at certain cervical intervertebral disc levels. Conclusions DTI may be a promising modality for providing additional information beyond that of conventional magnetic resonance imaging in pediatric central nervous system disorders.
* Izlem Izbudak [email protected]
Section of Pediatric Neuroradiology, Division of Pediatric Radiology, The Russell H. Morgan Department of Radiology, The Johns Hopkins University School of Medicine, 600 N Wolfe Street, Phipps B-126B, Baltimore, MD 21287, USA
Department of Biostatistics, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA
Keywords Diffusion tensor imaging . Fractional anisotropy . Mean diffusivity . Pediatric . Spinal cord
Introduction Diffusion tensor imaging (DTI) is an advanced magnetic resonance imaging (MRI) technique that enables the qualitative and quantitative characterization of the magnitude, shape, and direction of diffusion of water molecules along three principal eigenvectors. It is therefore particularly useful for the in vivo evaluation of white matter tracts. It has been predominantly studied in the cervical spinal cord (CSC) of adults, recently in that of children [1–6]. However, DTI of the CSC presents several technical challenges. There often is a low signal-tonoise ratio given the small CSC volume. Different tissue interfaces involving bone, soft tissue, and fluid may also create significant susceptibility artifacts [3, 7, 8]. Lastly, cerebrospinal fluid (CSF) pulsations, blood flow, respiratory and cardiac movements, swallowing, and other motion-related factors produce significant artifacts and may degrade the image quality. Moreover, increased motion artifacts are especially problematic in evaluation of the pediatric CSC, making accurate and reproducible measurements of DTI metrics particularly challenging. Techniques such as cardiac gating and respiratory compensation may be used to mitigate these artifacts. However, they often increase the acquisition time. Sedation to reduce motion artifacts is also not particularly ideal in the pediatric population [3, 9, 10]. Therefore, reliably fast, high-resolution DTI of the pediatric CSC is required, and various techniques of DTI have been investigated in children [3–6, 11]. Age-related decreases in mean diffusivity (MD) and increases in fractional anisotropy (FA) in the CSC and upper thoracic cord have been observed in a cohort of healthy, developing children . FA
Childs Nerv Syst
is a measure of the degree of anisotropy, and MD is the trace of the diffusion tensor matrix, and DTI metrics of different pathologies allow more accurate characterization of intrinsic integrity of tissues including cellular density and architecture . In this study, we illustrate the DTI metrics of CSC with several disorders in children, including hypoxic brain injury, he ad tr aum a, an d c ra nio ce rvi cal jun cti on (CCJ) malformations along with their clinical background and outcomes.
Methods Patients This was a retrospective study approved by the Johns Hopkins Medicine Institutional Review Board. DTI of the CSC of children have been acquired between January 2013 and September 2013. We included patients who met the following criteria into the study: (1) age less than 18 years old, (2) MRI and DTI data were obtained exclusively at Johns Hopkins Hospital, (3) clinical confirmation or documentation of a disorder of the central nervous system (CNS), or trauma, and (4) positive imaging findings. Exclusion criteria were (1) non-diagnostic image quality, (2) incomplete clinical and imaging information, and (3) spinal instrumentation or other metal artifacts at multiple spinal levels. Imaging techniques All MRI studies were obtained on a 1.5-T clinical MR scanner (Magnetom Aera, Siemens Healthcare, Germany) using a standard 4-element neck matrix coil and the standard 32element spine matrix coil. Routine conventional sequences of the spine were performed including axial and sagittal T1and T2-weighted turbo spin-echo imaging and sagittal short [inversion time (TI)] inversion recovery (STIR) sequences. Intravenous contrast was administered when appropriate for the clinical indication of the study. Sagittal DTIs of the CSC were obtained using a multi-shot (2 shots), interleaved, multi-slice, spin echo, echo planar imaging (EPI) pulse sequence in combination with parallel imaging with an acceleration factor of two. DTI images were acquired using the following scan parameters: TR= 3200 ms; TE=73 ms; field-of-view (FOV)=200 mm; slice thickness=2.5 mm; imaging matrix=176×176 with an inplane resolution of 1.2×1.2 mm. Balanced pairs of gradients were used to minimize eddy current effects. Saturation bands anterior to the cervical spine were used to decrease the susceptibility and motion artifacts from adjacent tissue interfaces. A total of 20 directions were sampled by diffusion-encoding gradients with b values 0 and 750 mm2/s without respiratory
and cardiac gating. The DTI data sets were post processed with calculation of MD, FA, and color-coded FA maps using the standard Siemens Neuro 3D post-processing software. Data analysis The conventional MRI images and post-processed DTI maps were visually evaluated for quality by two experienced pediatric neuroradiologists (TAGMH and II). Any spinal cord level that showed distortion, artifacts, or inhomogeneity of signal intensities was excluded. Region of interest (ROI)s were manually placed at the center of the cervical and upper thoracic spinal cord at intervertebral disc levels defined on a midsagittal trace image. Meticulous ROI placement to avoid partial volume averaging effect with the CSF was performed at mid-sagittal images at the center of the CSC. The ROIs were placed from C1–C2 through C7–T1 intervertebral disc levels at the CSC by the first author of the manuscript. Size of the ROIs ranged from 2–4 mm in diameter. Mean FA and MD values and corresponding standard deviations were extracted for the selected voxels. DTI metrics with more than 30 % of the standard deviation were discarded. The CSC DTI results were compared with values from a normal pediatric CSC DTI database .
Results Sixteen patients (9 girls and 7 boys) were included in the study. The mean age of the children was 4.41 (between 0.17 and 11) years. Patients were classified into three groups based on their main diagnosis. The clinical history, timeline, and outcomes were summarized in the table. Group 1 Group 1 comprised of three patients with history of hypoxic brain injury with different degree of severity. Patient 1, who suffered prolonged asphyxia and cardiopulmonary arrest 9 months prior to imaging, demonstrated significantly lower FA and higher MD values when compared to those of normal age-matched controls (Fig. 1). In contrast, conventional MRI of the CSC was normal (Fig. 2a). Severe diffuse chronic ischemic changes were seen on brain MRI (Fig. 3). Group 2 Group 2 comprised of nine patients with CCJ malformations. Patients 4, 8, and 12, with histories of Chiari 1 malformation, demonstrated abnormal FA values when compared to that of normals, especially at the first two cervical spinal levels (Fig. 4). Conventional MRI revealed tonsillar herniation and confirmed compression of the upper CSC (Fig. 5a).
Childs Nerv Syst
Fig. 1 Group 1 fractional anisotropy (a) and mean diffusivity (b) values are shown at each cervical intervertebral disc level in comparison to respective mean values of the control database with corresponding 95 % confidence intervals
Fig. 3 Magnetic resonance imaging of the brain of patient 1 performed 3 days following anoxic injury reveals restricted diffusion involving the subcortical white matter in bilateral centrum semiovale, bilateral occipital lobes (a), and elevated FLAIR signal in the bilateral posterior putamina and thalami (b). Magnetic resonance imaging 9 months after anoxic injury reveals global cerebral volume loss with ex-vacuo dilatation of the ventricles and abnormal signal within the gray matter (c, d)
Patient 5, who had a history of achondroplasia as demonstrated on MRI (Fig. 6), was noted to have abnormal FA and MD values at all cervical spinal levels (Fig. 4). The most significant changes were at the first two cervical spinal levels.
Patients 6, 9, and 10 are those with Chiari 1 malformations, subsequent occipital decompressions, and hydromyelia of the central canal (Table 1). Patient 6 had normal FA and MD values at all levels except for
Fig. 2 Magnetic resonance imaging of the cervical spinal cord of an eleven-year-old boy (patient 1) with asthma exacerbation and diffuse anoxic brain injury secondary to prolonged cardiopulmonary arrest 9 months prior reveals normal signal on sagittal T2-weighted imaging
(a) and apparent diffusion coefficient map (b). Seven region of interests are placed on color-coded fractional anisotropy maps (c) at each intervertebral disc level. Tractography (d) reveals intact fibers
Childs Nerv Syst
Patient 7 and 11 are those with Arnold-Chiari 2 malformation related to lumbar myelomeningocele. Patient 7, with hydromyelia extending from C6 to T2, revealed abnormal FA values at level C4–C5 and caudally (Fig. 4a). Patient 11 had abnormal FA and MD compared to that of the normal group throughout all cervical levels. Group 3 Group 3 comprised of four patients with histories of either minor or major head trauma. The time interval between trauma and imaging ranged from 1 to 3 days (Table 1). Decreases in FA occurred at the first two cervical spinal levels in all patients (Fig. 7a). MD changes were more subtle beyond the first two disc levels except patients 13 and 14 (Fig. 7b). The final clinical outcomes of the patients are also given in the Table.
Fig. 4 Group 2 fractional anisotropy (a) and mean diffusivity (b) values are shown at each cervical intervertebral disc level in comparison to respective mean values of the control database with corresponding 95 % confidence intervals
those with involvement of hydromyelia from C6 to T2 (Fig. 4). In contrast, patient 9 (no cervical involvement of hydromyelia) and patient 10 (C2 to L3 involvement) had abnormal FA and MD values throughout all cervical spinal levels (Fig. 4). Fig. 5 Sagittal T2-weighted imaging of a two-year-old girl (patient 8) with Chiari I malformation and no prior occipital decompression reveals crowding of the cerebellar tonsils at the foramen magnum with cerebellar tonsillar descent 9 mm below the foramen magnum (a). Fractional anisotropy (b), apparent diffusion coefficient (c), and color-coded fractional anisotropy (d) maps reveal similar tonsillar descent. Tractography (e) reveals intact fibers
DTI is a non-invasive functional MRI technique that exploits the three-dimensional diffusion characteristics of water in the CNS. Diffusion is considered isotropic if it shows no directional dependence (e.g., in gray matter) or anisotropic if directional dependence is present (e.g., in white matter). From this technique, various DTI scalars can be calculated, including MD, which is a marker of overall diffusion and also known as apparent diffusion coefficient (ADC), and FA, which quantifies the degree of anisotropic diffusion. The application of this technique in the brain to characterize the microstructural changes secondary to ischemia, infection, and neoplasm has been well established. More recently, studies have also demonstrated the potential of DTI in characterizing SC pathologies of adults and children [3–6, 12]. However, there are significantly fewer DTI studies evaluating the pediatric SC.
Childs Nerv Syst
Fig. 6 Sagittal T2-weighted imaging (a) of a nine-year-old girl with achondroplasia (patient 5) demonstrates craniocervical junction narrowing and spinal cord compression (arrow). Sagittal T2-weighted imaging (b) of a four-year-old boy with Chiari 1 malformation (patient 4) reveals spinal cord compression and descent of cerebellar tonsils
15 mm below the foramen magnum (arrow). Sagittal T2-weighted imaging (c) of a nine-year-old girl with suboccipital craniectomy (arrow) for Chiari 1 malformation (patient 6) reveals normal position of cerebellar tonsils and extension of hydromyelia from C6 to T2
Clinical background of patients
Patient group Patient number Patient age History
Level of Time interval Time interval Clinical outcomes hydromyelia between prior between operation insult and DTI and DTI
Anoxic brain after status asthmaticus
Post-op CM 1
Post-op CM 2
Pre-op CM 1
Post-op CM 1
Post-op CM 1
Post-op CM 2
Pre-op CM 1
Head trauma after MVA
Head trauma following fall
Head trauma, non-accidental Head trauma following fall
CM Chiari malformation, MVA motor vehicle accident
Wheelchair-bound, unable to talk, significantly increased tone throughout Quadriplegia, dysphagia, seizure disorder, developmental delay
Neck pain and headaches
Hydromyelia in the spinal cord between T2 and T10 levels without any symptoms
Hydromyelia in the spinal cord between C6 and T2 levels without any symptoms
Shunted hydrocephalus, central apnea: deceased
N/A 4 months 1.5 years
Back pain and headaches Continuous headaches but less severe than pre-op Hydromyelia in the spinal cord between C2 and L3 levels without any symptoms Urinary incontinence Headaches
Recovery Impaired ambulation, decreased strength
Childs Nerv Syst
Fig. 7 Group 3 fractional anisotropy (a) and mean diffusivity (b) values are shown at each cervical intervertebral disc level in comparison to respective mean values of the control database with corresponding 95 % confidence intervals
We acquired CSC DTI by using sagittal small FOV multishot EPI technique with 20 diffusion gradient angles in approximately 5 min of scan time, which is a reasonable time to include in our routine pediatric clinical CSC MRI protocol. Nevertheless, CSC DTI scanning in axial plane takes longer causing an increase in total scanning time; therefore, we have favored obtaining only sagittal DTI acquisition. In our previous study with 47 healthy pediatric controls, which served as the normal database in this study, we have used the sagittal DTI technique and found consistent FA and MD values at each spinal cord level . Our results also correlate well with previously published limited number of papers [3, 5, 12, 13]. Each patient in Group 1 (hypoxic brain injury) experienced a different interval between the time of insult and time of MRI. Patient 1, with a 9-month history of chronic hypoxic changes to the brain, demonstrated FA reduction and MD elevation at all cervical spinal levels, despite a normal conventional cervical MRI (Figs. 1, 2, and 3). MD and FA values were not as nearly elevated and reduced, respectively, in patients with recent hypoxic insults, as demonstrated in patients 2 and 3. This
may suggest that with time, prolonged hypoxia to the brain is also associated with changes in the CSC. Indirect injury via Wallerian degeneration may perhaps play a role. We present here one patient which is intriguing for further research to understand the mechanism of microstructural changes in the spinal cord after a brain hypoxic event. In Group 2, patients with CCJ malformations with no posterior fossa decompression demonstrated abnormal FA and MD values compared to normals at the upper CSC levels (Fig. 4) (Table 1). A recent study by Bosemani et al. similarly showed decreased FA and increased MD values in the lower brainstem of eight children with achondroplasia . These changes likely reflect early compressive myelopathy at the CCJ. DTI may be especially useful in younger children with CCJ malformation, who may be difficult to evaluate clinically. Another study demonstrated that the duration of the compression significantly affected clinical outcomes in patients with Chiari I malformation and hydrocephalus . Further investigation of the clinical utility of DTI in this patient population may potentiate it as an additional tool to monitor progression. It is also conceivable that certain threshold metrics may be utilized to help optimizing timing of the surgery to mitigate development of hydrocephalus. In Group 3, all patients had head trauma with different mechanisms of injury (Table 1). Younger children tend to have relatively larger and heavier heads in relation to the rest of the body when compared to adults [16, 17], thereby increasing the risk for spinal cord injury at CCJ. For example, patient 15 in this study with non-accidental trauma and normal conventional CSC MRI findings had abnormal FA and MD values at the first two cervical levels. The anatomical characteristics of the infant neck increase the likelihood of cervical flexion, extension, and rotational injury. Axonal damage in the corticospinal tracts was shown histopathologically to support the hypothesis of stretching injury to the spinal cord during flexionhyperextension trauma to the neck in association with nonaccidental injury . Although there was no evidence of direct injury to CSC based on clinical history or conventional MRIs in Group 3, CSC may have been affected by rotational or flexion-extension forces at the CCJ during head trauma or shaking. We are aware of some limitations in this study. First, the small number of patients within each group prevents appropriate statistical comparisons between controls and patients, and thus, nonparametric analysis was not conducted in this study. This is further obscured by the heterogeneity of the nature and mechanism of specific disorders within each group. But we aimed to share our CSC DTI experience in distinct pathologies with intriguing results in this preliminary report. Second, due to the retrospective character of the study, there was no standardization of timing of imaging with respect to the prior insult. Therefore, we provided the time interval between the insult and imaging. Lastly, in spite of its pervasive
Childs Nerv Syst
use in DTI studies, manual ROI placement may not provide the desired inter-rater or intra-rater reliability, obfuscating the ability to reproduce DTI measurements. In conclusion, in this study, we have shown that pediatric cervical spinal cord DTI is feasible and may provide additional microstructural information in diseases such as Chiari 1 malformation, hypoxic brain injury, or accidental/nonaccidental head trauma. Further improvements in methods of data acquisition and MR scanning time may enable clinical use in the future. Acknowledgments We would like to thank Scott Pryde (Section of Pediatric Neuroradiology, Division of Pediatric Radiology, The Russell H. Morgan Department of Radiology The Johns Hopkins University School of Medicine) for his technical support in this study. Grant sponsor Ximin Li and Carol Thompson’s work was supported by the National Center for Research Resources and the National Center for Advancing Translational Sciences (NCATS) of the National Institutes of Health through Grant Number UL1T000424. Conflict of interest The authors declare that they have no competing interests.
References 14. 1.
Vargas MI, Delavelle J, Jlassi H, Rilliet B, Viallon M, Becker CD, Lovblad KO (2008) Clinical applications of diffusion tensor tractography of the spinal cord. Neuroradiology 50:25–29 2. Thurnher MM, Law M (2009) Diffusion-weighted imaging, diffusion-tensor imaging, and fiber tractography of the spinal cord. Magn Reson Imaging Clin N Am 17:225–244 3. Barakat N, Mohamed FB, Hunter LN, Shah P, Faro SH, Samdani AF, Finsterbusch J, Betz R, Gaughan J, Mulcahey MJ (2012) Diffusion tensor imaging of the normal pediatric spinal cord using an inner field of view echo-planar imaging sequence. AJNR Am J Neuroradiol 33:1127–1133 4. Barakat N, Mulcahey MJ, Shah P, Samdani A, Krisa L, Faro S, Mohamed FB (2012) Diffusion tensor imaging in pediatric transverse myelitis: a case study. J Pediatr Rehabil Med 5:281–286 5. Mohamed FB, Hunter LN, Barakat N, Liu CS, Sair H, Samdani AF, Betz RR, Faro SH, Gaughan J, Mulcahey MJ (2011) Diffusion tensor imaging of the pediatric spinal cord at 1.5 T: preliminary results. AJNR Am J Neuroradiol 32:339–345
Mulcahey MJ, Samdani A, Gaughan J, Barakat N, Faro S, Betz RR, Finsterbusch J, Mohamed FB (2012) Diffusion tensor imaging in pediatric spinal cord injury: preliminary examination of reliability and clinical correlation. Spine 37:797–803 Figley CR, Stroman PW (2007) Investigation of human cervical and upper thoracic spinal cord motion: implications for imaging spinal cord structure and function. Magn Reson Med 58:185–189 Kharbanda HS, Alsop DC, Anderson AW, Filardo G, Hackney DB (2006) Effects of cord motion on diffusion imaging of the spinal cord. Magn Reson Med 56:334–339 Malisza KL, Martin T, Shiloff D, Yu DC (2010) Reactions of young children to the MRI scanner environment. Magn Reson Med 64: 377–381 Rosenberg DR, Sweeney JA, Gillen JS, Kim J, Varanelli MJ, O’Hearn KM, Erb PA, Davis D, Thulborn KR (1997) Magnetic resonance imaging of children without sedation: preparation with simulation. J Am Acad Child Adolesc Psychiatry 36:853–859 Kumar A, Juhasz C, Asano E, Sundaram SK, Makki MI, Chugani DC, Chugani HT (2009) Diffusion tensor imaging study of the cortical origin and course of the corticospinal tract in healthy children. AJNR Am J Neuroradiol 30:1963–1970 Singhi S, Tekes A, Thurnher M, Gilson WD, Izbudak I, Thompson CB, Huisman TA (2012) Diffusion tensor imaging of the maturing paediatric cervical spinal cord: from the neonate to the young adult. J Neuroradiol 39:142–148 Izbudak I, Dumrongpisutikul N, Thompson CB, Singhi S, Tekes A, Thurnher M, Huisman TAGM (2011) Trends and differences in DTI metrics across ages and spinal cord levels in normal children. International Society of Magnetic Resonance in Medicine, Montreal, Quebec, Canada, p 4203 Bosemani T, Orman G, Carson KA, Meoded A, Huisman TA, Poretti A (2014) Diffusion tensor imaging of the brainstem in children with achondroplasia. Dev Med Child Neurol 56:1085–1092 Deng X, Wu L, Yang C, Tong X, Xu Y (2013) Surgical treatment of Chiari I malformation with ventricular dilation. Neurol Med Chir 53(12):847–852 Pinto PS, Poretti A, Meoded A, Tekes A, Huisman TA (2012) The unique features of traumatic brain injury in children. Review of the characteristics of the pediatric skull and brain, mechanisms of trauma, patterns of injury, complications and their imaging findings— part 1. J Neuroimaging 22:e1–17 Pinto PS, Meoded A, Poretti A, Tekes A, Huisman TA (2012) The unique features of traumatic brain injury in children. Review of the characteristics of the pediatric skull and brain, mechanisms of trauma, patterns of injury, complications, and their imaging findings— part 2. J Neuroimaging 22:e18–41 Kemp A, Cowley L, Maguire S (2014) Spinal injuries in abusive head trauma: patterns and recommendations. Pediatr Radiol 44: 604–612