| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Technical Developments |
1 From the Departments of Clinical Physics (C.v.P., P.F.F.W.), Neonatology (J.B.), and Radiology (F.G.R.), Máxima Medical Center Veldhoven, De Run 4600, PO Box 7777, 5500 MB Veldhoven, the Netherlands; and Departments of Applied Physics (C.v.P., P.F.F.W.) and Biomedical Engineering (A.V.), Eindhoven University of Technology, Eindhoven, the Netherlands. Received September 3, 2004; revision requested November 11; revision received February 3, 2005; accepted March 11; final version accepted August 1. Address correspondence to C.v.P.(e-mail: C.vanPul{at}mmc.nl).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
© RSNA, 2006
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The process of myelination can be followed by using conventional magnetic resonance (MR) imaging techniques, such as T1- and T2-weighted imaging (4), or by using the relatively recently introduced technique, diffusion-tensor imaging (5–7). A preference of water molecules for diffusion in a particular direction (ie, anisotropic diffusion) reflects the underlying tissue structure. Diffusion-tensor imaging is based on the acquisition of a tensor to describe diffusion. The main eigenvector of the diffusion tensor defines the preferred direction when the underlying tissue structure (eg, white matter) is linear (8,9). Various methods for visualization of the diffusion anisotropy and the preferred direction are available and include anisotropy maps that display the anisotropy index (10) and color maps that depict the diffusion direction with color-coded pixels (11–13). Fiber tracking, a relatively recently introduced technology, is a three-dimensional visualization technique in which the underlying linear structure defined by the diffusion tensor is reconstructed (9,14–16).
White matter in the adult brain is highly myelinated, and visualization of the main direction of diffusion corresponds to known white matter pathways on anatomy atlases, as outlined in detail on a recently published brain fiber atlas (13). In neonates, the white matter is still developing. Diffusion anisotropy has been reported to occur before myelination occurs and to increase with increasing myelination (5–7,17).
The preliminary results of fiber tracking with 3.0-T MR imaging in newborns obtained by Zhai et al (18) show that this technique is feasible in the neonatal brain, although only the major white matter tracts could be traced. In a case report of one neonate and two children aged approximately 2 years, fiber tracking was shown to yield additional information about the white matter configuration during dysgenesis of the CC (19). Furthermore, asymmetry in the pyramidal tracts has been observed in children with congenital hemiparesis (20).
A high incidence of severe motor problems in newborns with hypoxic-ischemic injury in the white matter and/or injury in the basal ganglia has been reported (1,21,22). Thus, the purpose of our study was to retrospectively investigate the feasibility of fiber tracking at birth and 3 months in infants with hypoxic ischemia to detect disturbances in white matter development.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Infant Patient Group
The study group was recruited from a group of neonates who were born at term with perinatal hypoxic ischemia and were treated in our neonatal intensive care unit. Perinatal hypoxic ischemia was diagnosed when clinical symptoms of neonatal encephalopathy were present with two or more of the following risk factors during the first days of life: fetal heart rate abnormality, umbilical artery pH level lower than 7.10, meconium-stained fluid, and Apgar score 5 minutes after birth lower than 7. Included in our study group were all infants from this group in whom a diffusion-tensor image suitable for fiber tracking at birth and at 3 months was obtained. All babies with congenital abnormalities, neurometabolic disease, and/or perinatal infection were excluded.
The data from a total of 10 full-term infants were included between November 2002 and June 2004. All included babies were born after an uneventful pregnancy. The onset of symptoms could be timed to the day of birth. The newborns (nine male, one female) had a mean gestational age of 40.2 weeks ± 1.6 (standard deviation 
in the group) and were imaged at ages 1–9 days (n = 9); one baby was imaged 17 days after birth. Follow-up MR imaging was performed a mean of 3.4 months ± 0.5 later. Neurodevelopmental evaluations were performed when the infants were between 6 and 12 months of age, according to the protocol of Amiel-Tison et al (23). The Bayley Scales of Infant Development (edition 2, Dutch version) were used to test the psychomotor development of the infants when they were aged 1 year or older. According to the Bayley classification system, a psychomotor development score lower than –2 indicates severe delay and a score between –2 and –1 indicates moderate delay. Neurodevelopmental examinations were performed in the infants at ages 6 months (n = 5) and 1 year (n = 5).
Reference Values
Because we did not have a control group of healthy newborns, we obtained reference values from seven full-term newborns in whom MR imaging was performed for clinical purposes during the same period that the newborns in the patient group were imaged. The reference values were obtained from full-term newborns who had various clinical symptoms; the values were not chosen by using the criteria used to select the hypoxic-ischemic encephalopathy group. The newborns in this reference group had clinical symptoms of Sarnat stage 1 or 2 perinatal hypoxic ischemia (five babies), perinatal infection (one baby), and mild persistent neonatal pulmonary hypertension (one baby, who needed endotracheal ventilation for 3 days and nitric oxide therapy for 1 day and had a normal clinical short- and long-term follow-up). All of these newborns were imaged 4–13 days after birth. MR imaging revealed normal findings (n = 4), minor signal intensity abnormalities in the cortex (ie, high signal intensity on the T1-weighted image [22]) with no signal intensity abnormalities in the white matter (n = 2), or a minor abnormality in the cerebellum (n = 1). This reference group consisted of one female and six male neonates with a mean gestational age of 39.0 weeks ± 1.6. Their brain development and myelination were assessed at T1- and T2-weighted imaging and found to be normal. It was verified that no abnormalities were visible on the diffusion-weighted images, all of which were acquired within the first week after birth (by J.B., F.G.R.). The same neurodevelopmental examinations that were performed in the patient group were performed in these children at ages 6 months (two babies) and 1 year (five babies). All of these babies were deemed to be developmentally healthy. Two babies in this group were examined at follow-up MR imaging, including diffusion-tensor imaging, at age 3 months.
MR Imaging Protocol
A 1.0-T MR imaging unit (Gyroscan; Philips Medical Systems, Best, the Netherlands) was used. The imaging protocol included the acquisition of T1-weighted spin-echo (568/18 [repetition time msec/echo time msec]), T2-weighted fast spin-echo (4381/120), and inversion-recovery (3436/18/400 [repetition time msec/echo time msec/inversion time msec]) MR images. The neonatal patients were sedated with chloral hydrate and fixated by using a vacuum pillow. For all brain imaging, a standard birdcage head coil was used.
The imaging protocol also included diffusion-tensor imaging, which was performed by using pulsed field gradients in six directions with single-shot echo-planar MR imaging. In the neonatal brain, 20 adjacent sections were recorded within 3 minutes with 3595/82 and a voxel size of 1.56 x 1.56 x 3 mm. Two signals were acquired. Three b values (0, 400, and 800 sec/mm2) were used to calculate the apparent diffusion coefficient (ADC). A duration of one gradient lobe of the pulsed field gradients of 20.06 msec, a time between the onset of the first gradient lobe and the onset of the second gradient lobe of 40.5 msec, and a gradient lobe strength of 19.6 mT/m were used with a b of 400 sec/mm2, and a gradient lobe strength of 27.8 mT/m (by combining the x, y, and z gradients) was used with a b of 800 sec/mm2. We chose to use b values lower than those used in the adult brain (typically 1000 sec/mm2) because the ADC in the white matter of neonates is higher than that in the white matter of adults. The optimal b value was calculated by using the method of Jones et al (24).
Postprocessing and Image Analysis
ADC maps were constructed by using software on the MR imaging unit. The diffusion-tensor images were converted into the file format used in the fiber-tracking program by using Mathematica 4.2 software (Wolfram Research, Urbana, Ill). For display of global information, color mapping and fiber tracking were implemented. Color mapping was performed by mapping the first eigenvector of the diffusion tensor directly to the red-green-blue channel and thereby color coding the diffusion direction. Fiber tracking is based on the line propagation technique (9), which involves the use of the second-order Runge-Kutta method for numerical integration. Stopping criteria, which are used in the fiber-tracking procedure to determine when the tracking is to be stopped, could be defined by the user. Examples of user-defined stopping criteria would be an anisotropy value smaller than the threshold value and an 
angle (maximal angle between connected vectors during fiber tracking) larger than the threshold angle. For visualization, the user could also choose to display only fibers of a certain length. The stopping criteria that were used in this study are listed in Table 1.
|
We have implemented a volume-tracing method of fiber tracking in which all voxels are considered as possible seed points. This approach yields long and evenly spaced fibers. The user defines how dense the visualized fibers must be by choosing values for the minimal distance between the seed points and the fibers and for the minimal distance allowed between the fibers once tracking has started. Tracking of a given fiber stops if the tracking starts to occur too close to other fibers—that is, within a distance that is less than the minimal distance allowed between the fibers during tracking. This algorithm has been described in detail previously (25).
For detailed visualization of a particular white matter structure, multiple ROIs for seed point selection can be defined. ROIs that are sampled uniformly to generate seed points are called "Or ROIs" derived by using the "Or" criterion. ROIs defined to show the fibers that pass through it are called "And ROIs" derived by using the "And" criterion. If no ROI derived with the "Or" criterion is defined, then the entire volume is seeded and just the fibers that pass through the "And" ROI are shown. This method prevents structures from being missed owing to undersampling of the ROI. It must be noted that, in contrast to conventionally displayed radiologic images, fiber-tracking images cannot always be displayed from the feet to the head. Therefore, for all images, the viewpoint has been indicated.
The color maps were used to evaluate the directionality and anisotropy of several white matter structures by means of visual inspection. The structural information observed on the color maps was compared with the structures observed on the color maps used by Makris et al (11), Wakana et al (13), and Jellison et al (26). Quantitative anisotropy maps were used to determine whether anisotropy was increased in affected regions and thus to determine whether this increased anisotropy could influence the fiber tracking. Volume tracing was used to determine, in a user-independent way, which fiber structures were visible. The obtained fiber-tracking images were compared with the fiber-tracking atlases defined by Wakana et al (13) and Jellison et al (26). The color maps, anisotropy maps, and fiber-tracking images were evaluated and interpreted in consensus by three authors (C.v.P., J.B., and F.G.R., all with 5 years experience in pediatric brain MR imaging and/or diffusion-tensor imaging). The fiber structures observed in the babies with standard MR imaging findings served as a reference.
The fiber structures observed in the babies at 3 months were compared with the structures observed in the same babies at birth. Some tracts, like the inferior fronto-occipital fasciculus and the optic radiation, have been reported to course close together and to be oriented in a similar direction (26–28). These tracts could not be detected separately by using color mapping or volume tracing and were therefore considered to be a combined tract (ie, fronto-occipital fasciculus and/or optic radiation [IFO/OR] tract); thus, "IFO/OR" indicated that the fiber path detected could have corresponded to either or both tracts. These tracts can be separated by using ROIs with the "And" criterion; however, this process was not performed in this study. We used the term fiber pathology to describe an abnormality in which a visible asymmetry in the fiber pattern between the left and right hemispheres or a large deviation in the fiber pattern, as compared with the fiber pattern in the reference infant group, was observed. After determining which structures were visible, with use of an ROI-based tracing method, we defined multiple ROIs to visualize the structures observed at volume tracing in more detail.
We chose the anisotropy index Cl—not fractional anisotropy or relative anisotropy—as a stopping criterion because it is more sensitive to linear anisotropy and less sensitive to planar anisotropy, as determined by using normalized eigenvalue space plots (29). The values for Cl were based on the values measured in neonatal white matter and CC (30). To avoid fiber information cluttering, the chosen value at which fiber volume tracing was stopped was larger than that at which ROI-based tracing was stopped (Table 1).
With both volume tracing and ROI-based tracing, fibers were displayed only if they were longer than a user-defined length, because visualization of the smaller fibers could have resulted in additional information cluttering. For neonates at birth, a minimal length of greater than 3 mm was used, and for babies at 3 months of age, a minimal length of greater than 5 mm was used (Table 1).
Quantification of PLIC Fibers
The motor and somatosensory pathways in the CR pass perpendicular to the transversal plane through the PLIC (31). Herein, these fibers are referred to as PLIC fibers. To determine the length and volume of the PLIC fibers, two ROIs were drawn roughly around the PLIC in two adjacent transversal sections. We used the "And" criterion: Fiber tracking was initiated from all points in the volume, and only those fibers passing through both ROIs were displayed. This technique was similar to the method used by Mori et al (31). If an abnormality was suspected to be disturbing fiber tracking, the pattern of disturbance was checked by using volume tracing. The detected disturbance pattern would be used to determine whether normal ROI selection should be used or the ROI selection should be defined in other nearby transversal sections to prevent fibers from being missed owing to ROI misplacement. Analysis of the PLIC fibers was performed in consensus by three authors (C.v.P., J.B., F.G.R.). After visualization of the PLIC fibers, the mean length of all the fibers was calculated. It must be noted that not the true length of the white matter fibers, but rather an indication of the length of the linear diffusion path of diffusing water molecules, was determined. The fiber volume was the volume of all pixels through which one or more fibers passed. In all infants, the mean fiber lengths and volumes calculated with three Cl values (0.08, 0.10, and 0.12) were determined at birth and at 3 months.
Statistical Analyses
To compare the PLIC fiber lengths measured in the left and right hemispheres, a Student t test was used. A paired Student t test was used to compare the PLIC fiber lengths measured in neonates at birth with those measured in the same babies at 3 months. P < .05 was considered to indicate significance (32). Statistical analyses were performed by using Mathematica 4.2 software. These analyses were performed by two authors (C.v.P., P.F.F.W.).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
In the infant group with fiber abnormalities, the color maps showed a disturbed pattern compared with the fiber pattern observed in the reference infant group. In the affected areas, there was increased anisotropy, which was apparent in white matter areas and in the gyri.
Fiber Tracking
In the infants with standard MR imaging findings, several white matter tracts could be appreciated on fiber-tracking images at birth and at 3 months (Fig 1). No differences between the right and left hemispheres were observed in this group. The following main white matter structures could be visualized with fiber tracking in these infants at birth: the CR in all seven babies, the genu and splenium of the CC in all seven; the entire CC in two; the posterior, middle, and frontal regions of the IFO/OR in all seven, six, and two babies, respectively; the fornix in all seven; and the cingulum (directly above the CC) in five. All other fibers defined on published fiber-tracking atlases (13,26) could not yet be detected at birth. It must be noted that in some newborns in the reference group, structures that were not well visualized on the color maps were visible on the fiber-tracking images. On the other hand, not all structures that were visible on the color maps were traceable at birth—for example, usually only a small part of the cingulum could be traced. In the two infants with standard MR imaging results available at 3 months, the structures observed at birth were more developed at 3 months, although newly developed structures could not be detected.
|
|
|
|
|
|
One of the problems with using volume tracing is that information cluttering can occur during the tracing owing to artifacts on the MR images (Fig 3a), which do not occur when the ROI-based tracing method is used (Fig 3b). The data set displayed in Figure 3 is one of our worst examples. In general, the artifacts appeared as less dense lines. They could be easily distinguished from real fibers because they occurred only at the air-tissue boundaries. With our fiber-tracking protocol, it is possible to rotate the structures, use thick or thin fibers, and "look through" the cluttering.
|
|
|
in the group) for both hemispheres and the mean volume was 3.0 cm3 ± 1.0 for both hemispheres. Of the infants with fiber abnormalities, four had significantly reduced PLIC fiber lengths compared with the fiber lengths measured in the reference infant group (Fig 5a; values outside the shaded area indicate P < .05). In five patients, the measured fiber volumes were reduced significantly compared with the reference values (P < .05) (Fig 5b). The mean fiber length and volume in the patient group at 3 months were significantly increased compared with the values at birth (P < .01). For three infants (patients 6–8), the mean fiber volume (in one or both hemispheres) at 3 months deviated largely from the mean value for the other patients (Fig 5c, 5d). However, no comparable normal values were available because only two infants from the reference group were imaged at 3 months (Fig 5).
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Diffusion anisotropy has been observed before myelination occurred (5,6). Neil et al (5) divided the white matter regions visualized at birth into three types: (a) those packed into myelinated parallel bundles (type 1), (b) those with mostly nonmyelinated parallel bundles (type 2), and (c) those that are neither closely packed nor myelinated (type 3). Type 1 white matter tracts—for example, the central part of the CR that is myelinated in infants at birth—can be easily traced (2,6,33). The splenium of the CC and the optic radiation become myelinated in infants within the first 3 months and are type 2 tracts. These tracts can be visualized in infants at birth by using fiber tracking owing to the parallel orientation of the cell membranes. Type 3 white matter tracts cannot be visualized.
In the infants with fiber abnormalities, we observed disturbed fiber patterns. We must remark that when fibers are missed during fiber tracking, it does not mean that the white matter fibers are disconnected. This is, however, an indication that processes that have disturbed the diffusion have occurred.
It appears that the minor white matter abnormalities seen with fiber tracking (ie, asymmetry between hemispheres in a part of a traced fiber structure, the IFO/OR in particular) in our study tended to resolve at 3 months, while marked changes such as large fiber asymmetry persisted. All four patients who had large fiber abnormalities in the CR at 3 months had a poor outcome (ie, moderate to severe motor problems).
Several study investigators have reported that patients with hemiparesis have decreased anisotropy in the CR and the pyramidal tracts, which has been associated with loss of myelination and wallerian degeneration (20,34). In our study, the loss of fibers in the CR at birth was also visible at 3 months in four of the five patients in whom disturbed CR fibers were visible on the fiber-tracking images. In two of these infants, this fiber loss was associated with wallerian degeneration.
In a previous study, ischemia was observed to result in increased anisotropy in lesions in newborns (34). In two infants with ischemia in the splenium of the CC in our study, the increased anisotropy resulted in longer fibers being traced. The implications of these observations are not clear.
One of the limitations of our study was related to the criteria used to select the reference group. Although all of the infants in this group had standard MR imaging findings and normal short-term follow-up results, we realize that the white matter in these babies could have been affected because they all were examined with MR imaging because of clinical indications. However, the general fiber pattern observed in these infants was similar, with no asymmetry; therefore, we assumed that this group could be used to obtain reference values.
Quantification of PLIC Fibers
Because the PLIC is myelinated at birth (2,6,20), ROI selection in this area results in a clear pattern of fibers, which include the motor and somatosensory tracts (31). Quantification of the lengths and volumes of PLIC fibers appears to be useful for detecting abnormalities, although the results must be carefully interpreted because they depend on the stopping criteria used. In most cases, the abnormality is directly visible owing to the asymmetry seen on the fiber-tracking images; however, in some patients the abnormality is less clear. Furthermore, quantification could be useful for identifying patients with global myelination delays in both hemispheres. The implications of detecting this type of abnormality for further neurologic development of the infant are not yet clear and require further study.
Fiber-tracking Methods and Study Limitations
Volume tracing enables user-independent fiber tracking. The advantages of using this method are that all fibers are traced and no fibers are missed as a result of ROI misplacement. Furthermore, if fibers appear to be disrupted, tracking from a single ROI will result in only part of the fibers being traced. In contrast, with use of volume tracing, all fibers, if present, will be detected as long as they display anisotropic diffusion. The main problem with the volume-tracing method is that typical diffusion-tensor and echo-planar imaging artifacts, such as image distortions due to eddy currents and air-tissue susceptibility effects, can result in erroneously large anisotropy values. The large anisotropy values result in what we call fiber artifacts, which can clutter the image. ROI-based tracing results in images with fewer artifacts, but it requires prior knowledge about the structures in the data set for correct ROI placement. By first using volume tracing to determine the presence of fiber structures and then performing ROI placement on the basis of the knowledge obtained with volume tracing, the best aspects of both techniques can be used.
One of the main limitations of all currently available fiber-tracking methods—this limitation was reported not only in our current study but also in other investigations (13)—is that the tracking is stopped in regions with planar anisotropy because no main direction of diffusion can be determined in these regions. With ROI-based tracing in particular, fibers might be missed owing to this limitation.
It must be noted that the fiber-tracking results reported herein should be carefully interpreted. First, the technique is not yet validated; however, in the myocardium, a good correlation between the direction of the first eigenvector and the histologic findings has been shown (35–37). In terms of white matter, good correspondence between the fiber-tracking results and the findings on anatomy atlases and/or in dissected anatomic sections has been observed (13,26,38).
Second, the fiber-tracking technique is susceptible to several errors, as extensively discussed in the articles of Basser et al (39), Tournier et al (40), and Mori and van Zijl (9). In our clinical practice, we use a 1.0-T MR system and relatively large voxels, in which partial volume effects occur. Furthermore, anisotropy is lower in the neonatal brain than in the adult brain. The inherently lower signal-to-noise ratio and lower anisotropy affect the accuracy of fiber tracking. One problem that can occur owing to the low signal-to-noise ratio is tract jumping, in which the fiber path is determined erroneously (39). Furthermore, the anisotropy in a voxel can be obscured owing to partial volume effects. Tournier et al, by using simulations on synthetic data, showed that to identify small fibers, the signal-to-noise ratio needs to be high (40). This is probably the reason that we cannot distinguish small fiber tracts that are close to each other, like the fornix and the anterior commissure. Although several optimization steps still have to be taken, the main fiber tracts in the neonatal brain can be identified by using fiber tracking. Furthermore, the topologic features of the identified fibers correspond to expected white matter structures.
In conclusion, we have demonstrated the feasibility of fiber tracking—a user-independent volume-tracing method in particular—to study the neonatal brain. Abnormalities related to perinatal hypoxic ischemia resulted in fiber patterns that were different from the fiber structures detected in infants with standard MR imaging findings. Fiber tract abnormalities occurred frequently in the CR; this was expected because a high incidence of severe motor problems in asphyxiated newborns has been reported (1,21,22). With use of a quantification fiber-tracking method, an abnormality could be clearly seen as a difference between the hemispheres. Furthermore, significant differences in fiber lengths and volumes were observed between the infants with abnormalities in the basal ganglia and those with standard MR imaging findings. Further investigations are necessary to determine whether fiber abnormalities correlate with neonatal patient outcomes; our patient group was too small for us to make such conclusions.
|
FOOTNOTES |
|---|
Abbreviations: ADC = apparent diffusion coefficient • CC = corpus callosum • Cl = linear anisotropy index • CR = corona radiata • IFO/OR = inferior fronto-occipital fasciculus and/or optic radiation • PLIC = posterior limb of internal capsule • ROI = region of interest
Authors stated no financial relationship to disclose.
Author contributions: Guarantor of integrity of entire study, P.F.F.W.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, C.v.P., J.B., A.V.; clinical studies, C.v.P., J.B., A.V., F.G.R.; statistical analysis, C.v.P., P.F.F.W.; and manuscript editing, C.v.P., J.B., A.V.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  J.I. Berman, H.C. Glass, S.P. Miller, P. Mukherjee, D.M. Ferriero, A.J. Barkovich, D.B. Vigneron, and R.G. Henry Quantitative Fiber Tracking Analysis of the Optic Radiation Correlated with Visual Performance in Premature Newborns AJNR Am. J. Neuroradiol., January 1, 2009; 30(1): 120 - 124. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. J. Korzeniewski, G. Birbeck, M. C. DeLano, M. J. Potchen, and N. Paneth A Systematic Review of Neuroimaging for Cerebral Palsy J Child Neurol, February 1, 2008; 23(2): 216 - 227. [Abstract] [PDF]
|
|
|
|
|
|
J.H. Gilmore, W. Lin, I. Corouge, Y.S.K. Vetsa, J.K. Smith, C. Kang, H. Gu, R.M. Hamer, J.A. Lieberman, and G. Gerig Early Postnatal Development of Corpus Callosum and Corticospinal White Matter Assessed with Quantitative Tractography AJNR Am. J. Neuroradiol., October 1, 2007; 28(9): 1789 - 1795. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
= left hemisphere, 
= right hemisphere.
