| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Neuroradiology |
1 Departments of Neurology (N.G., J.M.G., P.B.B., R.A.K.)
2 Radiology (R.M.S., E.M.S., J.P.W.), Henry Ford Hospital and Health Sciences Center, K-11, 2799 W Grand Blvd, Detroit, MI 48202.
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: Six healthy adults aged 19–42 years underwent thin-section gradient-echo sampling of free induction decay and echo magnetic resonance (MR) imaging at 3.0 T. Imaging covered the mesencephalon and basal ganglia.
RESULTS: Relaxation rates (mean ± SD) were highest in globus pallidus (R2 = 25.8 seconds-1 ± 1.1, R2' = 12.0 seconds-1 ± 2.1) and lowest in prefrontal cortex (R2 = 14.4 seconds-1 ± 1.8, R2' = 3.4 seconds-1 ± 1.1). Frontal white matter measurements were as follows: R2 = 18.0 seconds-1 ± 1.2 and R2' = 3.9 seconds-1 ± 1.2. For gray matter, both R2 and R2' showed a strong correlation (r = 0.92, P < .001 and r = 0.90, P < .001, respectively) with [Fe]. Although the slopes of the regression lines for R2' versus [Fe] and for R2 versus [Fe] were similar, the iron-independent component of R2' (2.2 seconds-1 ± 0.6), the value when [Fe] = 0, was much less than that of R2 (12.7 seconds-1 ± 0.7).
CONCLUSION: The small iron-independent component of R2', as compared with that of R2, is consistent with the hypothesis that R2' has higher iron-related specificity.
Index terms: Brain, iron, 10.919 • Brain, MR, 13.121411, 13.121412, 13.12146, 14.121411, 14.121412, 14.12146 • Iron, 131.99, 14.99 • Magnetic resonance (MR), high-field-strength imaging, 13.121411, 13.121412, 13.12146, 14.121411, 14.121412, 14.12146 • Magnetic resonance (MR), relaxometry, 13.12146, 14.12146
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Assessment of brain iron has typically involved the measurement of the proton transverse relaxation rates, R2 (1/T2) or R2' (1/T2'). Transverse relaxation is thought to be influenced by inhomogeneities in the local magnetic field resulting from the presence of iron, particularly that stored as ferritin (2), although it is possible that other iron-containing substances may contribute to this relaxation.
In brain tissue, almost all iron other than that bound to hemoglobin in blood is nonheme iron (3). Levels of nonheme iron are not uniform throughout the brain. They are typically highest in the globus pallidus, substantia nigra, and red nucleus (4). The majority of nonheme iron is protein bound, and roughly one-third of all nonheme iron in brain may be in the form of ferritin (4), although this fraction may be much higher in certain high-iron, deep GM regions such as globus pallidus (2).
In studies limited to GM in healthy subjects (humans and/or other primates), several authors (5–8) have observed a relationship between R2 and regional nonheme iron levels. However, R2 alone is not a specific indicator of iron levels, particularly when pathologic tissue is involved, because it has contributions from iron-independent processes, such as tissue water content, in addition to the iron-related processes (9,10). Both of these may vary between tissue types (ie, GM vs white matter [WM]), between normal and diseased tissue, and, possibly, with age.
Methods to increase the specificity of R2 measurements have been developed. One such technique, which involves measurements at two magnetic field strengths, makes use of the increase in the iron-related contribution to R2 with increasing field strength (9–12). In a second method, performed at a single field strength, the difference between R2 measured with short and long interecho times, with a multiecho sequence, is used as an index of tissue iron content (13–15). This technique requires three acquisitions to obtain a single-section measurement. A third strategy (16,17) involves high-field-strength (eg, 3.0-T) measurements of R2' (R2' = R2* - R2). Only the latter method has demonstrated the ability to yield measurements from the substantia nigra. This region, which is difficult to study because of its small size, plays an important role in the neuropathology of Parkinson disease. Although measurements of R2' (and R2) from the substantia nigra were originally based on a single section requiring two acquisitions (16,17), multisection measurements can be performed with a single acquisition by using the gradient-echo sampling of free induction decay and echo (GESFIDE) sequence (18). We have applied this sequence to measure the transverse relaxation rates from several anatomic regions in healthy subjects as a preliminary step for its use in the study of neurodegenerative disorders.
In this work, we report high-field-strength (3.0-T) transverse relaxation rates (R2' and R2) for several anatomic regions of human brain in vivo. Prior publications (16,17) have provided transverse relaxation rates at 3.0 T for the substantia nigra. The purpose of our study was to use thin, contiguous sections obtained at MR imaging to determine the transverse relaxation rates from several GM regions (substantia nigra, red nucleus, globus pallidus, putamen, head of the caudate nucleus [hereafter, caudate head], prefrontal cortex) and from frontal WM in a group of six healthy young adults. In addition, we sought to determine the relationship between these transverse relaxation rates and regional levels of nonheme iron (according to subject age and anatomic region) obtained from a previous report (4) of postmortem iron measurements. This provides further insight into the iron-related sensitivity and specificity of these measures.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Imaging Sequence and Protocol
The GESFIDE sequence (18) provides the data required to construct multisection maps of R2*, R2, and R2' in a single acquisition. This sequence uses two section-selective radio-frequency pulses with tip angles that are typically, although not necessarily, 90° and 180°. An oscillating read gradient produces a train of gradient echoes between the 90° and 180° pulses and a further set of echoes after the 180° pulse. Phase encoding occurs before the first set of echoes. The decay shape of the transverse magnetization during the first set of echoes has the form exp[-(R2 + R2')t]. This becomes exp[-(R2 - R2')t] during the second set, because the 180° pulse reverses only the R2' decay. After R2 + R2' R2* and R2 - R2' R2- have been measured, the rates R2 and R2' can be obtained.
An important property of the GESFIDE sequence is that, in principle, it can be performed with two radio-frequency pulses that have an arbitrary flip angle (18), provided that sufficiently strong crusher gradients are applied around the refocusing pulse. Therefore, this method is insensitive to spatial variations in flip angle (eg, nonuniform radio-frequency magnetic field strength).
For our study, five gradient echoes, with a first-echo echo time of 9 msec and an interecho spacing of 8.4 msec, were acquired prior to the 180° pulse. The echo train after the 180° pulse contained six echoes separated by 8.5 msec. With this timing, the final echo produced a spin-echo image (echo time = 98 msec), that is, the time between the 180° pulse and the final echo was equal to that between the 90° and 180° pulses. Echoes were acquired only during one of the two polarities of the read gradient (as in reference 18), so that susceptibility-induced shifts in the read direction were in the same direction for all echoes. The duration of each of the "opposite polarity" read-gradient lobes, that is, the lobes during which data were not acquired, was 2.3 msec, which represents less than 30% of the total interecho time.
Sixteen thin (2.2-mm), contiguous sections that extended from the pontomedullary border to slightly above the superior border of the putamen were acquired. With the aid of a sagittal locator image, the sections were oriented parallel to a line that joined the anterior and posterior commissures. By using a 128 x 128 matrix and a 220-mm field of view, almost isotropic voxel dimensions (1.7 x 1.7 x 2.2 mm) were obtained. To avoid section interference, even and odd sections were obtained with separate acquisitions. The total imaging time with a repetition time of 1,500 msec was 6.4 minutes.
To minimize pulsatility artifacts associated with inflowing arterial blood in the lower mesencephalon (ie, midbrain), spatial presaturation pulses were applied just prior to the 90° radio-frequency pulse for each image section (approximately every 190 msec). We did not expect these arterial pulsatility artifacts to influence our measurements, because the ghost bands were displaced from our regions of interest (Fig 1). Nevertheless, presaturation was included as a precaution. The presaturated region was a 3.0-cm-thick slab offset inferiorly to leave a 12-mm gap between this slab and the most inferior image section.
|
|
|
For each subject, relaxation rates for the substantia nigra (hypointense region exclusively) (16), red nucleus, globus pallidus, putamen, and caudate head were obtained from multiple-section regions of interest (ROIs). The ROIs were outlined on the spin-echo images (final echoes in GESFIDE sequence) and were then transferred to the relaxation rate maps. For each ROI, the mean values of R2 and R2'—that is, the mean across all pixels in the ROI—were determined. Transverse relaxation rates (R2 and R2') for left-side and right-side ROIs were averaged together, with each side weighted according to the relative number of pixels in the ROI for that side. Although the 1,500-msec repetition time was shorter than that required for pure T2 weighting, contrast on the spin-echo images was sufficient to choose the ROIs (Fig 1).
The substantia nigra, globus pallidus, and putamen were clearly observed on four or five sections per subject, and the red nucleus typically was covered by two sections. Measurements for the caudate head were derived from three to four sections in which the border between the caudate head and putamen could be clearly distinguished. These multiple-section ROIs contained roughly 600 pixels for the globus pallidus and putamen, 200 pixels for substantia nigra and caudate head, and 100 pixels for the red nucleus. Measurements for frontal WM were obtained from an ROI (approximately 300 pixels) constructed on a single section containing the globus pallidus. For the prefrontal cortex, ROIs (approximately 30 pixels) were constructed on the most superior section to avoid susceptibility-related artifacts observed near the pre-frontal cortex regions of the lower sections.
The ROIs for each section were constructed by a single image analyst (R.M.S.) by using software known as EIGENTOOL, developed by the Division of Physics and Engineering, Image Analysis Laboratory, of our institution. The initial step in generating an ROI for the substantia nigra or red nucleus was the selection by the analyst of a group of approximately 5 pixels from within the darkest area of the GM structure. Next, the "automatic-fill" function of EIGENTOOL was used to extend the group of pixels to include all other contiguous pixels falling within 3 SDs of the mean signal amplitude of the original 5 pixels. The automatic-fill function was then applied iteratively until the size of the group did not increase further, indicating that all pixels bounded by a relatively steep border had been chosen. With each iteration, a new SD was calculated by using all pixels in the most recent group. The ROIs were then reduced in size by two pixels in every direction so that a dark ring of the structure encircled the ROI. This ensured that anatomic borders were not crossed. The automatic-fill procedure could not be applied to the globus pallidus, putamen, and caudate head, because the changes in signal amplitude at the borders were not sufficiently steep. In these cases, borders were delineated by the analyst and then reduced in size by two pixels in every direction. Finally, ROIs were inspected for accuracy by a neuroradiologist (E.M.S.).
Estimation of Regional Iron Levels
We investigated the relationship between our measured transverse relaxation rates and regional nonheme iron concentrations obtained previously (4) from postmortem iron measurements in 98 brains of varying ages. For each of our subjects, we estimated the nonheme iron concentration [Fe] (in milligrams of iron per gram fresh weight) according to subject age A for the globus pallidus, putamen, caudate head, prefrontal cortex, and frontal WM by using the following empiric formula (reported in reference 4):
, as well as mean nonheme iron concentration values (across all subjects) for several anatomic regions, are provided in reference 4. (Standard errors were approximately plus or minus 3 x 10-2 mg Fe per gram fresh weight for the globus pallidus and putamen, representing approximately 15% and 25%, respectively, of the nonheme iron concentrations at age 30 years.) For the substantia nigra and red nucleus, values of these parameters were not reported in reference 4, probably due to the large amount of scatter. However, the authors (4) did state that for the substantia nigra and red nucleus, nonheme iron concentrations increased rapidly during the first 20 years of life. Because this rapid increase also occurred for the globus pallidus, we estimated the iron concentrations for the red nucleus and substantia nigra by using an value obtained for the globus pallidus, but we scaled C0 and C1 to account for the lower mean concentration in the substantia nigra and red nucleus compared with that in the globus pallidus. This should be a reasonable approximation, because our subjects (age, 
19 years) would likely be past the rapidly increasing part of the iron concentration versus age curve. Statistical analysis was accomplished by using a linear regression to determine the relationship between the transverse relaxation rates and iron levels. A paired Student t test was used to compare mean values of the relaxation rates for frontal WM versus prefrontal cortical GM.
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
cm in the section-selective direction. Representative maps of R2*, R2, and R2' for two sections, one through the midbrain (Fig 1b) and the other at the level of the globus pallidus (Fig 1c), are illustrated in Figure 2. High values within the substantia nigra, red nucleus (Fig 2, top) and globus pallidus (Fig 2, bottom) are visible on all three maps. Hyperintense vessel-shaped regions (Fig 2, top), likely produced by venous deoxyhemoglobin and perhaps by flow-related signal loss, can also be observed. The high-signal-intensity region located near the anterior edge of the R2* and R2' maps (Fig 2, top) is caused by global magnetic field inhomogeneities associated with the close proximity of the sphenoid sinuses and nasal cavity (21). In one case, slight overlap between the artifact and the substantia nigra occurred. In a few cases, the susceptibility-related artifact could also be observed on R2' maps at the level of the bottom row images in Figure 2, and this artifact overlapped slightly with the caudate head.
|
|
|
|
) were found to be small (/SD 
-0.7 for R20' and less for others). The correlation coefficient increased slightly to 0.92 for R2' and decreased slightly to 0.91 for R2.
For R2', the regression line appears to represent the frontal WM data as well, whereas the R2 values for frontal WM tend to lie above the corresponding regression line. (The deviation of the mean frontal WM R2 value from the line is 1.7 times the SD of the frontal WM values given in Table 2.) Although the estimated iron concentrations for frontal WM are close to those for the prefrontal cortex, a paired Student t test showed significant differences (P < .001) between the R2 values for these two regions. This difference remained significant (P = .001) even after the prefrontal cortex values were corrected to account for the small difference in estimated iron concentration between the two regions ([Fe] = 0.013 mg Fe/g fresh wt). This correction, based on Equation (3), was accomplished by adding a small amount (m x [Fe] = 0.8 second-1) to each of the R2 values for the prefrontal cortex, where the value of m was obtained from Table 1. A similar procedure for R2' showed no significant differences between prefrontal cortex and frontal WM.
Representative signal decay curves for the substantia nigra and frontal WM are illustrated in Figure 4. The 11 data points correspond to the 11 echoes, each representing the mean signal amplitudes from a small single-section region in the left substantia nigra (approximately 40 pixels after zero-filling) and frontal WM (approximately 40 pixels) in one subject. The reversal of the R2' decay, represented by the difference in slope between each dotted line and the corresponding dashed line, is easily observed, particularly for the substantia nigra. Note also that whereas the R2* decay for the substantia nigra is much steeper than that for frontal WM, the R2- decays have similar slopes. This is consistent with the substantia nigra and frontal WM data in Table 2; that is, both R2 and R2' are increased in the substantia nigra compared to frontal WM, so that the values of the quantity R2 - R2' are similar for these two regions.
|
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
A comparison of the relaxation rates for frontal WM versus GM suggests that R2' may be less sensitive to differences between GM and WM than is R2 as measured at 3.0 T. A significant difference between the R2 values for frontal WM and those for the prefrontal cortex was observed, even when differences in estimated iron levels were accounted for. This exemplifies the lack of iron-related specificity of R2. On the other hand, for R2', no significant difference between frontal WM and prefrontal cortex was found. Bartzokis et al (9) found that their field-dependent measurement was highly correlated with estimates of iron content for four anatomic brain regions that included both GM and WM, suggesting that it is also insensitive to differences between GM and WM.
The higher iron-related specificity of R2' as compared with R2 has been demonstrated in previous studies (16,17) with high-field-strength (3.0-T) MR imaging of the substantia nigra in patients with Parkinson disease and age-matched control subjects. In patients with Parkinson disease, who are expected to have elevated iron levels in the substantia nigra (1), R2' was significantly greater than that in control subjects (P < .001). On the other hand, R2 was greater in the control group (P = .046), which suggests that the effect of changes other than increased iron may have outweighed the increased iron-related contributions to R2.
Comparison with Previous 3.0-T Data from Substantia Nigra
The values of R2 and R2' for the substantia nigra given in Table 2 (R2 = 24.1 seconds-1 ± 2.1, R2' = 11.6 seconds-1 ± 2.1) are comparable to those obtained previously (16) from single-section measurements in healthy control subjects as measured at 3.0 T (R2 = 29.3 seconds-1 ± 4.0, R2' = 8.2 seconds-1 ± 4.0). Relaxation rates for left and right substantia nigra were provided (16), but we have quoted the average of these values, since left versus right differences for normal subjects were not significant. We would not expect large age-related differences between relaxation rates obtained in the subjects in these earlier studies (mean age, 60 years ± 8.7) and the relaxation rates in the subjects in our current study (mean age, 30 years ± 8) because, for the substantia nigra, age-related increases in iron concentration (4,22) and R2 as measured at 1.5 T (22) appear to level off in early adulthood.
Comparison of R2 Measured at 3.0 T with R2 Measured at Other Field Strengths
Figure 5 illustrates a comparison of the iron-dependent slope of R2 (m in Eq [2]), which we obtained at 3.0 T (Table 1) versus previously measured (7) values obtained at 0.5 and 1.5 T. The lower-field-strength estimates of the iron-dependent variation of R2 provided by Vymazal et al (7) were obtained not by direct measurements of R2 but by measuring the ratio of signal intensities in several GM regions to that of WM on a large number of T2-weighted images (echo time of 100 [0.5 T] and 80 msec [1.5 T]) in healthy volunteers. Vymazal et al stated that error estimates from their nonlinear fit (shown by the error bars in Fig 5) are probably underestimates. Iron estimates used by Vymazal et al were also obtained from the work of Hallgren and Sourander (4). The in vivo measurements at three field strengths (Fig 5) are roughly consistent with those from in vitro studies of GM samples at 1.5 T and less, which show a linear field-dependent increase in R2 (6,23) and, more specifically, its iron-dependent component (6). Because the R2' and R2 decays both depend on variations in the local magnetic field strength, the iron-related sensitivity of R2' should also increase with field strength; that is, the R2' versus iron concentration slope (Table 1) would probably be less at lower field strengths.
|
|
|
Limitations of Measurements
A potential problem with the use of R2' as a measure of iron level is that this rate can be influenced by global magnetic field inhomogeneities. In particular, the region of high magnetic field distortion superior to the sphenoid sinus and nasal cavity can, in some cases, be in proximity to regions where a measure of iron levels may be desired. In the present study, this region of field inhomogeneity typically was well separated from most ROIs except for those in the caudate head. However, we found that omission of the caudate head measurements from the regression of R2' versus iron concentration did not significantly affect our results.
The contribution to R2' from global magnetic field inhomogeneities typically decreases with voxel size (26). Minimization of the section thickness, which often represents the largest voxel dimension, is of particular importance. In this study, relatively thin (2.2-mm) sections, with a 128 x 128 in-plane matrix (1.7 x 1.7-mm pixels), provided a signal-to-noise ratio that was sufficient for the easy detection of interregional variations in R2 and R2'. More work would be needed to determine whether further decreases in voxel size could be tolerated. It may be possible to acquire very thin sections and to average sections together to partially compensate for the reduced signal-to-noise ratio. This would be similar, in concept, to averaging of neighboring, thin, two-dimensional sections of three-dimensional gradient-echo images, which has been used to reduce susceptibility-related signal loss (27). As an alternative, a method of using the section-refocusing gradient to compensate for the dephasing associated with global magnetic field inhomogeneities (17,28) could be applied. However, the trade-off is a substantial increase in imaging time (eg, fourfold for the demonstration in reference 17). The susceptibility-related signal loss can also be reduced using quadratic phase radio-frequency pulses (29), but this method produces lower signal intensity from voxels in which the magnetic field is uniform. At lower field strengths (eg, 1.5 T), R2' would be less sensitive to susceptibility-related signal loss; however, as mentioned earlier, it likely would also be less sensitive to iron levels.
As demonstrated in Figure 2 (top), elevated transverse relaxation rates can occur in blood vessels. Fast transverse relaxation in vessels may be due to deoxyhemoglobin (venous blood), flow-related signal loss, or both. High R2' values in blood vessels may interfere with the accuracy of assessment of nonheme iron levels if these vessels are in proximity to the ROI.
In this work, we studied the relationship between transverse relaxation rates and nonheme iron levels. Tissue relaxation rates may also be influenced, to some extent, by heme iron in deoxyhemoglobin. Recently, Punwani et al (30) reported a linear variation of R2' with deoxyhemoglobin levels, corresponding to roughly 3 seconds-1 per 10% oxygen saturation change, in the GM of the parietal cortex during hypoxia in piglets at 7.0 T. However, under normal oxygen conditions, the mean R2' for GM in these piglets was too small to be considered significantly different from 0. Changes in R2 with oxygenation level were roughly 10 times smaller than for R2'. In human brain at 3.0 T, it is unlikely that deoxyhemoglobin plays an important role in determining interregional differences in R2' under normoxic conditions.
The transverse relaxation rates reported here were obtained by using a relatively long spin-echo echo time (98 msec). We expect that R2 would increase, to some extent, with increasing echo time, analogous to the variation of R2 with interecho time 
in a multiecho sequence (6,13,14). However, this variation, which increases with iron content, would probably be small over the range of echo times practical for the GESFIDE sequence. For example, a 10-fold decrease in produced a 2.4-second-1 decrease in the value of R2 for the globus pallidus at 1.5 T (14). This is equivalent to a decrease of only approximately 0.7 second-1 for a two-fold decrease in , if one assumes a logarithmic relation between R2 and , as shown in reference 13. At 3.0 T, the variation of R2 with would probably be larger than that at 1.5 T by a factor of less than two, because this variation is a measure of the iron-related contribution to R2 (13,14), which increases linearly with field strength (6). Finally, R2', which is a measure of the amount of reversible decay, may also vary with echo time, but in the opposite direction from R2, and this would also be a relatively small effect.
In the analysis of decay rates, we have approximated both the reversible and irreversible decay shapes by using exponential functions. On the basis of the exponential shapes of the curves in Figure 4, it appears that this approximation is sufficient to provide a useful quantitative measure of the decay rates. In general, the irreversible decay (ie, R2) for tissue may not be a single exponential function. For example, in WM the R2 decay may be represented by a spectrum of components (31). The actual shape of the reversible decay (ie, R2') would likely depend on factors such as the spatial distribution of magnetic particles (eg, ferritin molecules) producing the local magnetic field variations and the correlation times of fluctuations in these local fields. A theoretic analysis does predict that a random distribution of very strongly magnetic particles should produce an exponential decay, other than at very short times, if diffusion effects are ignored (32).
With the GESFIDE sequence, the image sections along the R2* decay are excited by a single 90° radio-frequency pulse, whereas the R2- images are exposed to a 90° pulse, as well as a 180° pulse. As a consequence, the section profiles of these two sets of images are not identical (33), and hence the measurements of R2* and R2- are based on slightly different sensitivity functions in the section-selective direction. This could contribute to errors the in determination of R2 and R2' near sharp borders, that is, where these parameters vary over distances that are short compared with the section thickness. This problem should be less important with thin sections as compared with thick sections. A modification of the GESFIDE method (33) involves sampling the R2* decay after the spin echo rather than after the 90° radio-frequency pulse, so that the R2* and R2- decays are acquired from the same section profile. The trade-off is a substantial reduction in the signal-to-noise ratio. For example, if the spin-echo echo time is approximately equal to T2, as suggested by Yablonskiy and Haacke (33), then the spin echo will have decayed by a factor of exp(-1) 0.37. On the other hand, the first gradient echo of the GESFIDE sequence, which occurs at a time TE1 after the 90° pulse, will have decayed by a factor exp(-TE1/T2*), which, in the present case (TE1 = 9 msec, T2* 25 msec), is not less than 0.7.
Implications for Studies of Neurodegenerative Disorders
Increased levels of brain iron appear to be involved in the pathology of several neurodegenerative disorders, including Parkinson disease (1). The ability to specifically image and to quantitatively assess regional brain iron content in human subjects noninvasively and longitudinally over time has potential use in (a) studying the pathogenesis, (b) monitoring the progression, (c) evaluating putative neuroprotective treatments, and (d) aiding in the differential diagnosis of these conditions.
The present work offers a clinically practical method to quantify brain iron on the basis of measurements of the transverse relaxation rate R2'. We have demonstrated the feasibility of performing high-resolution measurements of R2', as well as R2, from several anatomic brain regions with a reasonably short imaging time (<7 minutes). Evidence in support of the idea that R2' is a more specific index of brain iron deposition than is R2 has been provided. If one assumes that the field dependence of R2' is similar to that of R2, then measurements obtained at 3.0 T should be more sensitive to brain iron levels than those obtained at 1.5 T.
Finally, the study of neurodegenerative disorders, as well as the optimization of many high-field-strength (3.0-T) imaging procedures, in general, will require a knowledge of normative relaxation rates at this field strength. Our measurements of R2' and R2 for several anatomic regions in healthy young adults should provide a contribution to this goal. Future studies should be undertaken to investigate age-related changes in the high-field-strength transverse relaxation rates.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
J.M.G. supported by grant NS33341 from the National Institute of Neurological Disorders and Stroke.
Address reprint requests to J.M.G.
From the 1997 RSNA scientific assembly.
Abbreviations: GESFIDE = gradient-echo sampling of free induction decay and echo GM = gray matter ROI = region of interest WM = white matter
Author contributions: Guarantors of integrity of entire study, J.M.G., N.G.; study concepts, N.G., J.M.G., P.B.B.; study design, N.G., J.M.G., P.B.B., J.P.W., R.A.K.; definition of intellectual content, N.G., P.B.B., J.M.G.; literature research, N.G., P.B.B., J.M.G.; clinical studies, N.G., P.B.B., J.M.G.; experimental studies, N.G., P.B.B., J.M.G.; data acquisition, N.G.; data analysis, R.M.S., E.M.S., J.P.W., J.M.G., N.G.; statistical analysis, N.G.; manuscript preparation, N.G.; manuscript editing, N.G., P.B.B., J.M.G., R.A.K., E.M.S., J.P.W., R.M.S.; manuscript review, N.G., P.B.B., J.M.G., R.A.K., E.M.S., J.P.W.
Received November 7, 1997;
revision requested January 7, 1998; revision received July 22, 1998;
accepted September 9, 1998.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. McNeill, D. Birchall, S. J. Hayflick, A. Gregory, J. F. Schenk, E. A. Zimmerman, H. Shang, H. Miyajima, and P. F. Chinnery T2* and FSE MRI distinguishes four subtypes of neurodegeneration with brain iron accumulation Neurology, April 29, 2008; 70(18): 1614 - 1619. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. R. W. Martin, M. Wieler, and M. Gee Midbrain iron content in early Parkinson disease: A potential biomarker of disease status Neurology, April 15, 2008; 70(16_Part_2): 1411 - 1417. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S.L. Harder, K.M. Hopp, H. Ward, H. Neglio, J. Gitlin, and D. Kido Mineralization of the Deep Gray Matter with Age: A Retrospective Review with Susceptibility-Weighted MR Imaging AJNR Am. J. Neuroradiol., January 1, 2008; 29(1): 176 - 183. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Elolf, V. Bockermann, T. Gringel, M. Knauth, P. Dechent, and G. Helms Improved Visibility of the Subthalamic Nucleus on High-Resolution Stereotactic MR Imaging by Added Susceptibility (T2*) Contrast Using Multiple Gradient Echoes AJNR Am. J. Neuroradiol., June 1, 2007; 28(6): 1093 - 1094. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. E. Gamaldo and C. J. Earley Restless legs syndrome: a clinical update. Chest, November 1, 2006; 130(5): 1596 - 1604. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Ramani, J. H. Jensen, and J. A. Helpern Quantitative MR Imaging in Alzheimer Disease Radiology, October 1, 2006; 241(1): 26 - 44. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M.J. House, T.G. St. Pierre, J.K. Foster, R.N. Martins, and R. Clarnette Quantitative MR Imaging R2 Relaxometry in Elderly Participants Reporting Memory Loss. AJNR Am. J. Neuroradiol., February 1, 2006; 27(2): 430 - 439. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L.-A. Williams, N. Gelman, P. A. Picot, D. S. Lee, J. R. Ewing, V. K. Han, and R. T. Thompson Neonatal Brain: Regional Variability of in Vivo MR Imaging Relaxation Rates at 3.0 T--Initial Experience Radiology, May 1, 2005; 235(2): 595 - 603. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. S. Georgiades, R. Itoh, X. Golay, P. C. M. van Zijl, and E. R. Melhem MR Imaging of the Human Brain at 1.5 T: Regional Variations in Transverse Relaxation Rates in the Cerebral Cortex AJNR Am. J. Neuroradiol., October 1, 2001; 22(9): 1732 - 1737. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R.P. Allen, P.B. Barker, F. Wehrl, H.K. Song, and C.J. Earley MRI measurement of brain iron in patients with restless legs syndrome Neurology, January 23, 2001; 56(2): 263 - 265. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||
