| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Neuroradiology |
1 From the Magnetic Resonance Institute (R.B., R.S., J.S., H.O., S.R., F.E., F.F.) and the Depts of Neurology (R.B., M.A., H.O., S.R., P.K., F.F.) and Radiology (J.S., F.E.), University of Graz, Auenbruggerplatz 9, A-8036 Graz, Austria; Philips Medical Systems, Hamburg, Germany (H.K.); and the Institute of Electrical and Biomedical Engineering, Graz University of Technology (P.W.). Received Feb 5, 1998; revision requested Mar 25; revision received Sep 29; accepted Nov 5. Supported in part by grants from Gemeinnützige Hertie Stiftung (R.B., S.R.) and the Faculty of Electrical Engineering, Graz University of Technology (R.B.). Address reprint requests to R.S.
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: Apparent diffusion coefficient (ADC) measurements were performed in phantoms, and six healthy adult volunteers were examined to determine intrasubject (precision) and intersubject (reference range) variations in absolute ADC and relative ADC (rADC) measurements. DW imaging maps and lesion rADC values were also obtained in 34 consecutive stroke patients to evaluate the sensitivity and reliability of DW-interleaved echo-planar imaging for detection of ischemic brain damage.
RESULTS: Phantom and volunteer ADC values were in excellent agreement with published data. The intrasubject variation of rADC was 6.2%. The ADC precision ranged from 6.5% in the subcortical white matter in the frontal lobe to 12.9% in the head of the caudate nucleus. Interleaved echo-planar imaging enabled rapid acquisition of high-quality images of the entire brain without substantial artifacts. Within the 1st week, the sensitivity of DW-interleaved echo-planar imaging for detection of acute infarction was 90% (18 of 20 true-positive studies) and independent of lesion location.
CONCLUSION: DW-interleaved echo-planar imaging with phase navigation and cardiac triggering is robust, reliable, and fast. With high sensitivity for detection of early ischemic infarction, it is useful for examining stroke patients by using MR systems with conventional gradient hardware.
Index terms: Brain, infarction, 13.78 • Brain, MR, 13.121412, 13.121416, 13.12146 • Magnetic resonance (MR), diffusion study, 13.121412, 13.121416, 13.12146 • Magnetic resonance (MR), echo planar, 13.121416 • Magnetic resonance (MR), tissue characterization, 13.121412, 13.121416, 13.12146
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Several problems may limit the widespread clinical use of DW imaging. Because of the inherent motion sensitivity of DW imaging, a patient's involuntary head motion or brain pulsation (10,11) can degrade images substantially. When the standard two-dimensional Fourier reconstruction method is used, severe ghost artifacts that render the DW image useless can occur because the phase changes that are introduced may be different for each particular phase-encoding step. In addition to mechanical restraints, different methods have been proposed to overcome this problem. These methods include (a) single-shot (12–15), (b) line scanning (16), (c) projection reconstruction (17), (d) bipolar gradient (18), and (e) navigator-echo (19–24) techniques. The following strategies are used to avoid ghost artifacts in these methods: (a) maintaining equal phase error in each phase-encoding step, (b) data acquisition without phase encoding, (c) motion-insensitive k-space trajectories, (d) reduction of diffusion time or (e) registration of phase errors, and retrospective correction for each k line/interleaf.
The time needed for DW image acquisition has been another important issue in routine clinical practice. Prolonged examination times increase the probability of patient movement and may not be tolerable in acute stroke patients. Therefore, several studies have relied on single-shot echo-planar image acquisitions that require high-performance gradient hardware. Implementation of the DW imaging technique with conventional MR units for routine imaging often has been hampered by the lack of single-shot echo-planar capabilities of these systems or by a laborious retrofitting procedure for insert gradients.
To overcome these problems, a navigated DW-interleaved echo-planar imaging technique that enables high-quality DW imaging with a conventional gradient system has been implemented. In single-shot echo-planar imaging, the entire k space is acquired during a single evolution of the transverse magnetization. The sampling of different k-space locations at different times may cause severe off-resonance effects. These effects should be much smaller with the use of interleaved echo-planar imaging because of the increased k-space velocity in the phase-encoding direction. By using navigator echoes (21–24), the motion-induced view-to-view phase alterations between each k line/interleaf can be monitored and retrospectively corrected for motion either in k space (20) or in hybrid space (21,22). In addition, plethysmographic RR gating should be used to compensate for pulsatile brain movement, which could otherwise cause additional image degradation (22).
The purpose of this study was to demonstrate, even with conventional gradient hardware, the clinical reliability of the proposed DW imaging method in a consecutive series of patients who had had an ischemic stroke. We further wanted to test the accuracy and precision of ADC measurements by means of phantom studies and examinations in volunteers.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
is the gyromagnetic ratio; 
, the duration of each diffusion gradient; 
, the time between the onsets of the diffusion gradient pulses; G, the gradient strength; S0, the signal intensity without diffusion weighting; and b, the DW factor.
|
Because artifacts caused by the chemical shift between water- and fat-bound protons are not negligible in interleaved echo-planar imaging, a spectral presaturation pulse was used to suppress the signal intensity of adipose tissue (Fig 1). The echo time shift technique was used (26) to reduce discontinuities in phase error function due to segmented k-space acquisition (minimization of interleave ghost artifacts) (27). To refocus the magnetization during interleaved echo-planar image readout, a second 180° pulse with a short TE was used. The sum of the TEdiffusion and the TE can be interpreted as the composite TE (TEcomposite), which is relevant for T2-weighted image contrast (Fig 1). The half-Fourier imaging technique (28) was used to allow echo-planar image acquisition with a short effective TE, which leads to a slight improvement in the signal-to-noise ratio while preserving a moderate number of interleaves. Furthermore, it diminishes the adverse effect of motion between the navigator and imaging echo. Automatic shimming to cover the entire imaged volume was performed to reduce field inhomogeneity.
Calculation of the ADC
Pixel-by-pixel calculation of the ADC was performed by fitting the characteristic curve, log(S0/Sb) = ADCb, onto the DW image data, where S0 is the signal intensity obtained without diffusion weighting; Sb, the signal intensity obtained with diffusion weighting; and b, the DW factor. The parametric ADC maps were processed offline on a workstation (SPARCStation 10; SUN Microsystems, Mountain View, Calif) by using dedicated MR image processing software (29). Only those pixels that could be clearly differentiated from background noise were considered for calculation. Therefore, the selection threshold was set above 2.5 times the SD of the scaled Rayleigh distribution of the background noise from the S0 image. The trace of the diffusion tensor, Trace (D), was calculated by using the equation,
Phantom Studies
Plastic bottles filled with either water or acetone were placed in the standard circularly polarized head coil. Imaging was performed at constant room temperature by using a 23-cm field of view, 256 x 256 acquisition matrix, 70% rectangular field of view, and 75% reduced acquisition matrix, which yielded an inplane resolution of 0.89 x 1.19 mm. The additional imaging parameters were a section thickness of 6 mm, section gap of 1.5 mm, repetition time (TR) of 2,000 msec, TEcomposite of 133 msec (TEdiffusion = 115 msec, TE = 18 msec), two signals acquired, and an echo-planar imaging train length (ie, number of echoes per interleaf) of eleven. The acquisition of one interleaf took 22.9 msec by using a sampling bandwidth of ±48.9 kHz. To measure all profiles, 12 interleaves were necessary. To obtain different levels of diffusion weighting, either two or eight gradient steps were used for each phantom. With two gradient steps, the b values were approximately 0 sec/mm2 and 716 sec/mm2. The b values ranged from approximately 0 to 1,550 (0, 32, 126, 284, 505, 790, 1,137, and 1,550) sec/mm2 when eight gradient steps were used.
Human Studies
All imaging protocols were approved by our institution's committee on human studies. Informed consent was obtained, after the nature of the procedure had been fully explained, prior to performing the MR imaging procedures.
At four different times, we examined six healthy volunteers (four men, two women; age range, 25–58 years; mean age, 35 years) by using the sequence illustrated in Figure 1. DW images at two different levels of diffusion weighting were obtained in the readout, phase-encoding, and section-select directions, and the ADC was calculated for each direction. The corresponding ADC values were determined in the following anatomic regions: (a) head of caudate nucleus, (b) motor cortex at the level of the lateral ventricle, (c) subcortical white matter of the frontal lobe, (d) splenium of the corpus callosum, and (e) cerebrospinal fluid. In addition, the trace of the diffusion tensor was calculated by using Equation (2).
All studies were performed by using an acquisition matrix of 256 x 256, field of view of 24 cm, inplane resolution of 0.93 x 1.24 mm, and TR of approximately 1,500 msec (2 x RR wave). Phase encoding was performed in the left-right direction. All of the other imaging parameters were the same as those used in the phantom studies. Twenty sections were chosen to enable coverage of the entire brain, including the infratentorial compartment. For a cardiac frequency of 55 beats per minute, the imaging time was 3 minutes, 55 seconds for two b values and two acquired signals.
To test the clinical applicability of the method that we used, a consecutive series of 34 ischemic stroke patients (22 men [mean age ± SD, 66.4 years ± 13.1; age range, 41–85 years] and 12 women [mean age, 67.8 years ± 17.4; age range, 36–86 years]) were examined for 4 months. Patients with cerebral hemorrhage and other nonischemic findings such as brain tumor, herpes encephalitis, and trauma were excluded. This allowed an assessment of the diagnostic sensitivity (ie, true-positive fraction) of DW-interleaved echo-planar imaging of acute and subacute infarction. A determination of the specificity was beyond the scope of this study.
The time of onset of stroke was considered to be the time that the patient was last known to be without new neurologic deficits. Serial images were obtained in four patients. Patients were placed in the standard head holder of the circularly polarized head coil. No additional head restraints except foam pads and Velcro straps were used. The imaging protocol consisted of a T2-weighted turbo spin-echo sequence and T2-weighted cerebrospinal fluid–attenuated inversion recovery (FLAIR) turbo spin-echo sequence, as well as the described DW imaging sequence. The images obtained with each sequence were first interpreted separately, for the delineation of an acute lesion, by three independent readers (H.O., J.S., F.F.). Thereafter, lesion conspicuity was directly compared between all of the images by one of the radiologists (F.F.).
The imaging parameters for T2-weighted imaging were as follows: acquisition matrix, 256 x 256; field of view, 23 cm; reduced acquisition matrix, 80%; rectangular field of view, 80%; section thickness, 5 mm, with a 0.5-mm gap; echo train length, 17; two signals acquired; and 2,900/120 (TR msec/TE msec). For cerebrospinal FLAIR imaging, the parameters were as follows: acquisition matrix, 256 x 256; field of view, 23 cm; reduced acquisition matrix, 90%; rectangular field of view, 80%; section thickness, 6 mm, with a 1.5-mm gap; echo train length, 20; two signals acquired; 6,000/130; and inversion time, 1,900 msec. For both sequences, a linear k-space profile order was used.
In the human studies, two levels of diffusion weighting were used. First, slight diffusion weighting (ie, b 
0 sec/mm2) was used, and the resultant image contrast was more or less T2 weighted. Thereafter, a large b value (ie, 716 sec/mm2) was used, and the resultant images were interpreted to be diffusion weighted. All DW imaging examinations in humans were performed by using a maximum diffusion gradient strength of 10 mT/m, with a rise time of 900 µsec to 10 mT/m. The typical duration of the diffusion gradient was 49.4 msec, and the typical time between the onsets of the diffusion gradient pulses was 57.5 msec.
Statistical Analyses
To account for individual variations in bulk ADC values (eg, temperature and serum sodium concentration), we calculated the ratio of the ADC of the ischemic lesion (ipsilateral region) to that of the unaffected corresponding contralateral brain region (rADC) (30). For statistical analysis of ADC changes in ischemic stroke patients over time, the patients were grouped according to time of stroke onset, with the rADC as the dependent measure (6,31). Grouping was performed as follows: group A, patients examined 0–6 hours after stroke onset; group B, 7–24 hours after onset; group C, 25–96 hours after onset; group D, 97–240 hours after onset; and group E, more than 240 hours after onset. By using an rADC of 1 as the null-hypothesis value, the mean rADC values in the specific groups were assessed by using a one-sample Student t test (5,6,31). The results were considered to be significant when the P value was less than .05. Multiple observations were considered by using Bonferroni correction.
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Human Studies
The mean ADC and trace measurements obtained in different anatomic regions in the six volunteers are shown in Table 1. The highest diffusion anisotropy was found in the corpus callosum. Anisotropy of the subcortical white matter tracts was moderate because of the fibers of different tracts with different orientation merges. The intrasubject variation (± SD, expressed as a percentage of the mean value) in these measurements in the six volunteers ranged from 6.5% in the subcortical white matter tract to 12.9% in the head of the caudate nucleus. We also determined the deviation in rADC from 1 in the motor cortex and frontal white matter of these individuals. The intrasubject variation (ie, precision) in rADC was 6.3%.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
When we added the described DW imaging sequence to our conventional imaging protocol for stroke patients, approximately 4 additional minutes of examination time were required. Our former attempts to perform diffusion studies had been hampered by the lack of fast motion-insensitive sequences, limited number of sections to cover the brain, poor spatial resolution, or laborious setup for each patient when insert gradients were used.
With single-shot echo-planar imaging techniques, DW images can be obtained very rapidly, and the entire k space can be acquired with a single excitation. Therefore, when these techniques are used for DW imaging, they provide inherently reduced sensitivity to bulk motion. However, single-shot echo-planar imaging provides limited spatial resolution and yields a poor signal-to-noise ratio because of the large receiver bandwidth that is necessary. Furthermore, single-shot echo-planar imaging is prone to artifacts in regions with large susceptibility gradients, which occur at air- or bone-tissue interfaces (eg, skull base, posterior fossa, and sinuses).
By using interleaved echo-planar imaging with navigator motion correction and cardiac gating, many of these problems can be overcome. Measuring the navigator echo at the time of the spin echo of the first 180° pulse yields the optimal navigator signal, which minimizes errors in phase correction because phase noise is reduced. In general, the best navigator correction for motion (translational and rotational) can be achieved when the diffusion gradients and the navigator echo are applied in perpendicular directions. Because we acquired the navigator echo in the readout direction, diffusion weighting in the phase-encoding direction was optimal. The lack of rotational motion correction when diffusion gradients are used in the other principal directions can be overcome by acquiring a second navigator echo in the phase-encoding direction (24) or by implementing spiral navigator echoes (23). A factor that potentially limits the widespread use of spiral navigating with a conventional system might be the reason that a high-performance gradient system with time-varying gradient waveform capabilities and sophisticated methods for image reconstruction are required.
In this study, the trace of the diffusion tensor was acquired by obtaining three independent measurements with diffusion encoding in each principal direction. A sequence that can be used to acquire the trace during a single measurement with an optimized gradient pulse switching scheme, as proposed by Wong and colleagues (34), also is possible with interleaved echo-planar imaging (24). However, with a conventional gradient system, the TEcomposite has to be prolonged to approximately 170 msec to achieve identical diffusion weighting (b = 700 sec/mm2). This might critically impair the signal-to-noise ratio of the DW images and, consequently, the quality of the calculated ADC maps.
Cardiac gating in patients with cardiac arrhythmia is a potentially limiting factor of all spin-warp techniques in which the k space is not acquired in one shot. Because this can also occur with DW-interleaved echo-planar imaging, we chose the TR to be two RR intervals. With this strategy, for typical T1 values, variations of the TR caused by arrhythmia should be negligible because, with a TR of this length, spins are almost fully relaxed.
When echo-planar imaging is used, even small residual eddy currents can cause substantial artifacts, particularly in the phase-encoding direction, where the image bandwidth can be low (<10 Hz/pixel). Although image degradation (eg, shrink, shear, and shift in the phase-encoding direction) caused by gradient-induced eddy currents was less prominent with the described technique than with single-shot techniques, a reliable calculation of the tensor trace can be difficult with interleaved echo-planar imaging with fewer interleaves. Several correction strategies (35,36) have been recommended to overcome this. In our study, the maximum shift between selected anatomic landmarks on images with different DW directions was 1.8 mm.
The susceptibility artifacts known to be associated with single-shot echo-planar imaging occurred very mildly with interleaved echo-planar imaging and the used number of interleaves. Small S-shaped hyperintense artifacts in the area surrounding the temporal pole were rarely observed and did not impair image interpretation in any case. Image distortions at the base of the frontal lobe of the brain also were negligible, and the posterior fossa was always well depicted. Therefore, brain stem and cerebellar lesions were all detected with high sensitivity.
With the described DW imaging sequence, acute ischemic injury was reliably identified, even before the appearance of signal intensity changes on conventional images, except in one patient, who was examined 1.5 hours after the onset of stroke. Compared with the lesions missed on conventional images, neither acute nor subacute ischemic lesions were missed on DW images at blind readings. The characteristic temporal evolution of the ADC with the corresponding high signal intensity of ischemic lesions, primarily within the 1st week, also helped to distinguish between new and old lesions. These findings should also enable one to recognize acute but clinically occult associated ischemic lesions and provide important information about the pathophysiology of stroke. That is, these findings could further support the embolic cause of an infarct (37). The good agreement between the rADC time course observed in our sample of stroke patients and that in published data achieved by using isotropic DW imaging (21,31,38) provides further evidence of the reliability of this method.
One potential limitation of ADC maps for clinical purposes is the "pseudonormalization" period that occurs approximately 200 hours after stroke onset, that is, the time of transition from a decreased to increased rADC of ischemic tissue; the values during this period approach those of the normal brain parenchyma. During this period, the diagnostic sensitivity of the ADC is very low and must be compensated by interpreting simultaneously obtained T2-weighted images. The high precision of the rADC and good contrast-to-noise ratio of DW imaging provided by using the described technique are mandatory to keep the period during which ADC map acquisition may be noncontributory to a minimum.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Author contributions: Guarantors of integrity of entire study, R.S., F.F.; study concepts and design, R.S., F.F., R.B.; definition of intellectual content, R.S., F.F., R.B.; literature research, R.B.; clinical studies, M.A., J.S., H.O., P.K., F.E., F.F., R.B.; experimental studies, R.B., R.S.; data acquisition, R.S., S.R., H.K., R.B.; data analysis, R.B., P.W.; statistical analysis, R.B.; manuscript preparation, R.B.; manuscript review, R.S., F.F., F.E., P.W.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  B.M. Ellingson, J.L. Ulmer, S.N. Kurpad, and B.D. Schmit Diffusion Tensor MR Imaging in Chronic Spinal Cord Injury AJNR Am. J. Neuroradiol., November 1, 2008; 29(10): 1976 - 1982. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B.M. Ellingson, J.L. Ulmer, S.N. Kurpad, and B.D. Schmit Diffusion Tensor MR Imaging of the Neurologically Intact Human Spinal Cord AJNR Am. J. Neuroradiol., August 1, 2008; 29(7): 1279 - 1284. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Yamashita, T. Yoshiura, H. Arimura, F. Mihara, T. Noguchi, A. Hiwatashi, O. Togao, Y. Yamashita, T. Shono, S. Kumazawa, et al. Performance Evaluation of Radiologists with Artificial Neural Network for Differential Diagnosis of Intra-Axial Cerebral Tumors on MR Images AJNR Am. J. Neuroradiol., June 1, 2008; 29(6): 1153 - 1158. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Murakami, T. Sugahara, H. Nakamura, T. Hirai, M. Kitajima, Y. Hayashida, Y. Baba, N. Oya, J.-i. Kuratsu, and Y. Yamashita Malignant Supratentorial Astrocytoma Treated with Postoperative Radiation Therapy: Prognostic Value of Pretreatment Quantitative Diffusion-weighted MR Imaging Radiology, May 1, 2007; 243(2): 493 - 499. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Summers, P. Staempfli, T. Jaermann, S. Kwiecinski, and S. Kollias A preliminary study of the effects of trigger timing on diffusion tensor imaging of the human spinal cord. AJNR Am. J. Neuroradiol., October 1, 2006; 27(9): 1952 - 1961. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C.-Y. Chen, C.-W. Li, Y.-T. Kuo, T.-S. Jaw, D.-K. Wu, J.-C. Jao, J.-S. Hsu, and G.-C. Liu Early Response of Hepatocellular Carcinoma to Transcatheter Arterial Chemoembolization: Choline Levels and MR Diffusion Constants--Initial Experience Radiology, May 1, 2006; 239(2): 448 - 456. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Taouli, V. Vilgrain, E. Dumont, J.-L. Daire, B. Fan, and Y. Menu Evaluation of Liver Diffusion Isotropy and Characterization of Focal Hepatic Lesions with Two Single-Shot Echo-planar MR Imaging Sequences: Prospective Study in 66 Patients Radiology, January 1, 2003; 226(1): 71 - 78. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. M. Herneth, M. O. Philipp, J. Naude, M. Funovics, R. R. Beichel, R. Bammer, and H. Imhof Vertebral Metastases: Assessment with Apparent Diffusion Coefficient Radiology, December 1, 2002; 225(3): 889 - 894. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Mamata, Y. Mamata, C.-F. Westin, M. E. Shenton, R. Kikinis, F. A. Jolesz, and S. E. Maier High-Resolution Line Scan Diffusion Tensor MR Imaging of White Matter Fiber Tract Anatomy AJNR Am. J. Neuroradiol., January 1, 2002; 23(1): 67 - 75. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Cercignani, M. Inglese, E. Pagani, G. Comi, and M. Filippi Mean Diffusivity and Fractional Anisotropy Histograms of Patients with Multiple Sclerosis AJNR Am. J. Neuroradiol., May 1, 2001; 22(5): 952 - 958. [Abstract] [Full Text]
|
|
|
|
 |
|
 S. L. Keir and J. M. Wardlaw Systematic Review of Diffusion and Perfusion Imaging in Acute Ischemic Stroke Stroke, November 1, 2000; 31(11): 2723 - 2731. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Augustin, R. Bammer, J. Simbrunner, R. Stollberger, H.-P. Hartung, and F. Fazekas Diffusion-weighted Imaging of Patients with Subacute Cerebral Ischemia: Comparison with Conventional and Contrast-enhanced MR Imaging AJNR Am. J. Neuroradiol., October 1, |
). After 4–6 days of decreased mean rADCs, the values start to rise, with an increased mean rADC that is characteristic of that of old lesions. This mean rADC course contributes to the timing of infarction. Note the problem of pseudonormalization during the transitional phase.