M. Essig, MD; B. Stieltjes, MD
Abt. Radiologische Diagnostik und Therapie, Deutsches Krebs- forschungszentrum Heidelberg, Im Neuenheimer Feld 280, 69120 Heidelberg, Germany
W. Reith, MD
Radiologische Universitätsklinik, Abt. für Neuroradiologie, Kirrbergerstr. 1, 66421 Homburg/Saar, Germany
C O N T E N T S
33.1 Introduction 371 33.2 Diffusion MRI 371 33.3 Perfusion MRI 375 33.4 Conclusion 377
References 377
Imaging of CNS Diffusion and Perfusion 33
Marco Essig, Bram Stieltjes, and Wolfgang Reith
33.1
Introduction
The assessment of cerebral functions has long been the domain of positron-emission tomography (PET) and single-photon-emission computed tomography (SPECT). With the use of fast imaging sequences and the availability of contrast agents the assessment and monitoring of physiological and pathophysiological cerebral processes using magnetic resonance imaging (MRI) has become possible. Presently, T1-weighted, as well as T2*-weighted contrast-enhanced fast- imaging sequences, can be used for the assessment of tissue perfusion, vascularity and microcirculation.
Diffusion MRI allows the functional assessment of white matter viability and structure. To obtain physi- ological information from MRI the measurements have to be performed with a high temporal resolution in combination with suffi cient MR signal.
Most functional MR imaging methods are based on an echo-planar imaging (EPI) acquisition with the limitation of low signal-to-noise ratio (SNR) and a limited spatial resolution. Especially diffusion- weighted MRI and diffusion tensor imaging (DTI) with intrinsic low signal intensities due to the dif- fusion weighting are limited because of their strong link between voxel size and SNR; therefore, any improvement of the SNR will enable a better spatial resolution which is possible by use of higher mag- netic fi elds (e.g. 3 T).
Herein the application of parallel imaging in diffu- sion and perfusion MRI of the brain is described. For single-shot acquisitions, such as echo-planar imag- ing, parallel imaging is particularly advantageous since the reduction of the SNR is counterbalanced by the shortening of the echo-train length. Since advan- tages of parallel imaging are more obvious on high- fi eld systems, most reports are from 3-T systems or higher. Owing to largely independent physics, the two approaches can be readily combined. Consider- ing the specifi c advantages and disadvantages of high fi eld strength and parallel imaging, it is found that the combination is particularly synergistic. In the joint approach, the two concepts play different roles.
Higher fi eld strength acts as a source of higher base- line SNR, whereas parallel imaging acts as a means of converting added SNR into a variety of alternative benefi ts. This interplay holds promise for a broad range of clinical applications, as recently illustrated by several imaging studies at 3 T. As a consequence, clinical MRI at 3 T and higher is expected to rely more on parallel imaging than at lower fi eld strength.
33.2
Diffusion MRI
Diffusion-weighted imaging (DWI) has become one of the most important tools in the MR assessment of
urements are time-consuming, which may limit the results in restless and uncooperative patients. Also, to ensure acceptable acquisition times, the spatial resolution used for DWI is low; typically a 128×128 matrix and a minimum of 6-mm slice thickness are used to keep the SNR and the acquisition time within clinically acceptable limits.
To overcome these limitations, there are two major approaches: the combination of high-fi eld (≥3 T) systems is ideally suited to improve DWI; and DTI, because of the double SNR compared with 1.5 T (Hun- sche 2001). The high magnetic fi eld strength, however, will degrade the image quality because of the expo- nential increase of susceptibility artefacts, increased image distortion and blurring; these, however, are strongly reduced by parallel-imaging techniques (Fig. 33.2). Moreover, at higher fi eld strength the geometry factor remains close to 1 at higher reduction rates (Wiesinger 2004), thus increasing the potential
eight-element SENSE-compatible receive-only sur- face coil they compared image quality, SNR, relative signal intensity and lesion contrast for diagnostic accuracy of and confi dence in detection of appar- ent-diffusion-coeffi cient (ADC) lesions. The authors found no major SENSE-related reconstruction arte- facts. Using DWI/DTI in combination with SENSE, consistently and signifi cantly (p<0.001) higher image quality scores were achieved because of substantial reduction of image distortions and blurring especially in areas close to the base and the vertex of the scull where conventional DWI suffers from substantial geometric distortions. Lesion contrast was equivalent with both techniques; however, the diagnostic confi - dence for demonstration and exclusion of lesions was signifi cantly (p<0.001) higher at MRI with SENSE.
In 3 patients, small microembolic lesions were only prospectively diagnosed at MR imaging with SENSE, whereas they were masked by adjacent susceptibility
Fig. 33.1a,b. A 55-year-old patient with right temporoparietal glioblastoma. Diffusion-weighted imaging performed at 1.5 T without parallel imaging a and with parallel imaging b. Parallel imaging resulted in substantial better image quality. N/2 artefacts are reduced as well as distortion artefacts at the scull base (arrows).
b a
effects and therefore overlooked at MR imaging with conventional (full) phase encoding.
Jaermann et al. (2004) evaluated SENSE DTI at 3 T in a recent publication in nine healthy volunteers.
Susceptibility-related artefacts, most pronounced close to the skull base, frontal cavity and inner ear, could be effectively compensated by the use of paral- lel imaging with shortening of the EPI readout. They observed the same substantial artefact suppression at 3 T as previously described at a fi eld strength of 1.5 T (Bammer 2001, 2002). Another advantage of paral- lel imaging is the effect it has on actual resolution.
Reducing signal decay during the readout effectively broadens the associated k-space fi lter, leading to a narrower point spread function and higher actual resolution than with conventional DTI (cf. Chap. 10;
Jaermann 2004). This can be noted when comparing
images of the same matrix size and fi eld of view. Here images recorded using SENSE show sharper borders.
Although distortions are most pronounced at tissue
air borders, all fi eld inhomogeneities lead to slight dis- ruption in observed fi ber orientation. This may lead to alterations in overall fi ber direction and thus hamper the results, especially when using fi ber tracking. This effect again would be most pronounced in areas where two fi ber systems cross. In Fig. 33.3a fi bertracking of the corona radiata was performed. Note that the fi bers do not reach the motorcortex and are bent backwards, due to mixing of other fi ber systems. In the data set using parallel imaging with acceleration factor R=2 the fi bers do reach up to the cortex (Fig. 33.3b). This shows the effect of relatively small distortions on overall fi ber orientation. (For further details about application of parallel imaging in DTI see Chap. 34).
b c a
d e f
Fig. 33.2a–f. A 59-year-old man with brain metastases: diffusion-weighted MRI with b-value 1000 s/mm² a,c ADC maps. b,d Contrast-enhanced T1-weighted and e T2-weighted acquisition. All images were acquired at 3 T. Acquisitions a,b without and c,d with parallel imaging were compared. Parallel imaging reduces distortion and aliasing artefacts, partially obscuring the small metastasis in the non-accelerated image (arrows). Imaging parameters were: matrix 192×192; fi eld of view 230×230 mm²; slice thickness 5 mm. Timing without parallel imaging: TR/TE=4500/127 ms; with parallel imaging: TR/TE=3000/95 ms
Fig. 33.3a,b. Reconstruc- tion of the internal cap- sule a without and b with parallel imaging. Stream tubes were obtained using a FACT-based fi ber-tracking algorithm (Mori 1999). Fibers are overlaid on sagittal color- coded fractional-anisot- ropy map.
a
b
In conclusion, the combination of EPI and parallel imaging for DWI/DTI is extremely favourable, reduc- ing EPI-induced artefacts and small distortions. Fur- thermore, due to the strong reduction in readout echo-train length, the inherent reduction of SNR in SENSE imaging is counteracted increasing the effec- tive SNR and resolution. Combination of parallel imaging and higher fi eld strengths will even enhance these positive effects.
33.3
Perfusion MRI
Perfusion is physiologically defi ned as the steady- state delivery of blood to an element of tissue. The term “perfusion” is also used to emphasize contact with the tissue, or in other words, capillary blood fl ow.
Because perfusion and blood volume is disturbed in many disease processes, monitoring of this key physi- ological parameter can often provide insight into dis- ease. Consequently, the measurement of perfusion for medical purposes has been performed in almost all organs using many techniques.
During the past decade several methods have been described to non-invasively measure perfusion with MRI. Most effort in this context has been made in the perfusion imaging of the brain with two major approaches: contrast-enhanced techniques based on tracer-kinetic models and non-enhanced techniques based on arterial spin labeling. Whereas the con- trast-enhanced techniques are well established, the non-enhanced techniques are of increasing interest with the use of high-fi eld MR systems; especially in the latter the limited anatomic coverage and common EPI artefacts have been limitations in the past. With the use of parallel-imaging techniques the method might be improved substantially.
At a fi eld strength of 1.5 T the main approaches to assess brain tissue perfusion are dynamic contrast- enhanced techniques (DCE MRI): T2*-based per- fusion MRI and tracer-kinetic MRI.
T2*-based perfusion MRI is based on a rapid con- trast-media injection and the evaluation of the signal- intensitytime curve with spin-echo or gradient-echo EPI sequences. Compared with the previously used gradient-echo sequences, the EPI-based sequences have the advantage of a better anatomic coverage but the disadvantage of the EPI technique, namely a reduced spatial resolution and more artefacts, e.g. at
the scull base and close to bone/air interfaces. This is a major drawback, especially in the assessment of small structures at the temporal lobes which are of importance in the assessment of neurodegenerative diseases and in lesions (e.g. ischaemic or tumorous) close to areas with a high potential for artefacts. The- oretically the use of parallel imaging might reduce these artefacts and allows to measure at a higher spa- tial resolution. The encoding time reduction with the use of parallel imaging can be used to achieve a higher temporal resolution of a conventional dynamic scan or to increase the spatial resolution of the typically low resolution of dynamic scans. For EPI acquisitions the echo-train length could be reduced with reduc- tion of image distortion and blurring. Disadvantages of parallel imaging are the reduction of the SNR and some restrictions in scan prescriptions as well as additional and new artefacts induced by parallel imaging (Madore and Pelc 2001).
Stollberger and Fazekas (2004) evaluated the infl uence of parallel imaging on DCE MRI. In their review article they discuss both the T2*-based per- fusion MRI and the tracer-kinetic MRI in different clinical applications. With the currently used slice thickness of 5–7 mm it seems not to be possible with non-accelerated acquisitions to cover the whole brain as desired especially for perfusion stroke imaging.
Another problem is the determination of the arterial input function which suffers from partial-volume effects, arterial voxel shift, saturation effects of the arterial input function and blurring.
The application of parallel imaging seems to be an almost ideal method to reduce the above-mentioned problems. Firstly, the possible shorter echo-train length reduces image distortion and blurring. In brain-tumor imaging the authors describe an advan- tage especially for lesions close to the scull base and in the postoperative assessment. Images can be inter- preted even in the close vicinity of an artefact due to surgical material at the site of craniotomy. With their proposed protocol, it was possible to cover the whole brain at a very high temporal resolution of <1.5 s per repetition. The reduction of artefacts (Fig. 33.4), reduced blurring and the shortening of the echo time enabled better determination of the arterial input function. A further reduction of the echo time is possible with the implementation of interleaved EPI sequences. The combination of such sequences with parallel imaging allows the acquisition of a double- echo sequence in which the fi rst echo can be meas- ured with an echo time well below 10 ms, whereas the second echo can be optimized for an ideal weight-
ing of the parenchyma. The double-echo sequence is an elegant method for suppressing T1-enhancement in cases of leaky microvessels and contrast agent extravasation (Vonken et al. 2000; Uematsu et al.
2000).
In tracer-kinetic MRI, parallel MRI can be used to reduce the scan time or to increase the scan res- olution at the same scanning time or to achieve an optimal compromise of both (Stollberger 2004).
The advantage of parallel imaging for tracer-kinetic MRI has been shown for imaging of the breast and liver (Kuhl 2002; McKenzie 2004); however, for CNS indications, e.g. brain-tumor characterization, there are no data available.
For non-enhanced perfusion MRI, a variety of continuous and pulsed arterial-spin-labeling (ASL) MRI techniques have been demonstrated in recent years and the infl uence of parallel imaging on these techniques have been evaluated mainly on 3-T MRI systems.
Talagala and colleagues (2004) from the NIH presented a fi rst study on whole-brain 3D perfusion MRI at 3 T using continuous arterial spin labeling with a separate labeling coil. This approach provided an increased SNR in perfusion images because there are no magnetization transfer effects. In their work, they also demonstrated the ability to acquire per-
fusion images in all brain regions with good sensitiv- ity. Furthermore, they showed that the method can be performed safely on humans without exceeding the current RF-power-deposition limits because it uses a local labeling coil. The current method can be extended to even higher fi elds, and further improved by the use of multiple receiver coils and parallel- imaging techniques to reduce scan time or provide increased resolution.
In a recent study Wang et al. (2005) evaluated the feasibility and effi cacy of using an array coil and par- allel imaging in continuous-arterial-spin-labeling (CASL) perfusion MRI. An 8-channel receive-only physed-array head coil was used in conjunction with a surrounding detunable volume transmit coil. The SNR, temporal stability, cerebral blood fl ow (CBF) and perfusion-image coverage were measured from steady-state CASL scans using a standard volume coil, an array coil and a phased-array coil with two- and threefold accelerated parallel imaging. They describe that compared with the standard volume coil, the array coil provided three times the average SNR and higher temporal stability for the perfusion- weighted images, even with threefold acceleration.
Although perfusion images of the array coil were affected by the inhomogeneous coil sensitivities, this effect was invisible in the quantitative CBF images,
Fig. 33.4a,b. a Echo-planar imaging T2*-based perfusion MRI source image without parallel imaging. Note the pronounced frontal and occipital image distortion, especially at the bone/air/tissue interfaces (arrows) . b With the use of parallel imaging (acceleration factor R=2) the artefacts could be substantially reduced. The image is less blurred and less image distortion is seen (arrows).
b a
which showed highly reproducible perfusion values compared with the standard volume coil. The unfold- ing distortions of parallel imaging were suppressed in the perfusion images by pairwise subtraction, although they sharply degraded the raw EPI images.
Moreover, parallel imaging provided the potential of acquiring more slices due to the shortened acquisi- tion time and improved coverage in brain regions with high static fi eld inhomogeneity.
33.4 Conclusion
The assessment of tissue function is one of the most challenging goals in neuroimaging. Presently, struc- tural MRI allows not only for the evaluation of the tissue characteristics, but also the assessment of tissue integrity. Diffusion-weighted MRI and diffu- sion-tensor imaging have enabled to visualize neu- ronal pathways and structural details. Functional MRI allows to gather non- or minimally invasively information about the tissue perfusion, vascularity and metabolism.
Recent investigations have shown that most of the advanced structural and functional MR sequences benefi t substantially from the application of parallel- imaging techniques. Methodological and fi rst clinical studies describe a signifi cant artefact reduction, the possibility of faster acquisition and a more robust assessment of the structural and functional param- eters with the use of parallel imaging. Further devel- opments can be expected with new and optimized coil arrays in the near future.
References
Bammer R, Keeling SL, Augustin M, Pruessmann KP, Wolf R, Stollberger R, Hartung HP, Fazekas F (2001) Improved diffusion-weighted single echo-planar imaging (EPI) in stroke using sensitivity enconding (SENSE). Magn Reson Med 46:548–554
Bammer R, Auer M, Keeling SL et al. (2002) Diffusion tensor imaging using single-shot SENSE-EPI. Magn Reson Med 48:128–136
Bukte Y, Paksoy Y, Genc E, Uca AU (2005) Role of diffusion- weighted MR in differential diagnosis of intracranial cystic lesions. Clin Radiol 60:375–383
Conturo TE, Lori NF, Cull TS et al. (1999) Tracking neuronal fi ber pathways in the living human brain. Proc Natl Acad Sci USA 96:10422–10427
Hunsche S, Mosely M, Stoeter P, Hedehus M (2001) Diffusion- tensor MR imaging at 1.5 and 3.0 T: initial observations.
Radiology 221:550–556
Jaermann T, Crelier G, Pruessmann KP et al. (2004) SENSE- DTI at 3 T. Magn Reson Med 51:230–236
Kuhl CK, Morakkabati N, Sondermann E et al. (2002) Dynamic breast MR imaging using sensitivity encoding (SENSE).
Proc Int Soc Mag Reson Med 10:1410
Kuhl CK, Gieseke J, Falkenhausen M von et al. (2005) Sensitiv- ity encoding for diffusion-weighted MR imaging at 3.0 T:
intraindividual comparative study. Radiology 234:517–526 Madore B, Pelc NJ (2001) SMASH and SENSE: experimental and
numerical comparisons. Magn Reson Med 45:1103–1111 McKenzie CA, Lim D, Ransil BJ et al. (2004) Shortening MR
image acquisition time for volumetric interpolated breath- hold examination with a recently developed parallel imag- ing reconstruction technique: clinical feasibility. Radiology 230:589–594
Mori S, Crain BJ, Chacko VP et al. (1999) Three-dimensional tracking of axonal projections in the brain by magnetic resonance imaging. Ann Neurol 45:265–269
Stieltjes B, Kaufmann WE, van Zijl PC et al. (2001) Diffusion tensor imaging and axonal tracking in the human brain- stem. Neuroimage 14:723–735
Stollberger R, Fazekas F (2004) Improved perfusion and tracer kinetic imaging using parallel imaging. Top Magn Reson Imaging 15:245–254
Talagala SL, Ye FQ, Ledden PJ, Chesnick S (2004) Whole-brain 3D perfusion MRI at 3.0 T using CASL with a separate labeling coil. Magn Reson Med 52:131–140
Tsuchiya K, Fujikawa A, Suzuki Y (2005) Diffusion tractogra- phy of the cervical spinal cord using parallel imaging. Am J Neuroradiol 26:398–400
Uematsu H, Maeda M, Sadato N et al. (2000) Vascular per- meability: quantitative measurements with double-echo dynamic MR imaging. Theory and clinical application.
Radiology 214:912–917
Vonken EP, van Osch MJ, Bakker CJ et al. (2000) Simultaneous quantitative cerebral perfusion and Gd-DTPA extravasa- tion measurement with dual-echo dynamic susceptibility contrast MRI. Magn Reson Med 43:820–827
Wang Z, Wang J, Connick TJ, Wetmore GS, Detre JA (2005) Continuous ASL (CASL) perfusion MRI with an array coil and parallel imaging at 3 T. Magn Reson Med 54:732–737 Warach S, Gaa J, Siewert B, Wielopolski P, Edelman RR (1995)
Acute human stroke studied by whole brain echo planar diffusion-weighted magnetic resonance imaging. Ann Neurol 37: 231–241
Wiesinger F, Van de Moortele P-F, Adriany G, de Zanche N, Ugurbil K, Pruessmann KP (2004) Parallel imaging per- formance as a function of fi eld strength: an experimental investigation using electrodynamic scaling. Magn Reson Med 52: 953–964
