Intravoxel Incoherent Motion Perfusion MR Imaging Wake Up Call

Document Sample
Intravoxel Incoherent Motion Perfusion MR Imaging Wake Up Call Powered By Docstoc

                                                                                      Intravoxel Incoherent Motion
                                                                                      Perfusion MR Imaging: A

                                                                                      Wake-Up Call1

                           Denis Le Bihan, MD, PhD

                                                                                          n an article published in this issue of    How would it be possible to differenti-
                                                                                          Radiology, Luciani et al (1) report        ate liver tumors from angiomas? There
                                                                                          that perfusion, an important surro-        were no available contrast media at that
                                                                                      gate marker to evaluate the importance         time. I had a fuzzy intuition that perhaps
                                                                                      of liver fibrosis, was decreased in pa-         molecular diffusion measurement would
                                                                                      tients with chronic cirrhosis. Notably,        result in low values in solid tumors be-
                                                                                      the magnetic resonance (MR) imaging            cause of molecular movement restric-
                                                                                      perfusion measurements were not ob-            tion, while diffusion would be somewhat
                                                                                      tained with contrast agents, but with          enhanced in flowing blood. In a matter
                                                                                      diffusion MR imaging, or more exactly,         of weeks “diffusion MR imaging” was
                                                                                      intravoxel incoherent motion (IVIM)            conceived, born, and implemented. Un-
                                                                                      MR imaging. Another interesting finding         fortunately, the method never worked
                                                                                      was that water diffusion apparently re-        in the liver, at least not until the 1999
                                                                                      mains normal in those patients, in con-        landmark article by Yamada et al (4),
                                                                                      trast to results of earlier studies, which     which demonstrated that I was not com-
                                                                                      suggest that the apparent diffusion coef-      pletely wrong. First, the MR imager we
                                                                                      ficient (ADC) decreases in patients with        used operated at 0.5 T (the almost
                                                                                      cirrhosis. Those results are, indeed, not      homemade Magniscan imager by CGR
                                                                                      as contradictory as they appear and            [Buc, France]), and the signal was very
                                                                                      point to an important aspect of diffusion      low. Second, the gradient hardware
                                                                                      and IVIM MR imaging.                           barely allowed strengths beyond 8 or 10
                                                                                           This excellent article by Luciani et al   mT/m to be reached (and these still had
                                                                                      (1) transported me back to the debut of        large eddy currents), and b values
                                                                                      diffusion MR imaging in the mid-1980s,         larger than 100 mm/sec2 were not even
                                                                                      and I thought this would be a good op-         in sight. Third, there was no echo-
                                                                                      portunity to share some “historical” mo-       planar imaging, and we just used spin-
                                                                                      ments. Diffusion MR imaging has been           echo sequences that were available. Ac-
                                                                                      extraordinarily successful during the          quisition times necessary for diffusion
                                                                                      past 20 years (with more than 85 000           encoding were very long (close to 10
                                                                                      entries in Google Scholar for “diffusion       minutes per b value), and, as respira-
                                                                                      MRI”), but its main clinical domain of         tory gating was not available, motion
                                                                                      application has been neurologic disor-         artifacts were atrocious in the body. So,
                                                                                      ders, especially for the treatment of pa-      I gave up and switched to the brain, as it
                                                                                      tients with acute stroke. It is also rap-      was my background after all. That move
                                                                                      idly becoming a standard for white             resulted in a great achievement: Diffu-
                                                                                      matter disorders, as diffusion-tensor          sion MR imaging was established (5,6),
                           Published online
                                                                                      imaging can help reveal abnormalities in       and the rest is history.
                                                                                      white matter fiber structure and allow               I continued to investigate the idea
                           Radiology 2008; 249:748 –752                               outstanding maps of brain connectivity         that diffusion MR imaging could provide
                             From NeuroSpin, I2BM, Batiment 145, CEA Saclay-Cen-
                                                                                      (2) to be obtained (3), which may be of        information on perfusion. I came up
                           ter, 91191 Gif-sur-Yvette, France. Received July 28,       great potential in the evaluation of some      with the view that perhaps the move-
                           2008; revision requested July 28; revision received July   psychiatric disorders.                         ment of blood in the microvasculature
                           29; accepted July 30; final version accepted July 31.            However, it is perhaps not so well        could be modeled as a pseudodiffusion
                           Address correspondence to the author (e-mail:              known that, in my hands at least, diffu-       process on a macroscopic scale. In the
                                                                                      sion MR imaging started in the liver.          true (molecular) diffusion process, mol-
                           Author stated no financial relationship to disclose.        Back in 1984 (when I was a radiology           ecules move because of their own ther-
                                                                                      resident), Denis Lallemand, a pediatric        mal energy and can be considered to be
                           See also the article by Luciani et al in this issue.
                                                                                      radiologist from Necker Hospital in            colliding with each other. (Actual diffu-
                             RSNA, 2008                                               Paris, came to me with a challenge:            sion of water is, indeed, much more

                           748                                                                                                            Radiology: Volume 249: Number 3—December 2008
EDITORIAL: Intravoxel Incoherent Motion                                                                                            Le Bihan

complex; see reference 7 for a review.)         Maurice Gueron (at the Ecole Polytech-
                                                             ´                                  spiratory triggering, and state-of-the-
Each collision results in a change in the       nique where I was completing my PhD             art gradient hardware that reached 40
motion direction of each molecule, and          in physics), to come up with the concept        mT/m.
the overall process is well described by        of IVIM to cover the overall molecular               Several other groups have shown
a random walk, as first realized by Ein-         displacements to which “diffusion” MR           encouraging results in the brain
stein (8). Similarly, one may consider,         imaging could be sensitive (6). Hence, it       (9,15,17,18), in the kidneys (19,20),
at a macroscopic level, that in blood, in       was very clear that the results of diffu-       and even in the heart (21). However,
addition to diffusion, water molecules          sion measurement with MR imaging                other researchers (11,22) have ex-
follow the stream and change direction          could include perfusion effects, among          pressed concerns about the concepts
between each capillary segment. If              other things, and not only true diffusion,      beyond perfusion measurement with
those segments are distributed in space         as beautifully demonstrated by Yamada           the IVIM method and its ability to mea-
in a pseudorandom manner, the overall           et al (4) and now by Luciani et al (1),         sure “classical” perfusion, compared
movement mimics a random walk and               and the term ADC was introduced (6).            with tracer methods. “Perfusion” had to
the mathematical model used for diffu-          The theoretical framework for IVIM and          be redefined according to the viewpoint
sion should work as well. Although the          the demonstration of the validity of the        of the physiologist (blood flow) and the
difference in spatial scale between the         concepts in phantoms and in vivo was            viewpoint of the radiologist (vascular
processes of diffusion (nanometers) and         introduced in a seminal Radiology arti-         density). Indeed, the exact nature of
pseudodiffusion (tens of micrometers)           cle (10), accompanied by a terrific edi-         what is measured with IVIM MR imag-
extends across five orders of magni-             torial by W. Thomas Dixon (12). Inter-          ing deserves further investigation,
tude, it is amazing to observe that the         estingly, Dixon recently told me that           which is a point raised by the work of
associated diffusion and pseudodiffu-           this editorial was one of his most cited        Luciani et al (1). The authors found that
sion coefficients differ only by roughly         articles, and, indeed, it is fair to say that   D* was significantly reduced in patients
one order of magnitude (D, the molec-           IVIM has been a great subject of contro-        with cirrhosis, which, according to the
ular diffusion coefficient of water in tis-      versy.                                          IVIM model, points to reduced blood
sues, is about 1 10 3 mm2/sec, while                Probably because of this article            velocity (and flow). (Another theoreti-
D*, the pseudodiffusion coefficient asso-        (10), diffusion MR imaging has been as-         cal, rather unlikely interpretation would
ciated with blood flow, is about 10              sociated with perfusion imaging for             be that capillary segments become
10 3 mm2/sec in the brain [9] and 70            many years, hence the many diffusion            longer or straighter in those patients
10 3 mm2/sec in the liver [1]). This is         and perfusion sessions at meetings and          with liver fibrosis.) Interestingly, the
because those coefficients combine ef-           workshops, books (13), and Radiology            perfusion fraction, f, which is linked to
fects of elementary particle velocity and       journal keywords; however, diffusion            blood volume in the IVIM model, re-
distance (10,11). Molecular diffusion is        and perfusion refer to completely differ-       mained normal, confirming earlier re-
a very fast process, as far as molecular        ent phenomena, both physically and bi-          sults of Yamada et al (4). However,
distances are concerned, while blood            ologically. This unexpected association         blood volume is expected to be reduced
flow pseudodiffusion is comparatively            has been a little puzzling for some of my       in liver cirrhosis. One has to keep in
much slower but is over distances of            colleagues, and at some point, they             mind that IVIM imaging has a differen-
tens of micrometers.                            teased me with such aphorisms as “dif-          tial sensitivity to vessel types, according
     This might look like a great coinci-       fusion, perfusion,. . . confusion.” Any-        to the range of motion sensitization (b
dence. However, one may also consider           way, there were real technical issues.          values) that is used (23,24). Signal from
that perfusion and diffusion are the ma-        Separation of perfusion from diffusion          large vessels with rapid flow disappears
jor ways for most molecular moieties to         requires good signal-to-noise ratios,           quickly with very low b values, while
reach targets in tissues. If flow is too         which were difficult to reach with low-          smaller vessels with slower flow might
fast, molecules may not be able to reach        field-strength MR imaging systems and            still contribute to the IVIM signal ac-
vessel walls and diffuse into tissues. One      limited gradient hardware (14,15). It           quired with b values larger than 200
may speculate that perhaps some bal-            was not until the availability of echo-         sec/mm2. The very limited number of b
ance between molecular diffusion and            planar imaging on clinical MR imagers           values applied by Luciani et al, for prac-
perfusion has always been kept during           that diffusion and IVIM MR imaging              tical reasons, precluded a fine analysis
evolution, shaping up the vascular net-         could really take off (16), and, as results     of the different vascular compartments,
work. In any case, their proximity in           became much more reliable, be free of           some of which could be preserved by
values allows D and D* to be evaluated          motion artifacts. Luckily, Luciani et al        the fibrosis process or by compensatory
together with the same diffusion MR im-         (1) benefited from tremendous ad-                mechanisms as suggested by the au-
aging sequence, which is good news, but         vances in MR imaging technology, with           thors. Furthermore, the limited accu-
it also means that diffusion MR images          the combination of echo-planar imaging          racy of the estimation of f, as derived
are prone to contamination by blood             with parallel imaging by using 18 chan-         from only four b values, could mask
microcirculation effects. It took a great       nels, reduced echo times, acquisitions          real, but relatively modest, changes in
deal of brainstorming with my mentor,           less vulnerable to motion because of re-        blood volume. Clearly, there is room to

Radiology: Volume 249: Number 3—December 2008                                                                                          749
EDITORIAL: Intravoxel Incoherent Motion                                                                                             Le Bihan

improve the IVIM model and under-            membranes. Beyond the low b-value             sensitivity. The largest b value used by
stand better its relationship with the       range (above 600 sec/mm2), diffusion          Luciani et al (800 sec/mm2) was clearly
functional vascular architecture and its     plots still present a net curvature, al-      too small to allow those slow diffusion
biologic relevance.                          though perfusion effects are no longer        pools to have a noticeable effect on dif-
    Another very important practical         contributing. By using b-value ranges         fusion MR imaging signal, and it might
and conceptual point raised by Luciani       extending more than 3000 sec/mm2,             be that water diffusion is altered in liver
et al (1) is that water diffusion was also   one may reveal different diffusion be-        fibrosis. With further improvements in
unchanged in patients with cirrhosis.        haviors or compartments. The ADC              gradient hardware performance that
While other groups (25–27) have re-          concept applies here again, but now to        are expected, it is hoped that very high
ported decreased ADC values, Luciani         reflect the presence of different diffu-       b values can be reached in a clinical
and colleagues (1) rightly point out that    sion contributions instead of a single dif-   setting to more definitely answer these
the ADC contains both perfusion and          fusion coefficient. The origin of this cur-    questions.
diffusion terms. Furthermore, in a given     vature is still debated (30–32), but, as it        A major goal for IVIM imaging at the
voxel, diffusion may arise from different    can be well described by a biexponential      end of the 1980s was to produce maps
compartments with different diffusion        law, many groups (33,34) have sug-            of brain perfusion to investigate brain
coefficients. As a result, the dependence     gested that it arises from the presence       function. However, other competing
of the diffusion-weighted signal (in log     of two main diffusing water pools in          methods appeared at about the same
plots) on the b value is no longer           slow or intermediate exchange. One is         time, especially the blood oxygen level–
straight, as would be expected for free      associated with a somewhat fast diffu-        dependent (BOLD) concept (39). BOLD
diffusion, but it is curved, reflecting the   sion coefficient (about 1.3 10 3 mm2/          functional MR imaging was clearly much
multiplicity of the underlying processes.    sec in the brain), and the other is asso-     easier to implement and much more
Given the relative values for D* and D,      ciated with a slow diffusion coefficient       sensitive, so there was no real room for
perfusion is expected to contribute to       (about 1.3       10 3 mm2/sec); the ADC       the challenging IVIM method. Studies
this curvature in a biexponential mode       depends on the relative contribution of       (40–42), however, have proved the va-
(10) for b values in the very low range      those two compartments in each voxel.         lidity of the IVIM concept, with an in-
(0 –200 sec/mm2 or even higher for very      The fractions associated with those           crease in the IVIM perfusion parame-
slow flow). Any ADC estimation with           compartments, 70% and 30% for the             ters in brain-activated regions and the
only two b values (eg, 0 and 1000 sec/       fast and the slow components, respec-         potential of the approach to understand
mm2), as classically performed for clin-     tively, invalidate an initial assumption      different vascular contributions to the
ical studies, would miss the curvature,      that those pools could correspond to the      functional MR imaging signal. Interest-
include perfusion effects, and obviously     extra- and intracellular compartments,        ingly, IVIM MR imaging has also been
result in an ADC that is an overestima-      respectively. A recent, still speculative     used in the context of functional MR
tion of the true diffusion coefficient, D     model (7) suggests that the slow pool         imaging in a negative way. A limitation
(28). This perfusion contamination gets      could correspond to a thin layer of wa-       for BOLD functional MR imaging is its
larger when using even lower b-value         ter molecules bound to the cell mem-          spatial resolution, as flow increase in
ranges, as often is the case for body        branes through the particular distribu-       somewhat large arteries or veins feeds
diffusion MR imaging, where tissues          tion of their electrostatic charges. This     or drains large neuronal territories. By
with short T2 preclude long echo times,      model accommodates previous reports           inserting “diffusion” gradient pulses in
which in turn limits the gradient pulse      (35,36) linking changes in cell volume        the MR imaging sequence (correspond-
duration used for diffusion encoding.        (and associated membrane surface)             ing to low b values), one may crush the
The decreased ADC values in cirrhosis        with the ADC, such as the decrease in         contribution of the largest vessels (with
observed in the literature could thus re-    ADC observed during cytotoxic edema           high D* values associated with fast flow)
sult from a sole decrease in D* (perfu-      in acute stroke. It also provides a frame-    in the BOLD signal and improve the spa-
sion), while D remains intact, as re-        work to explain diffusion anisotropy in       tial resolution of the activation maps
ported by Luciani et al (1). In summary,     some tissues with elongated structures,       (43–47). Several groups have relied on
ADC values are meaningful only when          such as brain myelin fibers, by the dif-       this trick, however, not always consid-
reported with b values used for their        ference in the number of membrane             ering to refer to the IVIM concept. This
measurement. In the brain, ADCs tend         (and myelin) interfaces encountered by        IVIM concept has also been borrowed
to better reflect true diffusion, as larger   water molecules according to the mea-         to improve other applications, such as
b values can generally be used and blood     surement direction. It also predicts that     arterial spin labeling (48,49), or to sup-
volume is very small (2%– 4%) (29).          the ADC should decrease in tissue un-         press signal from extracellular flowing
    It remains that the conclusion by the    dergoing cell proliferation, as observed      fluid in perfused cell systems (50,51). It
authors that D was unchanged in liver        in cancer (37,38), as membrane density        remains that there are genuine potential
fibrosis can be challenged. Diffusion in      would rise. Diffusion MR imaging with         applications for IVIM MR imaging. Per-
tissues is not free, but largely impeded     very high b values may thus reflect            fusion is a very important surrogate
and restricted by obstacles, mainly cell     changes in tissue structure with greater      marker of many physiologic or patho-

750                                                                                             Radiology: Volume 249: Number 3—December 2008
EDITORIAL: Intravoxel Incoherent Motion                                                                                                                Le Bihan

logic processes. MR imaging perfusion                    scopic motion of water in tissues with MR         23. Lorenz CH, Pickens DR 3rd, Puffer DB,
parameters can be obtained by using                      imaging: a cat brain study. J Comput Assist           Price RR. Magnetic resonance diffusion/per-
                                                         Tomogr 1991;15:19 –25.                                fusion phantom experiments. Magn Reson
gadolinium-based contrast agents, ei-
                                                                                                               Med 1991;19:254 –260.
ther injected as a bolus (to determine               10. Le Bihan D, Breton E, Lallemand D, Aubin
blood flow, transit times, etc) or in a                   ML, Vignaud J, Laval Jeantet M. Separation        24. Kennan RP, Gao JH, Zhong J, Gore JC. A
                                                         of diffusion and perfusion in intravoxel inco-        general model of microcirculatory blood flow
steady-state mode (to address blood
                                                         herent motion (IVIM) MR imaging. Radiol-              effects in gradient sensitized MRI. Med Phys
volume, vessel permeability, etc). With                  ogy 1988;168:497–505.                                 1994;21:539 –545.
the rising concern of nephrogenic sys-
temic fibrosis, some patients cannot be               11. Le Bihan D, Turner R. The capillary               25. Ichikawa T, Haradome H, Hachiya J,
                                                         network: a link between IVIM and classical            Nitatori T, Araki T. Diffusion-weighted MR
examined with such an approach. IVIM                                                                           imaging with a single-shot echoplanar
                                                         perfusion. Magn Reson Med 1992;27:171–
MR imaging may then appear as an in-                     178.                                                  sequence: detection and characterization
teresting alternative to provide crucial                                                                       of focal hepatic lesions. AJR Am J Roentge-
                                                     12. Dixon WT. Separation of diffusion and per-
clues on perfusion in tissues, such as the                                                                     nol 1998;170:397– 402.
                                                         fusion in intravoxel incoherent motion MR
kidneys (52), the liver (1,4), or even the               imaging: a modest proposal with tremen-           26. Taouli B, Tolia AJ, Losada M, et al. Diffu-
placenta during pregnancy (53,54).                       dous potential. Radiology 1988;168:566 –              sion-weighted MRI for quantification of liver
Clearly, thanks to the work by Luciani                   567.                                                  fibrosis: preliminary experience. AJR Am J
et al (1), IVIM MR imaging is waking up                                                                        Roentgenol 2007;189:799 – 806.
                                                     13. Le Bihan D. In: Le Bihan D, ed. Diffusion and
after some time of hibernation; let us                   perfusion magnetic resonance imaging: ap-         27. Lewin M, Poujol-Robert A, Boelle PY, et al.
further explore its potential.                           plications to functional MRI. New York, NY:           Diffusion-weighted magnetic resonance im-
                                                         Raven, 1995.                                          aging for the assessment of fibrosis in
                                                                                                               chronic hepatitis C. Hepatology 2007;46:
References                                           14. Pekar J, Van Zijl PC, Moonen CT. Signal to            658 – 665.
 1. Luciani A, Vignaud A, Cavet M, et al. Liver          noise requirements for quantitative mea-
    cirrhosis: intravoxel incoherent motion MR im-       surements of diffusion and capillary perfu-       28. Sakuma H, Tamagawa Y, Kimura H, et al.
    aging—pilot study. Radiology 2008;249(3):891–        sion in brain using IVIM-MRI. Magn Reson              Intravoxel incoherent motion (IVIM) imaging
    899.                                                 Med 1991;23:1222–1129.                                using an experimental MR unit with small
                                                                                                               bore [in Japanese]. Nippon Igaku Hoshasen
 2. Le Bihan D. Looking into the functional ar-      15. Wirestam R, Brockstedt S, Lindgren A, et al.          Gakkai Zasshi 1989;49:941–943.
    chitecture of the brain with diffusion MRI.          The perfusion fraction in volunteers and in
    Nat Rev Neurosci 2003;4:469 – 480.                   patients with ischaemic stroke. Acta Radiol       29. Grubb RL Jr, Raichle ME, Eichling JO, Ter
                                                         1997;38:961–964.                                      Pogossian MM. The effects of changes in
 3. Hagmann P, Kurant M, Gigandet X, et al.                                                                    PaCO2 on cerebral blood volume, blood
    Mapping human whole-brain structural net-        16. Turner R, Le Bihan D, Maier J, Vavrek R,              flow, and vascular mean transit time. Stroke
    works with diffusion MRI. PloS ONE 2007;2:           Hedges LK, Pekar J. Echo-planar imaging of            1974;5:630 – 639.
    e597.                                                intravoxel incoherent motions. Radiology
                                                         1990;177:407– 414.                                30. Yablonskiy DA, Bretthorst GL, Ackerman
 4. Yamada I, Aung W, Himeno Y, Nakagawa T,                                                                    JJ. Statistical model for diffusion attenuated
    Shibuya H. Diffusion coefficients in abdomi-      17. Chenevert TL, Pipe JG, Williams DM, Brunberg          MR signal. Magn Reson Med 2003;50:664 –
    nal organs and hepatic lesions: evaluation           JA. Quantitative measurement of tissue perfu-         669.
    with intravoxel incoherent motion echo-pla-          sion and diffusion in vivo. Magn Reson Med
    nar MR imaging. Radiology 1999;210:617–              1991;17:197–212.                                  31. Chin CL, Wehrli FW, Hwang SN, Jaggard
    623.                                                                                                       DL, Hackney DB, Wehrli SW. Feasibility of
                                                     18. Neil JJ, Bosch CS, Ackerman JJ. An evalua-            probing boundary morphology of structured
 5. Le Bihan D, Breton E. Imagerie de diffusion          tion of the sensitivity of the intravoxel inco-       materials by 2D NMR q-space imaging. J
    in vivo par resonance magnetique nucleaire.
                 ´            ´           ´              herent motion (IVIM) method of blood flow              Magn Reson 2003;160:20 –25.
    C.R. Acad.Sc.Paris 1985; T. 301, Serie II:
                                        ´                measurement to changes in cerebral blood
    1109 –1112.                                          flow. Magn Reson Med 1994;32:60 – 65.              32. Stanisz GJ, Szafer A, Wright GA, Henkelman
                                                                                                               RM. An analytical model of restricted diffusion in
 6. Le Bihan D, Breton E, Lallemand D, Grenier       19. Powers TA, Lorenz CH, Holburn GE, Price               bovine optic nerve. Magn Reson Med 1997;37:
    P, Cabanis E, Laval Jeantet M. MR imaging            RR. Renal artery stenosis: in vivo perfusion          103–111.
    of intravoxel incoherent motions: application        MR imaging. Radiology 1991;178:543–548.
    to diffusion and perfusion in neurologic dis-                                                          33. Niendorf T, Dijkhuizen RM, Norris DG, Van
                                                     20. Pickens DR 3rd, Jolgren DL, Lorenz CH,                Lookeren Campagne M, Nicolay K. Biexpo-
    orders. Radiology 1986;161:401– 407.
                                                         Creasy JL, Price RR. Magnetic resonance               nential diffusion attenuation in various states
 7. Le Bihan D. The ‘wet mind’: water and func-          perfusion/diffusion imaging of the excised            of brain tissue: implications for diffusion-
    tional neuroimaging. Phys Med Biol 2007;52:          dog kidney. Invest Radiol 1992;27:287–292.            weighted imaging. Magn Reson Med 1996;
    R57–R90.                                                                                                   36:847– 857.
                                                     21. Callot V, Bennett E, Decking UK, Balaban
 8. Einstein A. Uber die von der molecularki-            RS, Wen H. In vivo study of microcirculation      34. Assaf Y, Cohen Y. Non-mono-exponential at-
    netischen Theorie der Warme geforderte
                              ¨                          in canine myocardium using the IVIM                   tenuation of water and N-acetyl aspartate
    Bewegung von in ruhenden Flussigkeiten
                                   ¨                     method. Magn Reson Med 2003;50:531–                   signals due to diffusion in brain tissue. J
    suspendierten Teilchen. Ann Phys (Leipzig)           540.                                                  Magn Reson 1998;131:69 – 85.
    1905;17:549 –569.
                                                     22. Henkelman RM. Does IVIM measure classi-           35. Sotak CH. Nuclear magnetic resonance
 9. Le Bihan D, Moonen CT, Van Zijl PC, Pekar            cal perfusion? Magn Reson Med 1990;16:                (NMR) measurement of the apparent diffu-
    J, Des Pres D. Measuring random micro-               470 – 475.                                            sion coefficient (ADC) of tissue water and its

Radiology: Volume 249: Number 3—December 2008                                                                                                               751
EDITORIAL: Intravoxel Incoherent Motion                                                                                                             Le Bihan

      relationship to cell volume changes in patho-   42. Jin T, Zhao F, Kim SG. Sources of functional        perfusion. Magn Reson Med 1997;38:232–
      logical states. Neurochem Int 2004;45:569 –         apparent diffusion coefficient changes inves-        237.
      582.                                                tigated by diffusion-weighted spin-echo
                                                                                                          49. Kim T, Kim SG. Quantification of cerebral
                                                          fMRI. Magn Reson Med 2006;56:1283–
36. van der Toorn A, Sykova E, Dijkhuizen RM,                                                                 arterial blood volume using arterial spin la-
    et al. Dynamic changes in water ADC, en-                                                                  beling with intravoxel incoherent motion-
    ergy metabolism, extracellular space vol-         43. Boxerman JL, Bandettini PA, Kwong KK,               sensitive gradients. Magn Reson Med 2006;
    ume, and tortuosity in neonatal rat brain             et al. The intravascular contribution of fMRI       55:1047–1057.
    during global ischemia. Magn Reson Med                signal change: Monte Carlo modeling and         50. Van Zijl PC, Moonen CT, Faustino P, Pekar
    1996;36:52– 60.                                       diffusion-weighted studies in vivo. Magn Re-        J, Kaplan O, Cohen JS. Complete separation
                                                          son Med 1995;34:4 –10.                              of intracellular and extracellular information
37. Takahara T, Imai Y, Yamashita T, Yasuda S,
    Nasu S, van Cauteren M. Diffusion-weighted        44. Song AW, Li T. Improved spatial localization        in NMR spectra of perfused cells by diffu-
    whole-body imaging with background body               based on flow-moment-nulled and intra-               sion-weighted spectroscopy. Proc Natl Acad
    signal suppression (DWIBS): technical im-             voxel incoherent motion-weighted fMRI.              Sci U S A 1991;88:3228 –3232.
    provement using free breathing, STIR and              NMR Biomed 2003;16:137–143.                     51. Zhao L, Sukstanskii AL, Kroenke CD, et al.
    high-resolution 3D display. Radiat Med                                                                    Intracellular water specific MR of mi-
    2004;22:275–282.                                  45. Michelich CR, Song AW, MacFall JR. Depen-
                                                                                                              crobead-adherent cells: HeLa cell intracellu-
                                                          dence of gradient-echo and spin-echo BOLD
                                                                                                              lar water diffusion. Magn Reson Med 2008;
38. Ross BD, Moffat BA, Lawrence TS, et al.               fMRI at 4 T on diffusion weighting. NMR
                                                                                                              59:79 – 84.
    Evaluation of cancer therapy using diffusion          Biomed 2006;19:566 –572.
    magnetic resonance imaging. Mol Cancer                                                                52. Tsuda K, Murakami T, Sakurai K, et al. Pre-
    Ther 2003;2:581–587.                              46. Duong TQ, Yacoub E, Adriany G, Hu XP,               liminary evaluation of the apparent diffusion
                                                          Ugurbil K, Kim SG. Microvascular BOLD               coefficient of the kidney with a spiral IVIM
39. Ogawa S, Tank DW, Menon RS, et al. Intrin-            contribution at 4 and 7 T in the human brain:       sequence [in Japanese]. Nippon Igaku
    sic signal changes accompanying sensory               gradient-echo and spin-echo fMRI with sup-          Hoshasen Gakkai Zasshi 1997;57:19 –22.
    stimulation: functional brain mapping with            pression of blood effects. Magn Reson Med
    magnetic resonance imaging. Proc Natl Acad            2003;49:1019 –1027.                             53. Moore RJ, Issa B, Tokarczuk P, et al. In vivo
    Sci U S A 1992;89:5951–5955.                                                                              intravoxel incoherent motion measurements
                                                      47. Lee SP, Silva AC, Kim SG. Comparison of             in the human placenta using echo-planar im-
40. Song AW, Wong EC, Tan SG, Hyde JS. Dif-               diffusion-weighted high-resolution CBF and          aging at 0.5 T. Magn Reson Med 2000;43:
    fusion weighted fMRI at 1.5 T. Magn Reson             spin-echo BOLD fMRI at 9.4 T. Magn Reson            295–302.
    Med 1996;35:155–158.                                  Med 2002;47:736 –741.
                                                                                                          54. Moore RJ, Strachan BK, Tyler DJ, et al. In
41. Gangstead SL, Song AW. On the timing              48. Silva AC, Williams DS, Koretsky AP. Evi-            utero perfusing fraction maps in normal and
    characteristics of the apparent diffusion co-         dence for the exchange of arterial spin-la-         growth restricted pregnancy measured using
    efficient contrast in fMRI. Magn Reson Med             beled water with tissue water in rat brain          IVIM echo-planar MRI. Placenta 2000;
    2002;48:385–388.                                      from diffusion-sensitized measurements of           21:726 –732.

752                                                                                                             Radiology: Volume 249: Number 3—December 2008