United States Patent [19]
`Mugler, III et al.
`
`111111111111111111111111111111111111111111111111111111111111111111111111111
`US005245282A
`[II] Patent Number:
`[45] Date of Patent:
`
`5,245,282
`Sep. 14, 1993
`
`[75]
`
`[56]
`
`[54] THREE-DIMENSIONAL MAGNETIC
`RESONANCE IMAGING
`Inventors: John P. Mugler, III; James R.
`Brookeman, both of Charlottesville,
`Va.
`[73] Assignee: University of Virginia Alumni Patents
`Foundation, Charlottesville, Va.
`[21] Appl. No.: 723,230
`[22] Filed:
`Jun. 28, 1991
`Int. CI.s ............................................. G01R 33/20
`[51]
`[52] u.s. Cl ..................................................... 324/309
`[58] Field of Search ............... 324/300, 307, 309, 312,
`324/318, 322; 128/653.2
`References Cited
`U.S. PATENT DOCUMENTS
`4,797,616 1/1989 Matsui et aJ ........................ 324/309
`4,801,884 1/1989 Oppelt et aL ....................... 324/309
`4,818,942 4/1989 Rzedzian ............................. 324/312
`4,830,012 5/1989 Riederer .............................. 128/653
`4,833,407 5/1989 Holland eta!. ..................... 324/309
`4,836,209 6/1989 Nishimura ........................... 128/653
`4,843,321 6/1989 Sotak ................................... 324/309
`4,856.528 8/1989 Yang eta!. .......................... 128/653
`4,895,157 1/1990 Nambu ................................ 128/653
`4,90!,019 2/1990 Wedeen ............................... 324/309
`4,940,941 7/1990 Rzedzian ............................. 324/312
`4,982,161 1/1991 Twieg ................................. 324/309
`4,984,573 1/1991 Leunbach ............................ 128/653
`4,986,272 1/1991 Riederer eta!. .................... 128/653
`4,991,586 2/1991 Mueller eta!. ...................... 128/653
`4,993,075 2/1991 Sekihara eta!. .................... 382/6 K
`
`5,072,182 12/1991 Derby eta!. ........................ 324/309
`5.084,675 1/1992 Reinfelder eta!. ................. 324/309
`5,087,880 2/1992 Bruder eta!. ....................... 324/309
`5,105,152 4/1992 Pauly ................................... 324/309
`5,122,747 6/1992 Reiderer et al ..................... 324/309
`
`Primary Examiner-Michael J. Tokar
`Attorney, Agent, or Firm-Sheldon H. Parker
`ABSTRACT
`[57]
`A new three-dimensional (3D) MR imaging pulse se(cid:173)
`quence can produce over 100 high-resolution, high-con(cid:173)
`trast images in as little as 6 minutes of imaging time.
`Without additional imaging time, this same image data
`can be post-processed to yield high-resolution, high(cid:173)
`contrast images in any arbitrary orientation. Thus, this
`new pulse sequence technique provides detailed yet
`comprehensive coverage. The method of this invention
`relates to a preparation-acquisition-recovery sequence
`cycle. The flrst step is magnetization preparation (MP)
`period. The MP period can emply a series of RF pulses,
`gradient fleld pulses, and/or time delays to encode the
`desired contrast properties in the form of longitudinal
`magnetization. A data acquisition period includes at
`least two repetitions of a gradient echo sequence to
`acquire data for a fraction of k-space. A magnetization
`recovery period is provided which allows T1 and T2
`relaxation before the start of the next sequence cycle.
`The MP, data acquisition and magnetization recovery
`steps are repeated until a predetermined k-space volume
`is sampled.
`
`44 Claims, 7 Drawing Sheets
`
`PREPARE
`
`ACQUIRE
`
`RECOVER
`
`RF
`
`GS
`
`GR
`
`2D
`
`GP
`AOC L __ _J__j-·
`
`128l111ES (30)
`
`128 l1IIES (20)
`
`General Electric Co. 1006 - Page 1
`
`

`
`TRIGGER
`(OPTIONAL)
`
`,r MAGNETIZATION
`r..,
`PREPARA TION(S)
`
`RAPID
`GRADIENT ECHO
`ACQUISmON
`
`MAGNETIZATION
`RECOVERY
`(SECONDARY PREPARATIONS)
`
`1--
`
`~ •
`00
`•
`
`~ s(cid:173)a
`
`00
`tD
`
`'F' ...
`~~ ...
`~
`
`ACQUIRE A FRACTION
`OF K-SPACE VALUES
`
`REPEAT UNTIL ALL
`DATA IS COLLECTED
`
`FIG 1
`
`00 =(cid:173)
`
`tD
`
`tD -...
`
`0
`"""
`.......
`
`...
`01
`~
`~
`...
`01
`~
`QC
`~
`
`General Electric Co. 1006 - Page 2
`
`

`
`U.S. Patent
`
`Sep. 14, 1993
`
`Sheet 2 of 7
`
`5,245,282
`
`a
`
`en
`~
`I=
`,..
`~
`
`iS e
`m
`:e
`I=
`~
`,..
`
`C\1
`-
`(!J
`u.
`
`a:
`w
`>
`0
`0
`w
`a:
`
`w
`a:
`
`-:::»
`
`0
`0
`<(
`
`Q
`Ct)
`
`00
`T"""
`
`<(
`
`w a:
`0.. w
`a:
`D.
`
`0
`0 en
`
`r-
`
`u.
`a:
`
`(/)
`(!J
`
`a:
`(!J
`
`0..
`(!J
`
`0
`~
`
`General Electric Co. 1006 - Page 3
`
`

`
`U.S. Patent
`
`Sep. 14, 1993
`
`Sheet 3 of 7
`
`5,245,282
`
`FIG3A
`
`FIG3B
`
`General Electric Co. 1006 - Page 4
`
`

`
`U.S. Patent
`
`Sep. 14, 1993
`
`Sheet 4 of 7
`
`5,245,282
`
`FIG3C
`
`FIG 3D
`
`General Electric Co. 1006 - Page 5
`
`

`
`U.S. Patent
`
`Sep. 14, 1993
`
`Sheet 5 of 7
`
`5,245,282
`
`FIG4A
`
`FIG4B
`
`General Electric Co. 1006 - Page 6
`
`

`
`U.S. Patent
`
`Sep. 14, 1993
`
`Sheet 6 of 7
`
`5,245,282
`
`FIG4C
`
`FIG4D
`
`General Electric Co. 1006 - Page 7
`
`

`
`U.S. Patent
`
`Sep. 14, 1993
`
`Sheet 7 of 7
`
`5,245,282
`
`FIG5
`
`FIG6
`
`General Electric Co. 1006 - Page 8
`
`

`
`1
`
`THREE-DIMENSIONAL MAGNETIC
`RESONANCE IMAGING
`
`5,245,282
`
`2
`(3) the flexibility of the sequence structure to be adapted
`to different imaging requirements. The 3D MP RAGE
`technique can improve the imaging capabilities in some
`clinical areas (e.g., brain imaging) and provide new
`5 clinical capabilities in other areas (e.g., 3D abdominal
`imaging).
`The method of this invention relates to a preparation(cid:173)
`acquisition-recovery sequence cycle. The first step is
`magnetization preparation (MP) period. The MP period
`10 can emply a series of RF pulses, gradient field pulses,
`and/or time delays to encode the desired contrast prop(cid:173)
`erties in the form oflongitudinal magnetization. At least
`one contrast property can be encoded by the magnetiza-
`tion preparation step. For example, Tl-weighting com(cid:173)
`bined with one of spatial or chemical presaturation can
`be encoded by the magnetization preparation step.
`A data acquisition period includes at least two repeti(cid:173)
`tions of a gradient echo sequence to acquire data for a
`fraction of k-space.
`A magnetization recovery period is provided which
`allows Tl and Tl relaxation before the start of the next
`sequence cycle. The magnetization recovery period can
`have a time of zero. The time period employed for
`magnetization recovery can also be employed for mag(cid:173)
`netization preparation.
`The MP, data acquisition and magnetization recovery
`steps are repeated until a predetermined k-space volume
`is sampled.
`Advantageously, at least some of the preparation-
`acquisition-recovery sequences cycles are initiated by a
`trigger signal, whereby the sequence is synchronized
`with an external temporal event, such as respiration or
`heart beat.
`Some or all of the RF pulses and/or gradient pulses
`applied during any of the steps can serve the purpose of
`stabilizing responses of the apparatus (such as eddy
`currents). In addition, or instead of the foregoing, some
`or all of the RF pulses and/or gradient pulses can be for
`the purpose of stabilizing the magnetization system,
`e.g., oscillations in signal strength.
`The duration of any of the steps can be constant;
`alternatively, or in addition, the duration of at least one
`of the steps can vary from sequence cycle to cycle.
`Some or all of the RF pulses can be spatially and/or
`chemically selective. The spatially selectivity can be in
`two or three dimensions. A given pulse can combine
`spatial and chemical selection.
`Some or all of the RF pulses can be spatially and/or
`chemically non-selective.
`The gradient-echo sequence can employ gradient or
`RF spoiling to reduce or eliminate the effects of residual
`transverse coherences. The gradient-echo sequence can
`employ a partially or fully rephased gradient structure
`55 and can employ flip angles which are constant or which
`vary within a given data acquisition period and/or be(cid:173)
`tween data acquisition periods. The gradient-echo se(cid:173)
`quence can employ an echo time and/or repetition time
`which is selected from the group consisting of constant,
`varying within a given data acquisition period, varying
`between data acquisition period, and varying both
`within and between data acquisition periods.
`The gradient-echo sequence can employ a data sam(cid:173)
`pling period which is either constant, varies within a
`65 given data acquisition period, varies between data ac(cid:173)
`quisition periods, or which varies both within and be(cid:173)
`tween data acquisition periods. The gradient-echo se(cid:173)
`quence can employ either symmetric or asymmetric
`
`BACKGROUND OF THE INVENTION
`1. Field of the Invention
`The invention relates to magnetic resonance imaging,
`and more particularly to a rapid process for producing
`three-dimensional magnetic resonance imaging.
`2. Description Of The Prior Art
`Magnetic resonance imaging (MRI) is a non-invasive
`medical diagnostic imaging modality that can produce
`high-contrast tomographic images of the interior soft(cid:173)
`tissue structures of the human body without the use of
`ionizing radiation. In many imaging applications, MRI 15
`has replaced the competing technology of X-ray com(cid:173)
`puted tomography (CT) as the imaging method of
`choice. For an MRI examination, the subject is placed
`in a very strong static magnetic field, and the informa(cid:173)
`tion necessary to create the images is generated using a 20
`series of magnetic field gradient pulses and radio-fre(cid:173)
`quency (RF) pulses. The exact manner in which the
`gradient and RF pulses are applied is called the pulse
`sequence. By changing the pulse sequence, the relative
`appearance of different tissues and pathologies can be 25
`changed. Thus, the pulse sequence can be optimized to
`highlight certain pathological conditions, and even to
`create images of flow. There are literally an infinite
`number of possible pulse sequences. The potential vari(cid:173)
`ety of pulse sequences and the ability of different pulse 30
`sequences to produce images which highlight different
`types of information are major advantages ofMRI com(cid:173)
`pared to other techniques. Critical to the success and
`acceptance of MRI as a primary imaging modality has
`been the continued development of new pulse sequence
`techniques which have both improved the imaging
`capabilities in existing areas of clinical use and provided
`new clinical areas of application.
`The importance of rapid imaging techniques is dis(cid:173)
`cussed in the Manual of Clinical Magnetic Resonance 40
`Imaging, (CMRI) by Heiken et al., Raven Press, New
`York, 1991. It is therein explained that the impetus for
`the development of rapid imaging techniques has been
`primarily twofold: to improve the efficiency of clinical
`MRI and to decrease artifacts that arise from cardiac, 45
`respiratory, and other patient motion. The synopsis of
`the more important rapid imaging techniques discussed
`in CMRI, at pages 24 through 39, is incorporated herein
`by reference, as though set forth in detail. At page 31, it
`is noted that steady state GE images with short TRs and 50
`low flip angles provide a myelogram effect in which the
`spinal cord can be easily differentiated from surround(cid:173)
`ing CSF.
`
`35
`
`SUMMARY OF THE INVENTION
`It has now been found that a new three-dimensional
`(3D) MR imaging pulse sequence can produce over 100
`high-resolution, high-contrast images in as little as 6
`minutes of imaging time. Without additional imaging
`time, this same image data can be post-processed to 60
`yield high-resolution, high-contrast images in any arbi(cid:173)
`trary orientation. Thus, this new pulse sequence tech(cid:173)
`nique provides detailed yet comprehensive coverage.
`Compared to existing 3D MR imaging pulse sequences,
`our technique, called 3D MP RAGE, will potentially
`provide significant improvements in (1) the contrast and
`resolution that can be obtained in a given imaging time,
`(2) the variety of possible image contrast behaviors, and
`
`General Electric Co. 1006 - Page 9
`
`

`
`5,245,282
`
`10
`
`3
`sampling of the echo thereby potentially acquiring only
`a half echo. The signal can be acquired in the presence
`of a single constant applied gradient, and the remaining
`spatial dimensions can be phase-encoded (standard Fou(cid:173)
`rier transform phase encoding).
`Further, the gradient echo sequence can acquire a
`plane, or a fraction of a plane, of k-space data during
`each sequence cycle. Alternatively, the k-space data
`collected by the gradient-echo sequence during a given
`sequence cycle might not be contained in any plane.
`The temporal order in which the k-space data is col(cid:173)
`lected for each sequence cycle is determined based on
`achieving selected properties in the image, such as con(cid:173)
`trast, or selected properties of the corresponding point
`spread function. The temporal order of k-space data
`collection can be fixed or can vary from sequence cycle
`to cycle. The gradient-echo sequence can acquire a
`fixed or a varying amount of k-space data during each
`sequence cycle. The gradient-echo sequence can ac(cid:173)
`quire data in the presence of from one to three time- 20
`varying applied gradients or in the presence of two or
`three constant applied gradients, and any remaining
`spatial dimensions employ standard phase encoding.
`The gradient-echo sequence can employ predetermined
`gradient waveforms to compensate, in the sampled sig- 25
`nal, for phase shifts due to flow and/or motion. The
`compensations can be specifically designed for at least
`one of velocity, acceleration and higher orders of mo(cid:173)
`tion.
`The data acquisition can be in the absence of any
`applied magnetic field gradients and from two to three
`spatial dimensions are encoded using standard phase(cid:173)
`encoding. Thus, one dimension of the three or four
`dimensional data set, contains chemical shift informa-
`tion.
`
`4
`improved the imaging capabilities in existing areas of
`clinical use (e.g., brain imaging) and provided new clini(cid:173)
`cal areas of application (e.g., magnetic resonance angi(cid:173)
`ography). The new techniques can be classified into two
`5 general categories, those which improve the imaging
`capabilities in an existing area of clinical use (e.g., brain
`imaging), and those which provide imaging capabilities
`in a new clinical area (e.g., the development of magnetic
`resonance angiography techniques).
`The three-dimensional (3D) MRI technique of the
`present invention employs a magnetization preparation(cid:173)
`data acquisition-magnetization recovery cycle as the
`basic sequence element. Our new pulse sequence tech(cid:173)
`nique generalizes and extends the basic, prepare-
`15 acquire, philosophy introduced by Haase et al in 1989
`with the snapshot FLASH technique.
`By employing a distinct magnetization preparation
`period, the determination of the image contrast can be
`largely separated from the data acquisition. The image
`data is acquired using a rapid gradient-echo sequence.
`Additional control over the image contrast is provided
`by varying the duration of the magnetization recovery
`period. For convenience, reference to the new tech(cid:173)
`nique will be by the acronym 3D MP RAGE for 3-
`Dimensional Magnetization-Prepared Rapid Gradient(cid:173)
`Echo imaging.
`In experiments with 3D MP RAGE, high-quality 3D
`image sets (128 X 128 X 256 voxels) of the abdomen,
`were acquired showing minimal respiratory artifacts in
`30 just over 7 minutes (voxel size 2.7X2.7X2.7 mm3), and
`3D image sets (128 X 128 X 256 voxels) of the head
`showing excellent gray matter/white matter contrast in
`Jess than 6 minutes (voxel size 1.0 X 2.0 X 1.4 mm3). The
`technique of the instant invention can produce high-
`35 resolution 3D image sets of the abdomen with minimal
`respiratory artifacts in an imaging period acceptable for
`routine clinical use.
`3D MP RAGE can be applicable as a general screen(cid:173)
`ing pulse sequence for certain anatomical areas, and can
`40 result in significant reductions in patient exam time, thus
`providing increased patient throughput and decreased
`examination costs.
`Since the magnetization is sampled during a transient
`that is dependent on the tissue Tl relaxation times,
`many aspects of the theoretical description and optimi(cid:173)
`zation of the sequence are even more difficult than was
`the case for existing steady-state imaging techniques.
`Before the 3D MP RAGE technique could be made
`available for widespread clinical application, it was
`50 essential that the intricacies of the contrast behavior be
`fully understood.
`
`OBJECTS OF THE INVENTION
`An object of the invention is provide improved imag(cid:173)
`ing capabilities and to thereby provide increased patient
`throughput and reduced examination costs.
`
`DRAWINGS
`FIG. 1 is a schematic representation of 3D MP
`RAGE.
`FIG. 2 is a timing diagram for a Tl-weighted 3D MP 45
`RAGE sequence which employs a 180° pulse followed
`by a delay for preparation, and a FLASH gradient-echo
`sequence for data acquisition.
`FIGS. 3A-3D are images produced in accordance
`with Example I.
`FIGS. 4A-4D are images produced in accordance
`with Example II.
`FIG. 5 is an image produced in accordance with
`Example Ill.
`FIG. 6 is an image produced in accordance with 55
`Example IV.
`
`DESCRIPTION OF THE PREFERRED
`EMBODIMENTS
`Since its introduction into general clinical use in the 60
`early 1980's, magnetic resonance imaging (MRI) has
`become a very important diagnostic tool that is em(cid:173)
`ployed routinely in the course of patient care. In many
`areas, MRI has replaced X-ray computed tomography
`(CT) as the diagnostic imaging study of choice. Critical 65
`to the success and acceptance of MRI as a primary
`imaging modality has been the continued development
`of new pulse sequence techniques which have both
`
`1. Three-Dimensional Imaging
`2D versus 3D
`Clinical magnetic resonance images are usually ac(cid:173)
`quired as either a 2-dimensional (2D) plane or 3-dimen(cid:173)
`sional (3D) volume of data. In either case, the image
`data is generally presented as a series of 2D slices. The
`reference axis determining the slice direction in the 3D
`case is based on the mechanics of the pulse sequence.
`Each discrete intensity value (assuming a magnitude
`representation) in the image data represents an integral
`of the proton density, weighted by the Tl and T2 relax(cid:173)
`ation times, over a small volume (neglecting flow or
`other effects). For the standard Fourier transform imag(cid:173)
`ing technique, the data values are equally spaced along
`each of2 (or 3) mutually orthogonal axes corresponding
`to the read out direction and phase-encoding direc-
`
`General Electric Co. 1006 - Page 10
`
`

`
`5,245,282
`
`5
`tion(s). In the 2D case, the integrand corresponding to
`the two in-plane directions is proportional to the inverse
`Fourier transform of any filter function applied to the
`spatial frequency data in the given direction. In the ideal
`case, assuming the data is not windowed with a smooth- 5
`ing function, the integrand is the weighted proton den(cid:173)
`sity times a sine function, as disclosed in Bracewell RN.
`The Fourier Transform and its Applications, 2nd ed.,
`McGraw-Hill, New York, 1978. For the axis perpendic(cid:173)
`ular to the image plane, the integrand is the weighted 10
`proton density times the slice profile for the image,
`which is determined by the net effect of the radio fre(cid:173)
`quency (RF) pulse or pulses in the sequence, as dis(cid:173)
`closed in Rosen B. R., Pykett I. L., Brady T. J. Spin
`Lattice Relaxation Time Measurements in Two-Dimen- 15
`sional Nuclear Magnetic Resonance Imaging: Correc(cid:173)
`tions for Plane Selection and Pulse Sequence. J Comput
`Assist Tomogr 8, 195-199, 1984, Young I. R., Bydder
`G. M. Some Factors Involving Slice Shape which Af(cid:173)
`fect Contrast in Nuclear Magnetic Resonance (NMR) 20
`Imaging, Ann Radio! (Paris) 28, 112-118, 1985, and
`Young I. R., Bryant D. J., Payne J. A. Variations in
`Slice Shape and Absorption as Artifacts in the Determi(cid:173)
`nation of Tissue Parameters in NMR Imaging. Magn
`Reson Med 2, 355-389, 1985. If multiple 2D slices are 25
`acquired by time-multiplexing the acquisitions for dif(cid:173)
`ferent slice positions as is usually done in standard 2D
`clinical imaging, the profile for a given slice becomes
`increasingly distorted as the distance between adjacent
`slices is decreased, as disclosed in Kneeland J.B., 30
`Shimakawa A., Wehrli F. W. Effect of Intersection
`Spacing on MR Image Contrast and Study Time. Radi(cid:173)
`ology 158, 819-822, 1986, Crawley A. P., Henkelman R.
`M. A Stimulated Echo Artifact from Slice Interference
`in Magnetic Resonance Imaging. Med Phys 14, 35
`842-848, 1987, Kucharczyk W., Crawley A. P., Kelly
`W. M., Henkelman R. M. Effect of Multislice Interfer(cid:173)
`ence on Image Contrast in T2- and Tl-weighted MR
`Images. AJNR 9, 443-451, 1988, and Schwaighofer
`BW, Kyle KY, Mattrey RF. Diagnostic Significance of 40
`Interslice Gap and Imaging Volume in Body MR Imag(cid:173)
`ing. AJR 153, 629-632, 1989.
`The cross-talk between closely spaced slices can be a
`disadvantage of 2D multi-slice acquisitions for closely
`spaced or contiguous slices, but we note that a tremen- 45
`dous amount of research effort has been dedicated to
`optimizing RF inversion, excitation, and refocusing
`profiles to minimize slice-to-slice interference, as dis(cid:173)
`closed for example in Warren W. S., Silver M., in Ad(cid:173)
`vances in Magnetic Resonance, Academic Press, 12, 50
`248, 1988.
`In the 3D case (neglecting any effects of the RF
`pulses), the integrand for all directions is proportional
`to the inverse Fourier transform of the corresponding
`filter function in spatial frequency space. Assuming 55
`ideal conditions and no data windowing, the multiplica(cid:173)
`tive term (i.e., the point spread function or PSF) for the
`weighted proton density has the same form for each
`direction. This fact is advantageous if the image data is
`acquired with isotropic, or nearly isotropic, resolution 60
`and the 3D volume of data is reformatted to yield im(cid:173)
`ages in planes other than reference orientation. How(cid:173)
`ever, if the slice thickness (spacing in the second phase(cid:173)
`encoding direction) is large compared to the in-plane
`resolution, truncation artifacts arising from the side- 65
`lobes of the PSF will be significantly worse in the slice
`direction as disclosed in Carlson J., Crooks L., Orten(cid:173)
`dahl D., et al., and Signal-to-Noise Ratio and Section
`
`6
`Thickness in Two-dimensional versus Three-dimen(cid:173)
`sional Fourier Transform MR Imaging. Radiology 166,
`266-270, 1988.
`Truncation artifacts in the third dimension usually
`become pronounced with slice thicknesses greater than
`about 2 to 3 mm, as disclosed in Carlson J., Crooks L.,
`Ortendahl D., et al. Signal-to-Noise Ratio and Section
`Thickness in Two-dimensional versus Three-dimen(cid:173)
`sional Fourier Transform MR Imaging. Radiology 166,
`266-270, 1988.
`Three-dimensional volume techniques can provide
`several advantages over two-dimensional multi-slice
`techniques. As discussed above, the 3D acquisition in(cid:173)
`herently provides contiguous slices and the functional
`form of the slice profl.le does not change with the spac(cid:173)
`ing between the slices. If the 3D acquisition employs
`isotropic, or nearly isotropic, resolution, the volume
`data set can be reformatted to yield high-resolution
`contiguous image slices in any arbitrary orientation, as
`disclosed in Lai C-M, Lauterbur P. C. True Three-Di(cid:173)
`mensional Image Reconstruction by Nuclear Magnetic
`Resonance Zeugmatography. Phys Med Bioi 5,
`851-856, 1981, Buonanno F. S., Pykett I. L., Brady T.
`J., et al. Clinical Relevance of Two Different Nuclear
`Magnetic Resonance (NMR) Approaches to Imaging of
`a Low-Grade Astrocytoma. J Comput Assist Tomogr
`6, 529-535, 1982, and Pykett I. L., Buonanno FS, Brady
`T. J., Kistler J. P. True Three-Dimensional Nuclear
`Magnetic Resonance Neuro-Imaging
`in
`Ischemic
`Stroke: Correlation ofNMR, X-ray CT and Pathology.
`Stroke 14, 173-177, 1983.
`In a 3D acquisition, the signal-to-noise ratio increases
`as the square root of the number of slices, since the
`slices are acquired through phase-encoding. However,
`the use of a second phase-encoding direction generally
`increases the sensitivity of 3D images to motion induced
`artifacts.
`Whether 2D or 3D is more efficient in a given imag(cid:173)
`ing situation depends on the repetition time TR, which
`is chosen based on the desired contrast behavior and the
`properties of the pulse sequence, and the minimum time
`for an excite-acquire cycle, TRmin, which is also de(cid:173)
`pendent on the properties of the pulse sequence. The
`relative values of TR and TRmin determine how many
`different slice acquisitions can be time-multiplexed
`within TR. For the pulse sequence techniques in clinical
`use today, TRs greater than approximately 100 to 200
`ms are usually best suited for 2D multi-slice imaging,
`whereas TRs significantly less than 100 ms are best
`suited to 3D volume imaging. There is of course an
`intermediate region where a hybrid approach, multiple
`3D volume imaging, is applicable as disclosed in Wilk
`RM, Harms SE. Temporomandibular Joint: Multislab,
`Three-Dimensional Fourier Transform MR Imaging.
`Radiology 167, 861-863, 1988.
`3D Clinical Imaging
`In the early 1980's, 3D imaging results were reported
`for excised organs and human brains in-vivo. Initial
`human applications used saturation recovery and inver(cid:173)
`sion recovery sequences, both of which employed 90
`RF pulses for excitation, as disclosed in Buonanno FS,
`Pykett I. L., Brady T. J., et al. Clinical Relevance of
`Two Different Nuclear Magnetic Resonance (NMR)
`Approaches to Imaging of a Low-Grade Astrocytoma.
`J Comput Assist Tomogr 6, 529-535, 1982 and Pykett I.
`L., Buonanno FS, Brady T. J., Kistler J.P. True Three(cid:173)
`Dimensional Nuclear Magnetic Resonance Neuro-
`
`General Electric Co. 1006 - Page 11
`
`

`
`5,245,282
`
`7
`8
`Fast Three-Dimensional MR Imaging of the Knee:
`Imaging in Ischemic Stroke: Correlation of NMR,
`Comparison with Arthroscopy. Radiology
`166,
`X-ray CT and Pathology. Stroke 14, 173-177, 1983.
`865-872, 1988, Spritzer C. E., Vogler J. B., Martinez S.,
`To achieve the desired contrast properties with these
`et a!. MR Imaging of the Knee: Preliminary Results
`sequences, TRs of 200 ms or longer were necessary. A
`whole-head isotropic high-resolution (1 to 3 mm) data 5 with a 3DFT GRASS Pulse Sequence. AJR 150,
`set required imaging 3D volume acquisitions were very
`597-603, 1988, Haggar A.M., Froelich J. W., Hearshen
`D. 0., Sadasivan K. Meniscal Abnormalities of the
`high, the development and refinement of 2D multi-slice
`methods, combined with the relatively long imaging
`Knee: 3DFT fast-scan GRASS MR Imaging. AJR 150,
`times required for high-resolution large-volume 3D
`1341-1344, 1988 and SolomonS. L., Totty W. G., Lee
`acquisitions, diminished clinical interest in 3D tech- 10 J. K. MR Imaging of the Knee: Comparison of Three-
`Diques for several years.
`Dimensional FISP and Two-Dimensional Spin-Echo
`For the in-plane image matrix sizes commonly em-
`Pulse Sequences. Radiology 173, 739-742, 1989, Harms
`ployed (128 or 256), a TR ofless than 100 ms is required
`SE, Flamig D. P., Fisher C. F., Fulmer J. M. New
`for high-resolution large-volume (e.g., 64 or more Method for Fast MR Imaging of the Knee. Radiology
`phase-encoding steps in the third dimension) 3D image 15 173, 743-750, 1989. Other applications include magnetic
`sets to be acquired in clinically reasonable times (less
`resonance angiography.
`than approximately 15 minutes). This sequence require-
`The 3D short-TR gradient-echo sequences can be
`ment was met with the introduction in the mid-1980s of
`divided into two general categories, those which em-
`the short-TR, partial flip angle gradient-echo sequen-
`ploy a steady state of only the longitudinal component
`ces, such as FLASH, FFE, GRASS, FAST and FISP. 20 of the magnetization vector (e.g., FLASH, FFE) and
`FLASH is disclosed in Haase A, Frahm J, Matthaei D,
`those which employ a steady state of the complete mag-
`et a!., FLASH Imaging. Rapid NMR Imaging Using
`netization vector (e.g., GRASS, FAST, FISP). The
`Low Flip-Angle Pulses. J Magn Reson 67, 258-266,
`major practical difference between the two sequence
`1986, and FFE is disclosed in Van der Meulen P.,
`categories is the resulting image contrast properties as
`Groen J.P., Cuppen J. J. M. Very Fast MR Imaging by 25 disclosed in van der Meulen P, Groen J.P., Tinus A.M.
`Field Echoes and Small Angle Excitation. Magn Reson
`C., Bruntink G. Fast Field Echo Imaging: An Over-
`Imaging 3, 297-299, 1985. GRASS is disclosed in Utz J.
`view and Contrast Calculations. Magn Reson Imaging
`A., Herfkens R. J., Glover G., Pelc N. Three Second
`6, 355-368, 1988) and Tkach J. A., Haacke E. M. A
`Clinical NMR Images Using a Gradient Recalled Ac-
`Comparison of Fast Spin Echo and Gradient Field
`quisition in a Steady State Mode (GRASS). Magn 30 Echo Sequences. Magn Reson Imaging 6, 373-389,
`1988. It is important to note that the 3D implementa-
`Reson Imaging 4, 106, 1986 (abstract), and FAST is
`disclosed in Gyngell ML. The Application of Steady-
`tions of the longitudinal steady-state sequences, which
`State Free Precession in Rapid 2DFT NMR Imaging:
`are employed if Tl-weighted contrast is desired, have
`FAST and CE-FAST Sequences. Magn Reson Imaging
`been prone to slice-to-slice intensity banding artifacts as
`6, 415-419, 1988. FlSp is disclosed in Oppelt A., Grau- 35 disclosed in Wood ML, Runge V. M. Artifacts Due to
`Residual Magnetization in Three-Dimensional Mag-
`mann R., Barfuss H., et a!. FISP-a New Fast MRI
`Sequence. Electromedica 54, 15-18, 1986. For example,
`netic Resonance Imaging. Med Phys 15, 825-831, 1988.
`a TR of 15 ms and a flip angle of ISO produced 1283
`In these sequences, some type of spoiling is employed to
`image sets of human hands and feet in only 4 minutes, as
`destroy the coherence of the transverse magnetization
`disclosed in Frahm eta!. (Frahm J., Haase A., Matthaei 40 after the echo signal is sampled. Therefore, the trans-
`verse magnetization generated by a give]! excitation
`D. Rapid Three-Dimensional MR Imaging Using the
`FLASH Technique. J Comput Assist Tomogr 10,
`pulse contributes only to the signal measured in the
`363-368, 1986). Three-dimensional sequences, domi-
`echo period immediately following the pulse. This is the
`ideal case, and if the spoiling is incomplete the residual
`nated by the 3D gradient-echo techniques, have shown
`promising results for clinical application in the head as 45 transverse magnetization may create artifacts in the
`disclosed in Runge V. M., Wood ML, Kaufman D. M.,
`image. Traditionally, various combinations of magnetic
`et a!. FLASH: Clinical Three-Dimensional Magnetic
`field gradients have been employed in an attempt to
`Resonance Imaging. Radiographies 8, 161, 1988, Hu X.
`eliminate these artifacts as disclosed in Wood ML,
`P., Tan K. K., Levin D. N., et a!. Three-Dimensional
`Runge V. M. Artifacts Due to Residual Magnetization
`Magnetic Resonance Images of the Brain: Application 50 in Three-Dimensional Magnetic Resonance Imaging.
`to Neurosurgical Planning. J Neurosurg 72, 433-440, Med Phys 15, 825-831, 1988 and Frahm J., Hanicke W.,
`1990), in the spine, as disclosed in Gallimore G. W. Jr, Merboldt K-D. Transverse Coherence
`in Rapid
`Harms S. E. Selective Three-Dimensional MR Imaging
`FLASH NMR Imaging. J Magn Reson 72, 307-314,
`of the Spine. J Comput Assist Tomogr 11, 124-128,
`1987 and Wood M. L., Silver M, Runge V. M. Optimi-
`1987 and Sherry C. S., Harms S. E., McCroskey W. K. 55 zation of Spoiler Gradients in FLASH MRI. Magn
`Spinal MR Imaging: Multiplanar Representation from a
`Reson Imaging 5, 455-463, 1987 and Crawley A. P.,
`Single High Resolution 3D Acquisition. J Comput As-
`Wood M. L., Henkelman R. M. Elimination of Trans-
`sist Tomogr 11, 859-862, 1987, and Tsuruda J. S., Nor-
`verse Coherences in FLASH MRI. Magn Reson Med 8,
`man D., Dillon W., eta!. Three-Dimensional Gradient-
`248-260, 1988. However, gradients alone generally
`Recalled MR Imaging as a Screening Tool for the Diag- 60 have been found to be incapable of totally preventing
`nosis of Cervical Radiculopathy. AJR 154, 375-383,
`the artifacts in the 3D case.
`1990. The use in joints, is disclosed in Wilk R. M.,
`More recently, RF spoiling has been suggested as a
`Harms S. E. Temporomandibular Joint: Multislab,
`method to eliminate these transverse coherence arti-
`Three-Dimensional Fourier Transform MR Imaging.
`facts, as disclosed, for example in Crawley A. P., Wood
`Radiology 167, 861-863, 1988, Harms S. E., Muschler 65 M. L., Henkelman R. M. Elimination of Transverse
`Coherences in FLASH MRI