Lumbar Spinal Imaging in Radicular Pain and Related Conditions

Understanding Diagnostic Images in a Clinical Context

Bearbeitet von
J.T Wilmink

1. Auflage 2009. Buch. x, 161 S. Hardcover

ISBN 978 3 540 93829 3
Format (B x L): 19,3 x 26 cm

Weitere Fachgebiete > Medizin > Sonstige Medizinische Fachgebiete > Radiologie,
Bildgebende Verfahren
Zu Inhaltsverzeichnis

schnell und portofrei erhältlich bei

Die Online-Fachbuchhandlung beck-shop.de ist spezialisiert auf Fachbücher, insbesondere Recht, Steuern und Wirtschaft.
Im Sortiment finden Sie alle Medien (Bücher, Zeitschriften, CDs, eBooks, etc.) aller Verlage. Ergänzt wird das Programm
durch Services wie Neuerscheinungsdienst oder Zusammenstellungen von Büchern zu Sonderpreisen. Der Shop führt mehr
als 8 Millionen Produkte.
 Imaging Techniques for the Lumbar Spine:
Conventional Radiology, Computed 2
Tomography; Magnetic Resonance Imaging

2.1 Introduction All techniques presently employed for spinal imag-

ing have shortcomings. Conventional X-ray images
All imaging techniques have one feature in common: the have the drawback that potentially harmful radiation is
basis is the interaction between energy and matter. This employed, in addition to possessing a limited contrast
applies even to a conventional photograph: light (elec- resolution. In the early decades of the last century,
tromagnetic radiation in the visible wavelength spec- various methods were developed to artificially enhance
trum) is reflected with different frequencies (colours) image contrast by injecting contrast substances with
and intensities (brightness) from the surface of an object, very low (air) or high radiographic density (usually
thus, producing an image visible to our eyes. This image iodinated fluids) into various soft tissue structures or
can then be reproduced on photographic film by a cam- compartments. In the spine, myelography is the best
era, or captured on canvas by an artist. known of these techniques.
In a medical diagnostic setting, ultrasound waves can The development of new diagnostic methods, such
be reflected from tissue interfaces within the body to pro- as computed tomography and magnetic resonance
duce an echographic image. Electromagnetic energy in imaging, has resulted in a dramatic improvement in
the high-energy X-ray part of the spectrum is capable of low-contrast resolution, coupled with the advantages
passing through the human body but is not entirely unaf- provided by sectional (tomographic) imaging. The
fected: the X-ray photons are weakened (attenuated) to a downside is an increase in irrelevant detail demon-
varying degree depending on their wavelength (hard- strated by these improved techniques. This applies par-
ness) on the one hand, and the electron density and thick- ticularly to spinal imaging. Even conventional X-ray
ness of tissues within their path on the other. The residual films of the spine often demonstrate age-related and
radiation which has passed through the body is registered degenerative changes which are not necessarily associ-
by an X-ray film or another type of photon detector, and ated with the presence of disease. An MRI study can
the distribution of grey shades (contrast) in the resulting present an even greater abundance of morphologic
image represents local variations in the tissue density. details whose pathologic relevance is unclear. False-
Besides being reflected from, or transmitted through positive interpretation of an incidental finding is an
the body, energy can also be emitted from the body ever-present pitfall in all imaging studies, and this is
itself, for instance, by injecting a substance containing especially the case when insufficient attention is paid
a radioactive isotope into the body. This principle is to the correlation of high-resolution CT and MR imag-
the basis of nuclear medical imaging techniques. ing findings with clinical signs and symptoms.
Another emission-based technique is magnetic reso-
nance imaging (MRI), in which the protons incorporated
in water molecules of the body tissues emit radiofre-
quency (RF) signals under the influence of a combina- 2.2 Conventional X-ray Studies
tion of a magnetic field enclosing the body and RF
energy which is beamed into the body from an external Plain films of the spine offer a quick and inexpensive
source, causing the protons to “resonate” in electromag- evaluation of bony structures and are frequently used
netic terms. as an initial screening examination in, for instance,

J. T. Wilmink, Lumbar Spinal Imaging in Radicular Pain and Related Conditions 9

DOI: 10.1007/978-3-540-93830-9_2, © Springer-Verlag Berlin Heidelberg 2010
10 2 Imaging Techniques for the Lumbar Spine

suspected fractures, malalignment, and congenital spi- Lumbar myelography (syn. radiculography, cau-
nal defects. Abnormal spinal curves can be assessed in dography). In this examination an iodinated radiologic
scoliosis and the anatomy of individual vertebrae can contrast fluid is injected into the dural sac so that the
be defined, although superimposition of anatomical cerebrospinal fluid is opacified, outlining the dural sac,
structures is a problem. Spondylolysis and spondylolis- the dural root sleeves and their contents (Bates and
thesis are well demonstrated. Spinal metastases can be Ruggieri 1991). Structures of interest are the conus
detected on plain X-ray films, but only in a late stage, medullaris of the spinal cord, whose tip is located
when cortical bony structures of the vertebrae are approximately at the L1–2 level, and the nerve roots
affected, or the vertebra is deformed or collapsed. forming the cauda equina which originate from the
Manifestations of spondylodiscitis are also detected conus medullaris and traverse the lumbar dural sac in
relatively late. craniocaudal direction. These nerve roots exit the dural
At present, plain film spinal imaging is still ordered sac by way of a dural root sleeve which accompanies
frequently in patients presenting with low back pain and the emerging dorsal and ventral root fibres over a vari-
neck pain, but the diagnostic value of the examination able distance (see Chap. 3). Lumbar disc herniations
in the evaluation of such complaints is low. Contrast which are located in the central, paracentral and subar-
resolution in conventional X-ray images is limited: only ticular regions of the spinal canal (see Chap. 4) can
four tissue densities, namely bone, water, fat, and air, produce impressions upon the dural sac and displace-
can be distinguished and soft tissue pathology such as a ment of the intradural nerve roots, as well as cut-off of
disc herniation cannot be visualised. On the other hand, contrast filling of the root sleeve (Fig. 2.1). Sometimes
so-called degenerative features such as disc space nar- also swelling of the nerve root proximal to the site of
rowing, spondylosis, and spondylarthrosis can be dem- compression is seen. The myelographic image of the
onstrated in asymptomatic as well as symptomatic nerve root ends when it leaves the contrast-filled suba-
individuals (Fullenlove and Williams 1957). rachnoid space. Thus, lateral disc herniations com-
The diagnostic yield of plain film studies in low pressing the dorsal root ganglion or nerve ramus inside
back pain is very limited unless so-called red flags or outside the intervertebral foramen, and which are
(indicators for specific disease conditions such as neo- reported to occur in around 10% of cases, (Abdullah
plasm, disc herniation or infectious disease) are pres- et al. 1988), will frequently be missed by myelography
ent (Staiger et al. 1999). As mentioned above, however, (Jackson and Glah 1987).
the sensitivity for early detection of specific pathology Contrast myelography is not a very invasive proce-
by plain films is low, and in such cases alternative tech- dure, but it is not completely innocuous (Bates and
niques with higher sensitivity, such as CT or MRI, are Ruggieri 1991; Wilmink et al. 1984), Even in experi-
preferable. enced hands, a lumbar puncture followed by injection
A plain film examination of the lumbar spine usu- of contrast fluid may be difficult and painful, especially
ally consists of a lateral and a postero-anterior view. when the dural sac is constricted or collapsed and the
Oblique views are sometimes performed of the isth- nerve roots are crowded together by a large herniation
mus region in case of spondylolysis, but these substan- or by narrowing of the spinal canal at the puncture site.
tially increase the X-ray dose to the patient, and are not The iodised oils which were initially employed for
always necessary. Studies of the spine in flexion myelography frequently gave rise to adhesive arach-
(kyphosis) and extension or retroflexion (lordosis) can noiditis resulting in crippling back complaints. The
be used in the assessment of post-traumatic or degen- water-soluble contrast media which were later intro-
erative instability. duced produced better images of the root sleeves but
the first generation of these agents possessed a high
osmolality and neurotoxicity and could also cause
adhesive arachnoiditis (Skalpe 1978). Modern low-
2.2.1 Contrast Studies: osmolality contrast media do not share these severe side
effects.
The following conventional X-ray studies featuring con- Nowadays, the most common indication to perform
trast injection are presently still performed in the lum- contrast myelography is when MRI is contraindicated
bosacral spine: or not available, and when CT does not provide an
2.2 Conventional X-ray Studies 11

Fig. 2.1 Lumbar myelogram (radiculogram, caudogram). Right Fig. 2.2 Lateral projection of L5-S1 discogram. Note opacifica-
oblique projection centred on the L4 vertebra, with a water- tion at site of nucleus pulposus, leakage of contrast medium to
soluble radiologic contrast medium outlining the dural sac, extruded disc (arrow)
intradural nerve roots and root sleeves. Note ventrolateral filling
defect at L4–5 disc level, mainly caused by swelling of distal
intradural L5 root (arrow), with non-filling of L5 root sleeve. disc. The purpose of this is twofold. Firstly the increase
Medial displacement of intradural S1 root without compression, in intradiscal pressure caused by the injection may
normal filling of S1 root sleeve (arrowheads)
reproduce or exacerbate the patient’s pain complaints,
thus confirming that the disc in question is the source of
adequate image of the dural sac and of possible intra- the pain. Secondly, X-ray images can show penetration
dural pathology. In these cases the conventional myel- of the contrast medium into fissures and defects in the
ographic study will almost invariably be followed by annulus fibrosus, and sometimes also into herniated disc
CT myelography (see below). material (Fig. 2.2). Disc herniations and nerve root
Myelography can be employed to produce images of compression are diagnosed more accurately by MRI
the dural sac and the cauda equina in the upright pos- and CT however.
ture, or in lumbar flexion and extension (Penning and Discography is a controversial diagnostic proce-
Wilmink 1981). dure, with outspoken proponents as well as antago-
Discography. This examination technique is employed nists. The examination can be tedious and unpleasant
primarily to localise painful intervertebral discs respon- for the patient, especially when muliple disc levels are
sible for lumbago or low back pain, and not to diagnose studied. Discography is used to localise painful discs,
lumbosacral nerve root compression causing sciatica. but there are reservations because false-positive pain
A water-soluble iodinated radiologic contrast medium responses can occur, even when care is taken to apply
is injected into the nucleus pulposus of the intervertebral a low injection pressure (Carragee et al. 2006), and
12 2 Imaging Techniques for the Lumbar Spine

subjective pain responses at discography should be normal and abnormal soft tissue features without the
interpreted with special caution in patients with chronic necessity of contrast injection into the dural sac
pain, social stressors and psychological disturbances (Fig. 2.4). Visualisation of intradural details by uncon-
(Carragee and Hannibal 2004). Annular tears or fis- trasted CT is limited; the spinal cord can sometimes be
sures can be demonstrated by discography, but MRI seen faintly, and intradural nerve roots not at all.
has shown these to occur in asymptomatic individuals CT and myelography are complementary techniques:
as well as in low back pain sufferers with painful discs the first is more suitable for assessing the cause of radic-
(Stadnik et al. 1998). ular complaints, herniated disc, spinal stenosis etc., while
Two other contrast examinations, peridurography the second is better for imaging the effect, the com-
and epidural venography are no longer performed. pressed intradural nerve root (Wilmink 1989). Techniques
These techniques relied on opacification of the epidu- which combine both these features are CT myelography
ral space itself, or the veins in this space respectively, and MRI with MR myelography (see below).
by contrast injection, either directly into the spinal CT can also be combined with discography to pro-
canal via the sacral hiatus (Luyendijk and van duce CT discographic images, thus improving the sen-
Voorthuisen 1966) or into the paraspinal and interver- sitivity with which small annular tears can be detected.
tebral veins via catheterisation of the iliac veins The sensitivity of CT for bony vertebral pathology
(Wilmink et al. 1978). The diagnostic principle of such as metastasis and fracture is better than that of
these methods was to demonstrate disc herniations by plain films.
their compressive effect on the epidural structures, and A significant development has been the introduc-
thus provide an alternative to lumbar myelography. tion of multi-slice spiral CT scanning with multi-planar
The contrast opacification was too irregular and unreli- reformatting. This technique permits rapid scanning of
able however, and with the advent of newer imaging a large tissue volume by a thin continuous spiral or
techniques these methods were relegated to obscurity. helical section, and this has proven to be of special
value, for instance, in case of spinal trauma where sub-
tle fractures and dislocations, especially in the poste-
rior spinal elements, can be detected with an ease and
2.3 X-Ray Computed Tomography accuracy unrivalled by any other imaging method.
CT can provide an acceptable diagnostic alternative
Computed tomography CT (Hounsfield 1973) revolu- to MRI in many cases with disc herniation or spinal
tionised medical imaging by its introduction in the stenosis (Figs. 2.3, 2.4). Soft tissue resolution by CT,
1970s. Three innovations were combined: however, is less than when MRI is employed, and some
disc herniations can be overlooked (see Sect. 2.4).
• Acquisition of sectional (tomographic) images by
Anatomical detail is also less in reformatted sagittal
the use of an X-ray tube rotating around the patient.
CT images when compared to direct sagittal MRI cuts;
This made it possible to study spinal anatomic rela-
in addition bone marrow pathology annular fissures
tionships in the axial plane which could not previ-
and other subtle changes cannot be detected by CT.
ously be visualised. A much better insight was
Intraspinal details usually are less well-depicted by CT
obtained in the morphology and classification of,
at the level of the vertebral pedicles and lamina, where
for instance, spinal stenosis (see Chap. 4).
the dural sac is entirely surrounded by a ring of bony
• Detection of smaller differences in X-ray attenua-
structures (see Chap. 3), where there is little epidural
tion (tissue density) by using more sensitive scintil-
fat to outline the dural sac and migrated disc fragments
lation detectors instead of an X-ray film, thus,
may be missed. At the disc level the structures border-
greatly improving soft tissue contrast resolution.
ing the spinal canal are ligamentous and less dense,
• Image reconstruction by a computer algorithm per-
and there is usually more intraspinal fat present to act
mitting selection of window and level settings
as a natural contrast agent.
appropriate for viewing bony or soft tissue struc-
CT has for many years formed the mainstay of diag-
tures as required.
nostic imaging in patients with radicular pain and related
The improved contrast resolution of CT made it pos- conditions, despite the drawback that compression of
sible to image disc herniations and other intraspinal the intradural nerve root could not be visualised directly.
2.3 X-Ray Computed Tomography 13

c d

Fig. 2.3 Spiral CT study with multiplanar reformat. Patient level of posterior disc (b), L5 endplate (c), and L5 lateral recess
with L4–5 disc herniation (a) mid-sagittal reformat showing (d) show extrusion migrating laterally towards right L5 root in
extrusion migrating below disc level (arrow), axial 2 mm cuts at lateral recess (arrow). Left L5 root normally outlined by fat

Indirect evidence of compression of the intradural root the dural sac at disc level, as well as displacement by the
in non-myelographic CT images can be derived from herniation and disappearance of the epidural fat adja-
features such as flattening of the ventrolateral angle of cent to the dural sac (Wilmink 1989).
14 2 Imaging Techniques for the Lumbar Spine

Fig. 2.4 CT of lumbar vertebra in developmental spinal steno-

sis. Note short pedicles and shallow spinal canal

b
In order to limit the radiation dose, only the lower
three lumbar disc levels are routinely scanned in sus-
pected lumbar disc herniation and this involves a risk
of missing a herniation which is situated at a higher
lumbar level (see also Chap. 3).
CT slices can either be acquired in separate sets with
the CT gantry angulated parallel to each disc, or as a
continuous or overlapping series of parallel slices. The
advantage of the first method is that there is less distor-
tion of sagittal dimensions of the spinal canal, but the
disadvantage is that portions of the spinal canal between Fig. 2.5 Slice positioning and angulation in spinal CT. When
the slice sets may be skipped, and migrated disc frag- slice sets are angulated parallel to each disc (a), there is fre-
ments in this region easily overlooked (Fig. 2.5). quently a “blind spot” at mid-vertebral level (asterisk) which is
not imaged but which may contain a migrated disc fragment.
Slice thickness in non-spiral lumbar CT is usually When slices are acquired in true axial plane without craniocau-
3–5 mm, with thinner 2 or 1 mm slices preferred when dal angulation (b) dimensions of the spinal canal are distorted:
spiral CT scanning with multi-planar reformatting is to bony sagittal diameter measured as 16.9 mm in true axial plane
be performed. The spiral datasets are acquired without compared to 15.6 mm in plane parallel to L4–5 disc
gantry angulation.
possible to clearly discern the spinal cord and individual
nerve roots within the dural sac, which is not possible
on non-contrasted CT images. These structures are pre-
2.3.1 CT Myelography sented in the axial plane, which is not possible on con-
ventional myelograms. The conventional myelographic
This technique which was first reported by Di Chiro and image of the nerve root ends after its departure from the
Schellinger (1976) is a useful adjunct to conventional dural root sleeve, whereas with CT myelography the
myelography as well as to non-contrast CT. The pres- root can first be followed through the CSF compartment
ence of an intrathecal contrast medium makes it where it is outlined by the contrast medium in the
2.4 Magnetic Resonance Imaging (MRI) 15

2.4 Magnetic Resonance

Imaging (MRI)

Imaging by nuclear magnetic resonance (NMR)

(Mansfield and Maudsley 1977), presently better known
as magnetic resonance imaging or MRI, produces com-
puted tomographic sections similar to X-ray CT, but
makes use of a different imaging principle. In X-ray
CT, image contrast is derived from differences in X-ray
attenuation due to variations in electron density in vari-
ous structures within the body. In MRI the protons of
the body are induced to act as radiofrequency (RF)
transmitters by being positioned in a magnetic field
and subjected to RF energy directed from an antenna,
or coil. The electromagnetic resonance of the protons
Fig. 2.6 CT myelography. 4.5 mm CT section through L5 end-
is analogous to the resonance of a tuning fork when
plate after intradural contrast injection. Note dural sac and root
sleeves opacified by contrast medium, with intradural details exposed to sound of the appropriate frequency. The
shown which are not depicted in plain CT: ventral and dorsal RF signals from the protons can be manipulated or
root components seen as dark dots within contrast-filled S1 root “weighted” to selectively amplify signal intensity of
sleeve (arrow) and S2 root in dural sac (arrowhead). Curved
various substances and structures within the body, and
arrow indicates right extradural L5 spinal nerve ramus exiting
foramen and outlined by fat and seen only faintly at wide win- are spatially encoded to produce an image.
dow setting An MR image in which contrast is dependent on
differences in longitudinal magnetic relaxation times
arachnoid space, then more distally through the fora- as defined by so-called T1 values between various tis-
men and beyond, where it is outlined by fat (Fig. 2.6). sues is called “T1-weighted”. When image contrast is
When MRI is not available or contraindicated, CT predominantly determined by differences in transverse
myelography can be used for the detection of intraspi- magnetic relaxation values (T2), the image is called
nal space occupying lesions: intradural (intramedul- “T2-weighted”.
lary neoplasm or cyst, extramedullary meningioma or For spinal imaging, MRI has significant advantages
nerve root tumour) extradural (disc herniation, verte- over CT: soft tissue contrast resolution is better (Fig. 2.7)
bral neoplasm or extradural hematoma) or both (dumb- and there are no artefacts due to high-density skeletal
bell schwannoma). Cord atrophy or transection can structures. The signal intensity of bony spinal struc-
also be demonstrated, but spinal cord lesions without tures is less bright in MR than in CT images, and the
mass effect, such as cord infarct or multiple sclerosis latter method is better for diagnosing bony cortical
plaques can only be visualised by MRI. Other draw- lesions such as in vertebral fractures. Although some
backs of CT myelography compared to MRI are the consider that spinal stenosis is better demonstrated
necessity for intrathecal contrast injection and the by CT than by MRI, in fact cortical bone can be well-
employment of ionising X-rays. distinguished as a dark line bordering the brighter
On the other hand, spatial resolution in CT myelo- bone marrow in T1-weighted MR images. Also, spi-
graphic images is usually better than in axial MR images, nal stenosis has an important ligamentous as well as
and this is especially important in diagnosis of nerve root a bony component. Even in cases with severe devel-
compression, for instance in the lateral recess of the spi- opmental stenosis, compression of the dural sac and
nal canal, where MRI often does not provide sufficient the cauda equina takes place mainly at the level of
detail. CT is more accurate than MRI for the assessment the intervertebral disc, and is not due only to bony
of calcified herniations as well as bony spurs emanating narrowing of the spinal canal but rather to superim-
from the vertebral bodies or encroaching upon the fora- posed ligamentous encroachment by bulging of the
men as well as for demonstrating the presence of gas in annulus fibrosus and hypertrophy of the flaval liga-
a degenerated disc or joint (see Fig. 4.18). ments (Fig. 2.8, see also Chap. 4, Fig. 4.1.b and d), and
16 2 Imaging Techniques for the Lumbar Spine

a b

c d

Fig. 2.7 CT compared to MRI. CT and MRI of same large L4–5 also well-depicted in sagittal T1-(c) and T2-weighted images (d).
disc extrusion. Axial 5 mm CT section (a) shows apparently nor- Note that in retrospect remnant of collapsed dural sac is very
mal L4–5 disc. Axial T1-weighted 4.5 mm MRI section (b) shows faintly visible on CT section
large extrusion almost completely collapsing dural sac (arrow),
2.4 Magnetic Resonance Imaging (MRI) 17

b c

Fig. 2.8 T1-weighted axial MR image (a) in patient with loca- details such as facets on the one hand, as well as soft tissue struc-
lised narrowing of spinal canal at L4–5 showing almost com- tures such as annulus fibrosus (arrow) and flaval ligament
plete CSF block on sagittal images (b, c) best seen on (arrowhead ) on the other
T2-weighted image (c). Note that axial cut clearly shows bony
18 2 Imaging Techniques for the Lumbar Spine

sometimes deformation of the spinal canal by degen- the sacral canal and in the intervertebral foramina has the
erative anterolisthesis. highest signal intensity in T1-W images of the spine, and
Subtle changes in shape and composition of the spi- T1-weighting is popularly said to produce a “fat image”.
nal cord can be demonstrated by MRI, and intradural In such images fat acts as a natural contrast medium, and
nerve roots can be seen without the necessity for con- structures bordered by fat are clearly outlined: dorsal and
trast injection into the dural sac (MR myelography, see caudal borders of the lumbar dural end-sac, foraminal
below). MR images can be acquired in any plane desired, borders and intraforaminal contents such as dorsal root
and are superior to reformatted sagittal or coronal CT ganglia, as well as laterally migrated disc extrusions.
images of the spine, especially for showing soft tissues. Due to the low signal intensity of CSF on T1-W
The largest single indication for spinal MR imaging images, the intradural nerve roots can only be faintly dis-
is presently in degenerative spinal disease, usually per- tinguished. The spinal cord can be seen, but not as well
formed to diagnose a possible disc herniation. as in T2-W images. The lack of contrast between the
A number of options for lumbar spinal MR imaging posterior disc surface and the anterior border of the dural
will now be discussed. It will be clear that there are sac, both dark, sometimes makes it difficult to discern
many methods to produce good-quality diagnostic spi- disc herniations in this location on T1-weighted images.
nal images (Ruggieri 1999), and the selection depends In the lumbar spine T1-W images are usually
upon the characteristics of the MRI system employed acquired by a so-called spin-echo (SE) or fast spin-echo
and personal preferences of the radiological user and (FSE) sequence.
clinical end-user. An example of a typical set of imag-
ing sequences for use in lumbar degenerative disc dis-
ease is given in Fig. 2.9.
2.4.2 T2-Weighted (T2-W) and
The discussion of the various techniques set out
below is not intended to be exhaustive, and reflects the T2*-Weighted (T2*-W) Images
personal experience of this author. For more detailed
information regarding technical aspects of MR imag- The bright signal intensity of water (CSF, nucleus pul-
ing and the various acquisition sequences mentioned posus) predominates in images with T2-weighting,
below, the reader is referred to specialised texts deal- and these are sometimes known as “water images”
ing with these subjects. A review of recent develop- (Fig. 2.9b). For this reason intradural features such as
ments in spinal MR imaging sequences is given by spinal cord and cauda equina are best seen with this
Vertinsky et al. (2007). technique, as are disc herniations impinging upon the
CSF-filled dural sac or the root sleeve.
T2-W lumbar spinal images are at present generally
acquired with a 2D fast spin-echo, syn. turbo spin-echo
2.4.1 T1-Weighted (T1-W) Images (FSE, TSE) sequence. Conventional spin-echo (CSE)
sequences are no longer routinely used because of the
In these images the CSF-filled dural sac is darker than lengthy scanning times necessary to produce sufficient
the disc and vertebrae (Fig. 2.9a). Normal adult bone T2-weighting with this technique. CSE and FSE do not
marrow has a light grey shade, with somewhat brighter produce identical T2-weighted images; in a CSE
signal intensity than that of the intervertebral disc. The sequence epidural fat and bone marrow fat have low
fat seen in the epidural pockets dorsal to the dural sac, in signal intensity, while FSE produces a much higher fat

Fig. 2.9 Images from normal lumbar spinal MRI examination at surrounding CSF (white arrow), also of dorsal root ganglion in
1.5 T. (a) Mid-sagittal 4 mm T1-weighted spin-echo image show- foramen by surrounding fat (white arrowhead). Borders of dural
ing bright signal from epidural and subcutaneous fat, dark CSF sac (small black arrows) are less well-defined, however (d).
signal. (b) Mid-sagittal 4 mm T2-weighted fast spin-echo image Right and left oblique MR myelographic images presenting 3D
showing bright fluid signal from CSF and nucleus pulposus, also projections of dural sac acquired with single-shot, single-slice
from epidural and subcutaneous fat. Note better depiction of pos- technique (see below). Note good depiction of intradural nerve
terior disc contour compared to (a). (c) Axial 4 mm T2-weighted roots and root sleeves. Vertebral structures not imaged due to
fast spin-echo image at L4–5 produced with 3D DRIVE tech- heavy T2 weighting
nique. Note good depiction of intradural cauda equina fibres by
2.4 Magnetic Resonance Imaging (MRI) 19

a b

c d
20 2 Imaging Techniques for the Lumbar Spine

signal. In FSE T2-weighted images epidural and CSE image, may be almost invisible on T2-weighted
foraminal fat may be almost iso-intense to CSF (Fig FSE images because the normal fatty bone marrow is
2.9b, c). This has the advantage that extradural disc now iso-intense to the lesions. Application of fat-sup-
fragments in the intervertebral foramen are well-out- pression can be useful here (see below).
lined by bright fat; almost as well as in T1-W images. An FSE T2-W 3D driven equilibrium technique
One could say that FSE T2-W images provide “water (DRIVE) presently used in our department for axial
contrast” as well as “fat contrast”. The disadvantage of spinal imaging employs a desaturating pulse after
this is that the CSF-filled dural sac can be difficult to acquisition of the spin-echo, in order to null residual
distinguish from the surrounding epidural fat, as both magnetisation and so reduce the repetition time. In this
are now bright (Fig. 2.9c).This can create a problem way heavy T2 weighting can be produced in a rapid
when assessing, for instance, abnormal increase in epi- acquisition (Fig. 2.9c).
dural fat (lipomatosis) on T2-W FSE images (see An alternative option for producing “water images”
Chap. 4). In addition, bone marrow lesions with high is by the use of a T2*-weighted gradient-echo (GRE)
water content, such as in certain degenerative changes, sequence which also produces a high water signal.
metastases or osteomyelitis, which classically appear This technique is sometimes used for axial spinal
hyperintense to normal bone marrow in a T2-weighted imaging, most frequently in the cervical region.

a1 a2 a3

b1 b2 b3

Fig. 2.10 Comparative axial images acquired by T1 SE, T2 foramina. Outline of dural sac less well shown, however. T2*-
DRIVE, and T2*BFFE techniques respectively. (a1–3) Case weighted BFFE image (a3) produced with gradient-echo sequence
with conjoint left L5 and S1 root sleeves. T1-weighted image shows good depiction of dural sac, root sleeve and intradural
(a1) shows no intradural detail due to low CSF signal. Dural sac, nerve roots, but foraminal structures are less well shown. Spurious
root sleeves and foraminal details well-depicted due to high fat image of L3–4 disc protrusion (arrowheads) shown in T1-weighted
signal. T2-weighted DRIVE image (a2) produced by FSE sequence image (b1), not shown in T2-weighted DRIVE image (b2) and
giving high CSF signal as well as high fat signal shows good depic- T2*-weighted BFFE image (b3). Spinal nerve (arrow) well seen
tion of intradural nerve roots as well as dorsal root ganglia in outlined by fat in (b1) and (b2), not in (b3)
2.4 Magnetic Resonance Imaging (MRI) 21

Figure 2.10 shows a comparison of imaging features of cannot be used in T1-W post-gadolinium MR imaging
three techniques for axial lumbar spinal imaging: because the bright gadolinium signal is suppressed by
T1-W fast spin-echo, T2-W DRIVE and T2*-W bal- this technique together with the fat signal.
anced fast-field echo (BFFE). We have found the sec-
ond option the most useful.

2.4.5 MR Myelography:

2.4.3 Proton Density-Weighted The purpose of producing MR myelographic images is

(PD-W) Images not to satisfy nostalgic feelings in elder colleagues but
to provide a better diagnostic image of the intradural
MRI is a highly versatile method for assessing various nerve root. The course of a traversing nerve root as it
tissue characteristics and transforming these character- passes from the dural sac into the root sleeve in the
istics into image contrast. Beside producing images lateral recess region of the spinal canal is often hard to
weighted for differences in T1 or T2 relaxation times, follow in sagittal or axial MRI sectional images.
the MR acquisition sequence can be so arranged that Sagittal sections suffer from partial volume effects in
neither of these two tissue parameters plays a signifi- the lateral recess region, and even thin axial cuts can
cant role in image contrast; variations in signal inten- fail to identify the root, especially when the lateral
sity (brightness) producing image contrast now depend recess is not roomy. Individual cauda equina fibres can
mainly on variations in proton density within the tis- be discerned on thin (2 mm)-section T1-weighted vol-
sues. Ligamentous structures containing bound pro- ume scans and traced over some distance in oblique
tons (ligaments, cortical bone) are then clearly reformats (Hofman and Wilmink 1995) but compari-
discernible by their low signal intensity. Ruptures in son of the aspect of a single nerve root and root sleeve
ligamentous structures such as the outer annulus fibro- with the contralateral root or the adjacent root above or
sus are very clearly seen (Fig. 2.11) but this is the only below is not possible with flat sections through a
especially useful diagnostic feature of proton density curved tubular banana-shaped object such as the lum-
weighting and the technique is at present not routinely bosacral dural sac. Curved reformatted sections can be
used in spinal imaging. constructed but the production is time-consuming.
A presentation of a virtual 3D image of the dural
sac and root sleeve allows a better assessment of the
course of the root, and an easier comparison with adja-
2.4.4 Fat-Suppressed Images cent and contralateral roots. MR myelographic images
are generally acquired with heavy T2-weighting, which
Suppression of bright fat signal in the MR image can produces a very bright water signal from the CSF in
be achieved in several ways. Short TI inversion recov- the dural sac and (virtually) no signal from other spinal
ery (STIR) is very effective in nulling the fat signal structures. The dural sac is then easily segmented by a
from epidural fat and bone marrow, and is helpful in maximum intensity projection (MIP) technique similar
the analysis of bone marrow signal changes (Fig. 2.12). to that used in MR angiography, and presented as a
Spectral fat suppression by pre-saturation (SPIR, fat- virtual 3D object with the root sleeves well shown and
sat) can also be used in T2-W fast spin-echo sequences the intradural roots visible as dark linear structures
to produce the same effects. (Krudy 1992; el Gammal et al. 1995; Ferrer et al.
Post-gadolinium T1-weighted images can be acquired 2004). This technique compares well with conven-
with a spectral fat-saturation pre-pulse (SPIR or fatsat). tional contrast myelography; (Ramsbacher et al. 1997;
This is useful when bright fat signal (bone marrow, epi- Kuroki et al. 1998), and patient acceptance of an MRI
dural fat) is a hindrance to assessing contrast enhance- study is better than is the case with conventional myel-
ment of vascular structures (Fig. 2.13), but also infectious ography (Albeck and Danneskiold-Samsoe 1995).
or metastatic bone marrow enhancement, or enhance- Figure 2.14 shows an example of adjacent oblique
ment of post-operative epidural scar tissue can be better T2-weighted MRI sections fused to produce a virtual
identified in this way (see also Chaps. 4 and 5). STIR 3D representation of the dural sac and emerging root
22 2 Imaging Techniques for the Lumbar Spine

a b

c d

Fig. 2.11 Imaging of annular rupture by proton density com- foramen with similar weighting (c, d) show ruptured annulus,
pared to T1-weighting. Midsagittal T1-(a) and proton density- more clearly in proton density weighted image (d) (arrow). Note
weighted images (b) show extruded disc material behind intact small fragment of annulus displaced upwards into foramen (long
L4–5 posterior longitudinal ligament. Lateral sagittal cuts through arrow)
2.4 Magnetic Resonance Imaging (MRI) 23

a b

Fig. 2.12 Fat suppression in degenerative bone marrow changes. marrow above L4 endplate indicating Modic type 1 degenerative
T1-weighted image (a) shows area of signal loss due to increase changes; also suppression of fat signal from area below L5 end-
in bone marrow water content above L4 endplate (arrow); area plate indicating Modic type 2 fatty degenerative changes here.
of increased bone marrow fat signal below L5 endplate (thin Decrease in bone marrow fat signal combined with increase in
arrow). T2-weighted fast spin-echo image (b) shows areas with water signal is seen in Modic type I changes but also in for
increased fat as well as water content now hyperintense. STIR instance metastasis or spondylitis. Follow-up in this case
fat-suppressed image (c) confirms high water signal in bone revealed no progression over time
24 2 Imaging Techniques for the Lumbar Spine

where its effect on the nerve root is not clear (Hofman

and Wilmink 1996; see also Chaps. 4 and 5).
The MR myelographic dataset shown in Fig. 2.14
were produced with a multi-slice, multi-shot technique
requiring a lengthy acquisition 6 min. 30s. A single-
shot technique can be employed to reduce acquisition
time (Karantanas et al. 2000), and a refinement of this
sequence is used in our department. When the echo-
train length of an FSE sequence is increased to equal
the number of acquired profiles, a strongly T2-weighted
myelographic image can be produced with only a sin-
gle excitation. Such an image possesses a poor signal-
to-noise ratio (SNR), however (Fig. 2.15). When
multiple, successive, single-shot excitations are now
performed to improve the SNR, an MR myelographic
image is produced requiring a total acquisition time of
only about 30 s, with an image quality comparable to a
much lengthier multi-slice, multi-shot acquisition
(Fig. 2.16).
There are other technical options available to pro-
duce MR myelographic images, using gradient-echo
T2*weighted sequences (Zisch et al. 1992; Schnarkowski
et al. 1993; Eberhardt et al. 1997; Baskaran et al. 2003).
These will not be discussed in detail here.
It must be stressed that MR myelography is ancil-
lary to the standard MRI investigation, and can never
replace it (Thornton et al. 1999; O’Connell et al. 2003).
MR myelography has the same drawbacks as conven-
tional contrast myelography: false negatives occur
when the root is compressed distal to the root sleeve, in
the foramen or the sacral canal, and false positives are
seen when non-filling of a root sleeve is not due to
compression (see Chap. 3, Fig. 3.5). The standard MRI
cuts and the MR myelographic images should always
Fig. 2.13 Fat-suppressed post-gadolinium imaging. Mid-sagittal
T1-weighted spin-echo image with SPIR fat suppression by spec- be carefully matched against each other, and also
tral pre-saturation, post-gadolinium injection. Note signal loss against the clinical presentation. If a small L5-S1 her-
from fat in subcutaneous and epidural regions, bright enhancement niation for instance is seen to be extending into the
of basivertebral veins (arrow) and epidural veins (arrow head) epidural fat ventral to the dural sac but the root sleeve
at the same level is normally depicted and filled with
CSF on the MR myelogram (see Fig. 4.3), the clinical
sleeves. Such a presentation makes MR myelography a signs and symptoms of the patient should be reviewed
valuable adjunct in cases where disc or canal pathology with extra caution because a chance finding of an
is seen to be present on the standard MR images, but asymptomatic herniation is then quite likely.

Fig. 2.14 MR myelography. Two sections (a, b) selected from a image of virtual dural sac (c); acquisition time 6 min. 30s. Better
4-mm overcontiguous multi-shot, multi-slice oblique T2-weighted detail of intradural roots in individual sections (a) and (b), but better
FSE MR myelographic acquisition, fused by MIP to produce a 3D appreciation of entire dural sac and all root sleeves in (c) (arrows)
2.4 Magnetic Resonance Imaging (MRI) 25

a b

c
26 2 Imaging Techniques for the Lumbar Spine

a b

Fig. 2.15 MR myelography. Single-shot, single-slice MR myel- individual (b), acquisition time 6 min. 30 s. Note much poorer
ographic images in normal spinal canal (a), acquisition time signal-to-noise ratio in single-shot image
1.5 s, compared to multi-shot, multi-slice MIP image in same

Summary enhancement is usually better seen with spectral

fat suppression.
Imaging sequences and features best shown by
these in degenerative conditions. › T2-W: Dural sac and contents; central and
paracentral disc herniations impinging on the
› T1-W: Epidural and foraminal fat; lateral and dural sac; water content of the nucleus pulpo-
foraminal disc herniations in regions contain-
sus; fissures in the annulus fibrosus.
ing fat (foramen, lumbosacral transition and
sacral canal). Disc herniations adjacent to › NB: When FSE is used for T2-W imaging,
dural sac are not well seen; intradural nerve epidural and foraminal fat signal is sufficiently
roots are not well seen. bright to outline disc herniations in foramen,
lumbosacral transition and sacral canal.
› T1-W + Gd: Epidural veins; enhancing annu-
lar fissures; inflammatory epidural reaction › Proton density-W: Rupture of outer annulus
around an extruded disc fragment, inflamed fibrosus.
nerve root, post-operative epidural scarring or › T2++W MR myelography: Cauda equina, root
spondylodiscitis. Epidural scar or bone marrow sleeves, normal and compressed.
2.4 Magnetic Resonance Imaging (MRI) 27

a b

Fig. 2.16 MR myelography. Single-shot, single-slice multi- (b), acquisition time 6 min. 30s. Note comparable image quality
excitation MR myelographic image (a), acquisition time 32 s despite much more rapid acquisition in (a)
compared to multi-shot, multi-slice image in same individual

2.4.6 Imaging Planes rupture in the outer annulus fibrosus and extends cra-
nial and/or caudal to the vertebral endplates bordering
As a general rule, in sectional imaging, anatomic sur- the disc space (see Chap. 4 and Fig. 4.3). The distinc-
faces or structures are best imaged in a plane which tion between diffuse disc bulging and broad-based her-
lies perpendicular to the surface of interest, and least niation is often difficult to make in the sagittal plane,
well in a plane parallel to this surface. Thus, vertebral and the lateral recesses and the root sleeves are not well
endplates are best seen in sagittal and coronal sections, imaged. The foraminal borders on the other hand are
and not well in the axial plane because of partial vol- best defined in sagittal images, as is cranial migration
ume effects. The inner pedicular borders are best seen of extruded material in the foramen with compression
in axial and coronal sections, and not well in sagittal of the dorsal root ganglion against the pedicle (see
cuts. Fig. 4.5). The mid-sagittal diameter of the spinal canal
Sagittal: Images in this plane are best for demon- is better assessed in the sagittal plane than on axial
strating disc herniations and distinguishing between images, as are spinal deformities such as anterolisthe-
“contained” protrusions (whose maximum height does sis and retrolisthesis. Isthmic fractures in spondyloly-
not exceed the height of the parent disc) and extrusions, sis may be detected, as well as increase or decrease in
in which the displaced disc material passes through a sagittal diameter of the spinal canal which is associated
28 2 Imaging Techniques for the Lumbar Spine

with spondylolytic and degenerative anterolisthesis, 2.4.8 Considerations of Field Strength

respectively (see Chap. 4 and Fig. 4.20).
Axial: Images in this plane permit the best classifi- Increasing the field strength of the magnet used for MRI
cation of the axial location and extent of a disc abnor- produces an equivalent increase in signal-to-noise ratio
mality, diffuse, broad-based or focal (see Chap. 4 and and hence in low-contrast resolution in the MR image.
Fig. 4.2). Migration of extruded disc material cranial or A study comparing image quality at 0.5, 1 and 1.5 T
caudal to the level of the endplates can be hard to assess showed image quality at the two higher field strengths
due to partial volume effects. The lateral recesses of to be superior to that obtained at 0.5 T (Maubon et al.
the spinal canal are best studied in the axial plane, as 1999). If desired, the increase in signal can also be
well as lateral encroachment upon the spinal canal due traded off against other image properties: thinner slices
to hypertrophy of the facets and flaval ligaments, and or an increase in matrix size to improve spatial resolu-
the passage of the traversing nerve roots through these tion, or a larger field of view to expand anatomic cover-
regions may be traced. Isthmic fractures in spondyloly- age without sacrificing image quality. Alternatively, the
sis can often be seen in axial images. When measure- acquisition time can be reduced.
ments of the sagittal diameter of the spinal canal are There are other factors beside field strength affect-
performed in the axial plane, error due to tilting of the ing image quality in MRI, the most important being
plane of section relative to the longitudinal axis of the the characteristics of the RF antenna or coil employed.
spinal canal should be taken into account (see Fig. 2.5). In addition, imaging at higher field strengths such as
The foramen and its contents can be studied in axial 3 T produces increased chemical shift artefacts, sus-
images, but less well than in the sagittal plane. ceptibility and flow artefacts and also problems with
Oblique: Sections can be acquired or reconstructed energy deposition in body tissues. A drawback is the
in the plane of the emerging root sleeve, usually 20–30° loss of fluid-tissue contrast in T1-W FSE images due
off-coronal. Left and right oblique sections are some- to increased T1 relaxation times at higher field
times difficult to position with exact symmetry, and strengths. T1-weighted images produced by GRE and
this can make it difficult to compare the course of left fluid-attenuated inversion recovery (FLAIR) sequences
and right root sleeves and nerve roots, especially with suffer less from this problem (Shapiro 2006).
thin sections. Oblique 3D virtual images of the dural
sac acquired in T2-weighted MR myelographic projec-
tions do not suffer from this drawback.
Coronal: This imaging plane is used only rarely in 2.4.9 Abbreviated Scanning Protocols
diagnosis of degenerative disease. Some spinal deformi-
ties such as scoliosis or hemivertebra are imaged best in The suggestion has been made to reduce the number of
the coronal plane. acquisition sequences per spinal MRI study, in the
interest of increasing patient throughput. In a study
comparing a rapid two-sequence screening protocol
lasting 2 min. 30s and a detailed four-sequence proto-
2.4.7 Upright Imaging col requiring 28 min, all moderate and severe bulges
and herniations were detected by the rapid protocol but
The introduction of open MRI systems has made more subtle changes were better seen in the detailed
upright weight-bearing MRI studies possible, with the examination (Robertson et al. 1996). Another study
additional option of dynamic flexion-extension imag- (Chawalparit et al. 2006) showed disc herniations to be
ing of the spine (Weishaupt and Boxheimer 2003; demonstrated equally well by the two imaging proto-
Jinkins et al. 2005). As discussed in detail in Chapters cols, but sensitivity for nerve root compression was
3 and 4, the effect of such postural changes on normal significantly poorer in the screening protocol. In a
and pathologic spinal anatomy makes this a valuable study comparing a rapid MRI examination with spinal
addition to our diagnostic arsenal, most likely to be radiographs in a group of 380 patients with low back
useful in cases with spinal developmental stenosis or pain (Jarvik et al. 2003), clinical outcomes were the
another form of narrowing of the spinal canal. same for both groups but costs were greater in the MRI
References 29

group while more patients were operated (10 in the Jackson RP, Glah JJ (1987) Foraminal and extraforaminal lumbar
MRI group versus four in the radiography group). disc herniation: diagnosis and treatment. Spine 12: 577–585
Jarvik JG, Hollingworth W, Martin B et al (2003) Rapid mag-
Costs of rapid MRI were about half those of a conven- netic resonance imaging vs radiographs for patients with
tional MRI study (Gray et al. 2003). low back pain: a randomized controlled trial. JAMA 289:
2810–2818
Jinkins JR, Dworkin JS, Damadian RV (2005) Upright, weight-
bearing, dynamic-kinetic MRI of the spine: initial results.
References Eur Radiol 15:1815–1825
Karantanas AH, Zibis AH, Papanikolaou N (2000) Single-shot
turbo spin-echo MR myelography: comparison with 3D-turbo
Abdullah AF, Wolber PG, Warfield JR et al (1988) Surgical man- spin-echo MR myelography and T2-turbo spin-echo at 1 T.
agement of extreme lateral lumbar disc herniations: review of Comput Med Imaging Graph 24:37–42
138 cases. Neurosurgery 22:648–653 Krudy AG (1992) MR myelography using heavily T2-weighted
Albeck MJ, Danneskiold-Samsoe B (1995) Patient attitudes to fast spin-echo pulse sequences with fat presaturation. AJR
myelography, computed tomography and magnetic resonance Am J Roentgenol 159:1315–1320
imaging when examined for suspected lumbar disc herniation. Kuroki H, Tajima N, Hirakawa S et al (1998) Comparative study of
Acta Neurochir (Wien) 133:3–6 MR myelography and conventional myelography in the diag-
Baskaran V, Pereles FS, Russell EJ et al (2003) Myelographic nosis of lumbar spinal diseases. J Spinal Disord 11:487–492
MR imaging of the cervical spine with a 3D true fast imaging Luyendijk W, van Voorthuisen AE (1966) Contrast examination
with steady-state precession technique: initial experience. of the spinal epidural space. Acta Radiol Diagn (Stockh) 5:
Radiology 227:585–592 1051–1066
Bates D, Ruggieri P (1991) Imaging modalities for evaluation of Mansfield P, Maudsley AA (1977) Medical imaging by NMR.
the spine. Radiol Clin North Am 29:675–690 Br J Radiol 50:188–194
Carragee EJ, Alamin TF, Carragee JM (2006) Low-pressure Maubon AJ, Ferru JM, Berger V et al (1999) Effect of field
positive Discography in subjects asymptomatic of significant strength on MR images: comparison of the same subject at
low back pain illness. Spine 31:505–509 0.5, 1.0, and 1.5 T. Radiographics 19:1057–1067
Carragee EJ, Hannibal M (2004) Diagnostic evaluation of low O’Connell MJ, Ryan M, Powell T et al (2003) The value of rou-
back pain. Orthop Clin North Am 35:7–16 tine MR myelography at MRI of the lumbar spine. Acta
Chawalparit O, Churojana A, Chiewvit P et al (2006) The limited Radiol 44:665–672
protocol MRI in diagnosis of lumbar disc herniation. J Med Penning L, Wilmink JT (1981) Biomechanics of lumbosacral
Assoc Thai 89:182–189 dural sac. A study of flexion-extension myelography. Spine
Di Chiro G, Schellinger D (1976) Computed tomography of spi- 6:398–408
nal cord after lumbar intrathecal introduction of metrizamide Ramsbacher J, Schilling AM, Wolf KJ et al (1997) Magnetic
(computer-assisted myelography). Radiology 120:101–104 resonance myelography (MRM) as a spinal examination
Eberhardt KE, Hollenbach HP, Tomandl B et al (1997) Three- technique. Acta Neurochir (Wien) 139:1080–1084
dimensional MR myelography of the lumbar spine: compara- Robertson WD, Jarvik JG, Tsuruda JS et al (1996) The compari-
tive case study to X-ray myelography. Eur Radiol 7:737–742 son of a rapid screening MR protocol with a conventional
el Gammal T, Brooks BS, Freedy RM et al (1995) MR myelogra- MR protocol for lumbar spondylosis. AJR Am J Roentgenol
phy: imaging findings. AJR Am J Roentgenol 164:173–177 166:909–916
Ferrer P, Marti-Bonmati L, Molla E et al (2004) MR-myelography Ruggieri PM (1999) Pulse sequences in lumbar spine imaging.
as an adjunct to the MR examination of the degenerative Magn Reson Imaging Clin N Am 7:425–37, vii
spine. Magma 16:203–210 Schnarkowski P, Wallner B, Goldmann A et al (1993) [MR-
Fullenlove TM, Williams AJ (1957) Comparative roentgen find- myelography of the lumbar spine using a PSIF sequence:
ings in symptomatic and asymptomatic backs. Radiology 68: first experiences]. Aktuelle Radiol 3:53–56
572–574 Shapiro MD (2006) MR imaging of the spine at 3T. Magn Reson
Gray DT, Hollingworth W, Blackmore CC et al (2003) Conven- Imaging Clin N Am 14:97–108
tional radiography, rapid MR imaging, and conventional MR Skalpe IO (1978) Adhesive arachnoiditis following lumbar myel-
imaging for low back pain: activity-based costs and reim- ography. Spine 3:61–64
bursement. Radiology 227:669–680 Stadnik TW, Lee RR, Coen HL et al (1998) Annular tears and
Hofman PA, Wilmink JT (1995) 3-D volume scanning. A new disk herniation: prevalence and contrast enhancement on MR
technique for lumbar MR imaging. Acta Neurochir (Wien) images in the absence of low back pain or sciatica. Radiology
134:108–112 206:49–55
Hofman PA, Wilmink JT (1996) Optimising the image of the Staiger TO, Paauw DS, Deyo RA et al (1999) Imaging studies
intradural nerve root: the value of MR radiculography. for acute low back pain. When and when not to order them.
Neuroradiology 38:654–657 Postgrad Med 105:161–162, 165–166, 171–172
Hounsfield GN (1973) Computerized transverse axial scanning Thornton MJ, Lee MJ, Pender S et al (1999) Evaluation of the
(tomography). 1. Description of system. Br J Radiol 46: role of magnetic resonance myelography in lumbar spine
1016–1022 imaging. Eur Radiol 9:924–929
30 2 Imaging Techniques for the Lumbar Spine

Vertinsky AT, Krasnokutsky MV, Augustin M et al (2007) Wilmink JT, Lindeboom SF, Vencken LM et al (1984) Relationship
Cutting-edge imaging of the spine. Neuroimaging Clin N between contrast medium dose and adverse effects in lumbar
Am 17:117–136 myelography. Diagn Imaging Clin Med 53:208–214
Weishaupt D, Boxheimer L (2003) Magnetic resonance imaging Wilmink JT (1989) CT morphology of intrathecal lumbosacral
of the weight-bearing spine. Semin Musculoskelet Radiol nerve-root compression. AJNR Am J Neuroradiol 10:
7:277–286 233–248
Wilmink JT, Penning L, Beks JW (1978) Techniques in trans- Zisch RJ, Hollenbach HP, Artmann W (1992) Lumbar myelog-
femoral lumbar epidural phlebography. Neuroradiology 15: raphy with three-dimensional MR imaging. J Magn Reson
273–286 Imaging 2:731–734