The Spine: Myelography, CT or MRI?
World Small Animal Veterinary Association World Congress Proceedings, 2003
Robert M. Kirberger, BVSc, MMedVet (Rad), DECVDI
Department of Companion Animal Clinical Studies, Faculty of Veterinary Science, University of Pretoria
Onderstepoort, Republic of South Africa


Dogs or cats presented with suspect spinal pathology require a good clinical and neurological examination prior to embarking on any diagnostic imaging. It is essential to rule out orthopaedic or central nervous system problems and to localise a spinal lesion the C1-4, C5-T2, T3-L3 and L4-S3 region. Survey radiographs are then made of the affected region centering the primary beam at the site of suspect pathology. At least 2 orthogonal well positioned good quality radiographs are required.

Survey films may provide a specific diagnosis such as a fracture, neoplasia or discospondylitis and no further imaging may be required or additional information about the extent or severity of the disease process may be needed. The findings may also be equivocal such as in thoracolumbar disc disease where there may be several radiological changes involving more than one disc region. Alternatively the radiographs may be completely normal and the pathological process involves the specific soft tissues which are not visible on survey films.

Additional imaging is now required and the choice of myelography versus advanced tomographic imaging by computed x-ray tomography (CT) or magnetic resonance (MR) imaging has to be made. The choice will often depend on matters not related to the specifics of each imaging modality but rather the familiarity of the clinician with the procedure, cost implications and availability of equipment.


Myelography is often the first choice in that the veterinarian can perform the procedure himself. However it is an invasive procedure, exposes the patient and workers to radiation, involves general anaesthesia and may result in complications such as worsening of the clinical signs and seizures. Modern water-soluble non-ionic contrast media are relatively safe and with experience myelography is a very useful procedure. If executed properly it provides a lot of information including circumferential location of pathology, a reason often cited for performing CT or MR imaging. An additional advantage is that a cerebrospinal fluid sample can be obtained for analysis.

Cervical myelography is relatively easy to perform but penetration of the cord may be fatal. Additionally it will not allow contrast medium to bypass a severe compressive lesion making evaluation of the caudal extent of the lesion impossible and thus requiring an additional lumbar myelogram.

Lumbar myelography is technically a bit more challenging but is much safer for the patient. If the contrast medium is injected under pressure with radiographs made immediately, the effect of cord swelling may be overcome and the lesion as well as its cranial extent is more accurately localised. It is essential to make lateral, ventrodorsal (or dorsoventral in caudal cervical region) and oblique views to accurately determine the circumferential location of pathology.

Computed tomography

Advanced imaging with CT or MR imaging is non-invasive and gives multiplanar images of structures being evaluated resulting in cross sectional images with no superimposition of structures such as ribs or ilial wings. Craniocaudal and circumferential extent of pathology can be determined accurately and nerve roots can be followed caudally on contiguous transverse slices.

If helical CT is used vertebral column examination is completed in under 5 minutes. Examinations using standard 3rd or 4th generation scanners may take up to 30 minutes. Transverse images of the vertebral canal are produced. These can be reformatted to sagittal and dorsal images but with some loss of spatial and contrast resolution. Three dimensional reconstruction can also be performed and is particularly useful for surgical planning. Bony changes are seen very well with CT but soft tissue differentiation is not as good as in MR imaging. In particular this applies to the spinal cord although surrounding fat provides some contrast and also allows individual cauda equina nerve roots to be seen. Contrast may be improved by concurrent myelography but then invalidates the advantage of the non-invasiveness of CT with the same disadvantages cited for myelography.

Positioning is usually in dorsal recumbency with limbs strapped in flexion. Contiguous slices, up to 5 mm thick, are made of the affected region, with thinner slices in areas where better resolution is required, e.g., the intervertebral region. Different window levels and window widths are used to enhance differentiation of various tissues (e.g., bone and soft tissue windows) allowing much greater definition of soft tissue opacities as compared to radiology. Tissue attenuation in a region of interest is expressed in Hounsfield units (HU) which usually range from -1000 for air to +1000 for bone. The normal spinal cord has intermediate attenuation, similar to that of the kidney. Degenerated extruded disc material will be hyperattenuated and can thus be readily seen. Blood has a slightly higher attenuation than the spinal cord and haemorrhage in this region can thus also be distinguished.

Magnetic resonance imaging

Magnetic resonance imaging results in excellent soft tissue detail and reasonable detail of bone and cartilage. The various imaging planes are readily acquired and have equal resolution as compared to CT where the reconstructed images are not always optimal. There is no radiation danger and artifacts are minimal. However, the main limiting factors are its expense, prolonged anaesthesia when compared to helical CT, poorer resolution in cats and small dogs, and ferromagnetic substances in the patient that may cause severe artifacts. Only a few veterinary facilities have their own equipment but access to human equipment may be readily available in larger centres.

The most common spin-echo sequences used are T1 weighted (TW1) and T2- weighted (T2W). Numerous additional sequences, such as fat suppression and fast spin-echo techniques, are available on newer and more powerful machines. Altering the spin or gradient echo sequence changes the contrast of various tissues allowing characterisation of their makeup. This then allows better evaluation of extradural, intradural extramedullary and intradural intramedullary lesions which may only be visible as filling defects on myelography. At the same time subtle intracord changes can be evaluated which cannot be seen with myelography. Excellent anatomical detail is obtained with T1W images but T2W images give better tissue contrast allowing distinction of discs, ligaments, grey and white cord matter, cerebrospinal fluid and fat. Distinction of the latter can be enhanced by using fat suppression sequences. Additional information is obtained on T1W images by administering a Gadolinium based (Gd-DTPA) paramagnetic contrast medium at 0.2-0.4 mmol/Kg intravenously. Damaged capillaries will outline the vascular portion of a lesion distinguishing it from any surrounding oedema and may also show up pathology that was not visible on any other imaging sequence. A similar effect may be obtained in CT images after administering an iodinated contrast medium intravenously. The post-contrast signal intensity is described as absent, mild, moderate or marked, with homogenous or heterogenous enhancement.

The patient is usually placed in dorsal recumbency with the area of interest closest to a suitable coil. General anaesthesia is required and if inhalation anaesthesia is used the equipment must be non-ferrous. Procedures may take up to 30 minutes or more. Typically sagittal and parasagittal images are made first and the latter are particularly important to show up dorso- or ventrolateral cord compressions and intervertebral foraminal lesion which may not haven been seen myelography. The field of view is set to the region of interest and can be up to 40 cm long. Transverse images are then acquired. To save imaging and anaesthesia time these are often only done at the affected sites seen on sagittal images.

Transverse images are usually magnified to give more detail of the relationship between pathology and the surrounding tissues. Dorsal images of the whole region of interest may also be made. Slice thickness may vary from 2.5-3.5 mm and may be contiguous, overlapping or with a gap of up to 0.5 mm.

Signal voids on MR images occur because of low hydrogen proton density associated with gas (signal void on all sequences), cortical bone, calcification, fibrous materials and rapidly flowing blood. Soft tissues generally have a variable hypointense appearance. Cystic fluid is typically hypointense on T1W and hyperintense on T2W images but intensity may vary depending on the content. Cervical articular facet synovial cysts and hydromyelia can thus be characterised. The nucleus pulposus in normal dogs is hyperintense on T2W images due to its water content. Loss of this intense signal implies dehydration or herniation of the disc. Hyperintense extradural fat (T2W less intense than T1W) is readily seen and its displacement or obliteration is a good indication of a mass exerting cord pressure (also applies to CT). At the same time subtle changes in cord shape may also be appreciated. Hyperintense intraparenchymal cord lesions on T2W images may represent oedema, gliosis or even a dilated central canal. Tumour bone invasion and discospondylitis is easily appreciated on MRI and are seen much earlier than on radiographs. They show loss of hypointense cortical bone signal, disruption of more hyperintense bone-marrow signal or the presence of neoplastic tissue in bone.

Patient presentation

Besides the availability of equipment and clinician expertise the choice of which imaging modality to start off with also depends on the case presentation. The history, breed and clinical signs may play an important role. A Dachshund with a T2-L4 myelopathy is likely to have a disc prolapse and a well executed lumbar myelogram will provide all the information required to successfully treat the patient in 95% of cases. A dog with clinical signs of discospondylitis and with normal survey radiographs should proceed directly to MR imaging as this is more sensitive to pick up early changes. In cervical-vertebral-malformation-malarticulation syndrome ("Wobbler") the age at presentation and breed may be important in deciding what modality to use. The 7 year old Doberman is likely to have a lower cervical disc problem with or without bony changes. Myelography is adequate in the vast majority of these cases. The 1 year old Great Dane has a much greater likelihood of having multiple lesions involving the articular facets and pedicles. Although cervical myelography will provide most of the information required, MR imaging will usually give additional information, such as intracord pathology, which will influence the choice of surgical decompression sites. In these cases MR imaging may thus be the initial imaging modality of choice. In a Cavalier King Charles Spaniel presented with persistent scratching of one side of the shoulder or neck region, syringohydromyelia would be the most likely diagnosis and this is best seen with MR imaging as it is an intramedullary lesion. A myelogram in these cases may inadvertently result in contrast medium being deposited into the dilated central canal. In some cases, e.g., lumbosacral pathology, the choice may be more difficult. Lesions caudal to the lumbar intumescentia are difficult to diagnose with myelography due to the rapidly narrowing dural end-sac. Additional contrast techniques such as epidurography or vertebral venography rarely give precise information as to the nature of a possible lesion causing cauda equina syndrome. Myelography will however rule out additional pathology cranial to the L4 region which may influence the prognosis of the case. As lumbosacral pathology is often lateralised and the intervertebral foramina have to be evaluated, CT or MR imaging may be the initial imaging modality of choice. If there is extensive bony involvement on the survey radiographs, or if dysnamic studies need to be done to evaluate instability, CT may be more advantageous than MRI.


Irrespective of the imaging modality used, the imaging findings must still be correlated to clinical findings and neurological status. Many osseous or other changes may be incidental findings. Increasing amounts of abnormalities are seen with improved diagnostic modalities and their significance has to be correlated with the clinical presentation.


1.  Kirberger R M, Roos C J, Lube A M. The radiological diagnosis of thoracolumbar disc disease in the dachshund. Vet Radiol & Ultrasound 1992;33:255-261.

2.  Tidwell A S, Jones J C. Advanced imaging concepts. Clin Tech in SAP 1999;14:65-111.

3.  Jones J C, Wilson M E, Bartels J E. A review of high resolution computed tomography and a proposed technique for regional examination of the canine lumbosacral spine. Vet Radiol & Ultrasound 1994;35:339-346.

Speaker Information
(click the speaker's name to view other papers and abstracts submitted by this speaker)

Robert M. Kirberger, BVSc, MMedVet (Rad), DECVDI
Department of Companion Animal Clinical Studies
Faculty of Veterinary Science
University of Pretoria
Onderstepoort, Republic of South Africa

MAIN : Imaging : Imaging the Spine
Powered By VIN