Imaging for Peripheral Nerve Disorders

Published on 26/03/2015 by admin

Last modified 26/03/2015

Print this page

This article have been viewed 1841 times

CHAPTER 235 Imaging for Peripheral Nerve Disorders

The development of most subspecialties in neurosurgery has been closely associated with developments in imaging—radiography, pneumoencephalography, and angiography, all of which have played a large role in guiding spine and brain surgical planning for many decades. This was before the advent of cross-sectional imaging with computed tomography (CT) and magnetic resonance imaging (MRI) greatly improved the accuracy and simplicity of making an image diagnosis in the 1970s and 1980s.

Peripheral nerve surgery, however, has developed in almost complete isolation from imaging, both because radiography, angiography, CT, and standard MRI have not proved widely useful or reliable for the diagnosis of nerve impingement or injury, and because electrodiagnostic studies have provided an effective alternative in many situations. Nonetheless, peripheral nerve surgery is one subspecialty where clinical physical examination and exploratory surgery to define the pathology have continued to play a dominant role.

The discovery of a series of magnetic resonance (MR) pulse sequence strategies for tissue-specific imaging of nerves in 1991 and 1992 has opened a new diagnostic world in which a wide variety of pathologies involving nerves can now be visualized directly.¹^–⁵ These techniques are grouped under the term MR neurography to emphasize the fact that it represents tissue-specific imaging of peripheral nerves in which the image signal arises from the nerve tissue itself and specialized imaging strategies are required.

Before these developments in 1992, it had generally been assumed that peripheral nerves simply could not be imaged reliably.⁶ The strategy of using diffusion-based MRI sequences to help produce linear neural images was first discussed by Filler and colleagues in 1991⁷ and emerged as a workable technique through discoveries by Filler, Howe, and Richards (in London and Seattle) during late 1991 and early 1992,^2,^4,⁵ both for peripheral nerve and for CNS neural tractography (diffusion tensor imaging).

The resulting neurograms then served as a model for discovering additional nondiffusion tractographic methods for peripheral nerves. By 1993, there were several major publications in these fields.⁸^–¹⁰ In the subsequent 14 years, more than 150 academic publications have reported on various aspects of this new imaging modality for peripheral nerves, including several large-scale formal outcome assessment trials.¹¹^–¹³ Nonetheless, most textbooks of radiology or neuroradiology do not devote any pages at all to nerve imaging.¹⁴

In a number of settings, MR neurography has proved to be more efficacious than electrodiagnostic studies for identifying nerve compression that will improve with surgical treatment. This is true both in diagnoses that are typically evaluated by electrodiagnostic studies, such as carpal tunnel syndrome,^12,¹³ and in diagnoses in which such studies have proved difficult to rely on, such as thoracic outlet syndrome (TOS), piriformis syndrome, and related sciatic nerve entrapment and pudendal nerve entrapment syndromes.^11,¹⁵^–¹⁷

Utility for MR neurography has now been established for the evaluation of entrapment syndromes,^11,¹⁸^–²¹ nerve injury/evaluation of repair,²² and nerve tumor assessment,²³^–²⁵ as well as in the setting of neuritis and a variety of neuropathies.²⁶ It has also proved effective for evaluating nerve disorders affecting young pediatric patients, such as obstetric brachial plexus palsy.²⁷

Technical Aspects

The basis for tissue-specific MRI of nerve derives from the existence of a variety of unique types of tissue water in nerves. Clinical use of MR neurography is based on the application of new findings about nerve image characteristics. It is also based on a specific strategy of using imaging to evaluate nerves.

The two major types of neurographic techniques that have been described are diffusion neurography and T2-based neurography. Diffusion neurography was the first reported.^1,^2,⁵ This method has extremely high selectivity for nerves and should be very sensitive to a variety of types of pathology. However, the technical demands necessary to perform diffusion neurography have delayed its clinical application until quite recently.²⁸^–³⁰ In addition, diffusion images tend to degrade in detail at locations of severe pathology.

T2-based neurography, however, can be applied on a reliable basis with many existing top-quality clinical scanners, with minor modifications. Even when the technical equipment requirements can be met, however, there may be limitations on the provision of MR neurography because of a lack of expertise in prescribing, postprocessing, interpreting, and acting on the MR data. Once the technical limitations are resolved, however, T2 neurography has the advantage of improvement in quality of detail as the severity of the nerve abnormality increases—the opposite of diffusion neurography.

Diffusion-Based Tractographic Techniques

Neural tissue is among several tissues in the body that demonstrate the property of linearly correlated water diffusion. Instead of diffusing freely in any direction (isotropic diffusion), the water molecules in neural tissue are restricted in their diffusion such that they tend to move linearly along the main longitudinal axis of the neural tissue (anisotropic diffusion). The detailed biophysical basis of this restricted diffusion is still not fully understood because water molecules should move freely through cell membranes.

In MRI, a radiofrequency pulse is used to synchronize the spin of protons in water molecules, and a return signal from the tissue then results. The signal decays as synchronization is lost. Tissue contrast arises on T1- and T2-weighted images because the spinning protons in water molecules interact with surrounding tissues and with each other and these interactions cause small changes in the spin rate so that the spins desynchronize. Loss of synchrony makes the signal decay, and such loss takes place at various rates in different tissues.

In diffusion-based imaging, tissue contrast also arises because it is possible to use the relative degree of diffusional isotropy and anisotropy in a given tissue to affect the rate of signal decay in that tissue. MRI can be sensitized to anisotropy by applying magnetic field gradients during the pulse sequence of the image acquisition process. When diffusion carries a proton to a new position in the gradient, it experiences a slightly different local magnetic field strength, and thus its spin rate changes, which results in desynchronization of the spins. When water diffuses equally in all directions in a tissue, the rate of this cause of signal decay is not affected by how the gradient is oriented.

In anisotropic diffusion, however, there is a large impact on the decay rate if the gradient is directed strictly perpendicular or strictly parallel to the direction of the nerve. When the gradient is parallel, the nerve protons move rapidly to new positions in the gradient and their signal decays more rapidly than in an isotropic tissue. When the direction of the gradient is strictly perpendicular, the protons stay in the same field strength region of the gradient as they diffuse along the longitudinal axis of the nerve, and they show little or no signal decay relative to surrounding isotropic tissues.

In a peripheral nerve, anisotropic water makes up just a small fraction of the total water in the nerve, and therefore the effect is seen only when both fat and isotropic water signals are suppressed—one of the discoveries made by Filler, Howe, and colleagues in 1992.¹ When they found a means of selectively suppressing the signal of isotropic water, the result was a pure nerve image in which all other tissues disappeared and only the image of the nerve was left.

For both peripheral nervous and CNS tissue, Filler and colleagues ^4,⁹ also pointed out that it was possible to check each voxel (three-dimensional pixel) in an image volume by using three, six, or more gradient directions to determine both the magnitude and the direction in three-dimensional space of the anisotropy and then use this information to depict actual curving neural tracts rather than simply displaying two-dimensional differences in contrast in cross sections.

Until this time, the relative amount of anisotropic diffusion had been used to assign contrast to a given image pixel in a cross section or to assign colors to regions of neural tissue that respond to a given gradient direction. In tractography, however, each voxel is represented by an arrow that is a vector in three-dimensional space—representing a tensor.

When the arrows are followed from voxel to voxel, the result is a linear tract image that moves through the three-dimensional volume of the tissue along the course of the neural tissue. A variety of three-dimensional computer graphic techniques are now used to assemble tractographic images from these linear anisotropy traces—the basis of tractographic diffusion tensor and vector imaging.

T2-Based Neurography

Once the diffusion method was understood, it was possible to show that structures with long decay times (imaged at a relatively long echo time) in fat-suppressed spin echo images were, in fact, nerves. Previously, nerves had been misinterpreted as exhibiting short decay times on T2-weighted imaging ³¹ because they are a mixture of different tissues, including protein-laden axoplasmic water, myelin, fatty interfascicular epineurium, and connective tissue. Older methods allowed the image signal from these various component tissues to mix. In a variety of different imaging techniques, the result of mixing the image signals was a featureless gray image of the nerve, which left it difficult to distinguish clearly in an image and caused confusion about the fundamental imaging characteristics of nerves.

The Physiologic Basis of T2 Neurography

Several pieces of evidence suggest that the low-protein endoneurial fluid is what is seen most prominently in T2 neurography images. Endoneurial fluid is a low-protein liquid that lies within the privileged space of the endoneurium, confined by the perineurial blood-nerve barrier, and bathes the axons.^32,³³ It has a bulk proximal-to-distal flow along the nerve³⁴ that may be disrupted by nerve compression and edema.

Although endoneurial fluid is responsible for only a fraction of the protons that can be imaged in a nerve, it is one of the most distinctive types of tissue water in nerves from the point of view of MRI. By applying a chemical shift–selective (CHESS) pulse, it is possible to not only suppress fat around nerves but also suppress much of the fat signal from within nerves. A similar effect can be produced with inversion recovery–type sequences that achieve fat suppression. Then, by selection of an appropriate echo time (around 90 msec), a T2 weighting can be achieved that results in suppression of muscle signal, thereby leaving most of the signal from endoneurial fluid intact. Any one of several methods can also be used to suppress bright fluid signals from flowing blood. For all MR neurography studies, echo times should be greater than 40 msec (usually 70 to 100 msec) to ensure that no magic angle effects could occur.³⁵

When all three measures are taken—fat suppression, T2 weighting, and blood suppression—conditions are created to allow the generation of selective nerve images in virtually any location in the body. However, these pulse sequence manipulations by themselves are not sufficient to reliably obtain useful images.

Optimizing Performance of the Magnetic Resonance Imaging Main Magnet

One source of variation in quality among MRI scanners that has a very large impact on image quality in neurography is the homogeneity of the main magnetic field of the imager. MRI scanners are designed to provide a precise level of magnetic field strength over a useful volume. At a minimum, this volume will be large enough to cover the entire brain when the head is positioned at the exact center of the magnet. However an imager that can do only that will be relatively ineffective in producing a brachial plexus image, which requires a larger field of view and will be off center in the magnet. Good versatile nerve imaging therefore requires an MRI scanner with good magnetic field homogeneity off center in the magnet and over a relatively large volume.

Another aspect of this design factor concerns fat suppression via a CHESS pulse, which is used for the most selective T2 neurography. CHESS pulses work well only when the main magnetic field is very accurate and very uniform—this is accomplished by a process called shimming. All MRI scanners come with an ability to shim the magnet to improve field performance. In some imagers, shimming is done as a weekly or daily maintenance function, which is generally adequate for routine brain, spine, and general imaging work. Top-performing MRI scanners for nerve imaging are actually shimmed at the beginning of each imaging sequence on the volume of interest with the patient in position. Nerve imaging is one of very few applications when per-patient local shimming is important, and some expensive top-of-the-line academic MRI scanners do not offer the necessary shimming capability and thus produce very poor-quality neurography images.

Spatial Resolution and Signal-to-Noise Performance

Useful imaging of peripheral nerves requires a balance between the ability to survey large areas and high-resolution imaging with a small field of view. Nerve imaging may be most effective when the referring physician has a clear idea of the suspected location of the lesion. When the required imaging volume can be reduced to an area that is between 10 and 20 cm on each side, it is possible to image the volume with maximum spatial resolution and provide good fascicular detail. Selecting an even smaller volume to image does not currently result in any further gain in spatial resolution because of limitations of the imaging system. These considerations do make 200-µm spatial resolution possible on the very best quality MRI scanners.

Phased-Array Coils

Signal-to-noise performance can be greatly enhanced with the use of a specialized class of radiofrequency antennas as the receiver coil for collecting the image data called “phased-array coils.”³⁶ The basic idea behind phased array is to use more than one antenna to collect the weak signal. Two phased-array receiver coils can be placed at two different positions within the imager. Each will collect data that contain both signal and noise, but a considerable part of the noise derives from physical aspects of the scanner, so the noise spectrum looks slightly different in different locations of the scanner. The actual data signal from the tissue, however, is effectively the same anywhere in the scanner. By comparing the information from the two antennas it is possible to discard data when it differs between the two antennas because it is likely to be noise but to keep data when it is identical in the two antennas. This strategy results in greatly enhanced signal-to-noise performance.

Phased-array coil technology may allow four or even more receivers to be run simultaneously to achieve even greater signal-to-noise enhancement. The most important issues for the referring physician to be aware of are that phased-array capability is critical for high-resolution imaging and that performance varies among phased-array coils from different manufacturers. Consequently, a specialized image center with the optimal equipment for neurographic imaging may do a much better job than a center that does not specialize in this very demanding type of imaging.

Intravenous Gadolinium Contrast

Intravenous gadolinium contrast material is helpful only when tumor is part of the initial differential diagnosis. Nerve enhancement with contrast agents has proved to be unpredictably variable in non–tumor-imaging applications. Schwannomas are seen extremely well with MR neurography protocols without the use of contrast agents.

Image Plane Orientation

For most types of routine MRI, the original scan can be collected in three standard planes—axial, coronal, and sagittal—and then printed on film and read. However, MR neurography is considerably more demanding to acquire and is essentially wedded to an electronic reading format. The initial scanning must be done with attention to the main orientation of the nerves of greatest interest so that a set of nerve-perpendicular and a set of nerve-parallel images can be obtained. The nerve-perpendicular plane is identified by obtaining an initial image in the coronal plane and then orienting a specialized oblique image plane to be perpendicular to the principal longitudinal direction of the nerve or plexus of interest. A nerve-parallel image plane can also be prescribed during the image study, depending on the individual patient and the nerve to be imaged.

Areas of nerve irritation can be confirmed by comparing the appearance of serial nerve cross sections in the nerve-perpendicular views with those seen in the nerve-parallel views. In nerve-perpendicular images, the fascicle pattern can generally be observed, and this adds further confidence to a finding of focal nerve hyperintensity noticed in the nerve-parallel views. It will demonstrate expansion of the fascicle compartment at the expense of the interfascicular compartment at areas of focal hyperintensity (Fig. 235-1). The nerve-parallel images provide a linear overview.

FIGURE 235-1 Fascicular-level swelling as the basis for nerve hyperintensity on imaging. The distal sciatic nerve in a 36-year-old bus driver with a focal painful neuritis of unclear cause in the left lower extremity was examined. A, Cross sections of the sciatic nerve as it divides into tibial (t) and peroneal (p) components. Use of phased-array coil allows evaluation of the fascicle pattern. Some fascicles become bright and swollen at the site of injury (A2) at the expense of the interfascicular epineurium. One fascicle in the peroneal nerve (A3 and A4) is particularly enlarged as a result of wallerian degeneration. B, Maximum intensity projection reconstruction of the tibial and peroneal nerves created from a selected volume containing the nerve cross sections. In this case, the precise location of severe fascicular disruption could be shown with great accuracy, and it could also be demonstrated that there was no persisting lesion near the nerve that might require surgical treatment.

(From Filler AG, Maravilla KR, Tsuruda JS. MR neurography and muscle MR imaging for image diagnosis of disorders affecting the peripheral nerves and musculature. Neurol Clin. 2004;22:643.)

Nerve Image Reconstruction, Three-Dimensional Reconstruction, and Partial Volume Averaging

In general, effective interpretation of nerve-parallel images depends on the ability of the MR neurography imaging pulse sequence to make the nerve brighter than surrounding tissues. However, partial volume effects at the edges of nerves can lead to the artifactual appearance of variation in image intensity within a nerve image. Even when image plane orientation is attended to during image collection, the raw image may capture only pieces of nerves in individual image planes.

The full image of the nerve can then be reconstructed in several ways. The first is oblique or multiplanar reformatting. This step is very helpful in interpreting most MR neurography studies. By shifting the effective image plane a few degrees and changing the effective slice thickness it is typically possible to reassemble significant lengths of nerve or plexus. Without this step it is far more challenging for the reader of the image to comment on local variations in nerve image intensity or distortions in the course of the nerve. Some reformatting is mandatory for nearly all MR neurography studies.

In the brachial plexus, multiplanar reformatting is usually sufficient to generate a series of images that can reliably confirm the existence of a focal change in nerve image intensity. This is aided by positioning the patient in the scanner in a manner that tends to straighten the plexus. When a change in fascicle pattern shows increased intensity in the nerve-perpendicular views that matches a change seen in nerve-parallel views (Fig. 235-1), there can be a very high level of confidence about the clinical reality of nerve edema at the location that appears abnormal in the image.

Yet another category of three-dimensional postprocessing is the use of curved reformatting. This step is often capable of producing an image of an extended length of nerve or nerve plexus. The process of curved reformatting can generate a trace of the course of the nerve (Fig. 235-2). This nerve trace is also very useful for interpreting the image because it documents any unusual deviations in the course of a nerve or plexus. However, this method does add some subjectivity to the image analysis process and should not be read without comparison to more objective comparison images such as the original image planes or multiplanar reconstructions.

FIGURE 235-2 Complete bilateral brachial plexus with curved reformatting. A, The plexus image extends from axilla to axilla, although this image does not follow the plexus into the proximal part of the arm. The patient had bilateral arm and hand pain with different sets of symptoms on the right and left, negative cervical spine magnetic resonance imaging findings, and no wrist or elbow entrapment on electrodiagnostic studies. This image demonstrates an exaggerated sinusoidal course (*) on the right side where the plexus passes below the clavicle and over the first rib. The left side of the image shows a very different configuration at the same location (**), with some focal narrowing of the upper trunk. The proximal C6 spinal nerve (***) is flat and broad. B, The image in A is produced by an experienced technician or physician manipulating the positions of the “X” marks until the plexus elements come into view—a process called curved reformatting. The resulting trace line documents posterior deviation of the plexus (**), which is at the location of the scalene muscle.

(From Filler AG, Maravilla KR, Tsuruda JS. MR neurography and muscle MR imaging for image diagnosis of disorders affecting the peripheral nerves and musculature. Neurol Clin. 2004;22:643.)

Conspicuity and Maximum Intensity Projection Images

Maximum intensity projections (MIPs) essentially stack up the image slices and have the effect of reassembling the nerves (Fig. 235-3). These MIP images can help in appreciation of the overall course of the nerve and variations in course, caliber, and intensity because they may vary along the length of a nerve.

FIGURE 235-3 Assembly of raw images by maximum intensity projection (MIP). This process can greatly enhance the diagnostic value of images. A, Series of four oblique coronal image sections each showing portions of the exiting cervical spinal nerves and proximal brachial plexus. Isolated segments of nerve elements appear at various locations, and some are difficult to identify reliably. B, An overlay MIP made possible because the nerves are among the brightest elements in the image. It is easy to match the dorsal root ganglia (d) with the cervical spinal nerves because of the appearance of physical continuity. Nerve hyperintensity (h) associated with motor symptoms is readily appreciated in the right C6 cervical spinal nerve because the lower intensity of all the other elements is easily seen. Evidence of downward distortion of the right C7 and C8 cervical spinal nerves (s) by an enlarged scalene muscle is also readily seen because of the assembled view of the entire proximal plexus bilaterally.

(From Filler AG, Maravilla KR, Tsuruda JS. MR neurography and muscle MR imaging for image diagnosis of disorders affecting the peripheral nerves and musculature. Neurol Clin. 2004;22:643.)

The greatest amount of additional diagnostic information in these analyses occurs in brachial plexus cases.³⁷ Because of the significant incremental amount of clinical information provided by three-dimensional analysis and the susceptibility of most studies to this, these are essential aspects of the diagnostic interpretation process in this group.