Magnetic Resonance Imaging of the Upper Cervical Spine

Detection of Atrophic Changes in Rectus Capitis Posterior Minor Muscles

Thanks to Rick Hallgren for the use of these articles!

We are currently using MRI to investigate the functional integrity of the upper cervical spine. We started out looking for hypertonic muscles in a population of patients who were suffering from chronic head and neck pain. Our hypothesis was that head and neck pain might be caused by increased muscle activity resulting in muscle tightness. Since muscle tension can be detected by an osteopathic physician, and increased metabolic activity can be quantified using MRI data, we felt that we should be able to separate hypertonic muscles from normal muscles at rest. We proposed to collect image data from two groups of subjects: those free from both pain and significant motion restrictions, and those having significant motion restrictions and pain thought to be caused by abnormal, neuromuscular conditions. We proposed to collect pixel intensity data and then analyze these data to see if there were significant statistical differences between the two groups. My first task was to collect MRI data and to identify suboccipital muscles within the MR images. After looking back and forth between the screen of my computer terminal and an anatomy text for two days, it became obvious that I needed some help. So I brought together a physician and an anatomy professor to see if they could help me out. Their comments were classic. The anatomy professor said, "The reason you can't find those muscles is because they are not there." The physician immediately responded by saying, "No wonder these patients don't get any better." I had been using images that were collected from a chronic pain patient, and it was apparent that the rectus capitis posterior minor muscles were missing. When we looked at images from a control subject it was very easy to locate these muscles. At that point, the focus of our research switched from looking for hypertonic muscles to comparing muscle density between the control group and the chronic pain group.
Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Biological tissues have an abundance of positively charged hydrogen nuclei. We know from physics that any moving charged particle will generate a magnetic field. Normally biological tissues do not generate a net magnetic field because of the random orientation of the nuclei (See Figure 1). However, if biological tissues are placed into a magnetic field, such as exists in an MR imager, the randomly oriented nuclei, spinning in three-dimensional space, will align either parallel or anti-parallel to the direction of the external magnetic field (See Figure 2). Even though the nuclei are in a state of equilibrium with the external magnetic field, the magnetic moment of protons contained in the tissue is not in stable alignment with the external magnetic field. Since the nucleus is spinning, it responds to the external magnetic forces in the manner of a gyroscope, with its axis rotating around the direction of the external magnetic field in a process called precession (See Figure 3). The angular frequency of spin is known as the Larmor frequency and is directly related to the strength of the external magnetic field. Consequently, if the external magnetic field increases linearly as a function of distance, the angular frequency of nuclei would also increase linearly as a function of distance. This characteristic is extremely important, allowing the MR device to selectively examine points in three-dimensional space.

Suppose we take a quick look at a simple device which could be used to detect magnetic resonance of protons in a sample of extracellular fluid (See Figure 4). The sample is placed between the pole pieces of a magnet that is used to generate a static magnetic field. A radio frequency (RF) generator is used to supply a pulse of energy to move the nuclei out of a state of equilibrium. The frequency of the RF generator is set to be equal to the Larmor frequency of the nuclei, which is determined by the magnitude of the static magnetic field generated by the fixed magnet. The energy from the RF generator causes the nuclei to rotate away from the axis of the magnetic field. Figure 5 shows a spinning proton that has been given enough energy to rotate 90 away from the axis of the magnetic field. Once the RF generator is turned off, the receiver, tuned to the Lamor frequency, can be turned to detect the signal that is generated by the nuclei as they once again align with the external magnetic field. At the instance that the RF generator is turned off, the nuclei are processing in phase at a common frequency. However, this is a transient condition since the net magnetic moment of the sample is no longer in equilibrium with the external magnetic field.

There are three characteristics that are primarily responsible for the intensity of the signal that is received from a sample:

1)   The number of proton nuclei contained within the sample;

2)   The speed at which the magnetic moment returns to equilibrium with the external magnetic field (T1 or longitudunal relaxation);

3)   The speed at which the nuclei lose phase (T2 or transverse relaxation).

While the nuclei remain in phase and as the net magnetic moment begins to swing back toward equilibrium, the signal detected by the tuned receiver is relatively large. But as the nuclei get out of phase with one another and as the net magnetic moment approaches equilibrium, the magnitude of the signal exponentially approaches zero. The magnitude of the signal, along with the rate at which it decays, are determined by physical characteristics of the sample.

Biological tissues have intrinsic characteristics that permit reliable identification of tissue types from MR image data. In T1-weighted images, gray scale intensity is inversely related to the value of T1, meaning that tissues having short values for T1 (fat and blood) produce bright images, while tissues having long values for T1 (edema and tumors) produce dark images. In T2-weighted images, gray scale intensity is directly related to the value of T2, meaning that tissues having short values for T2 (tendons, cartilage, and muscle) produce dark images, while structures having long values for T2 (urine, edema, and fat) produce bright images. By manipulating system parameters, the image intensity of skeletal muscle (T1 = 900 msec, T2 = 30 msec) can be changed relative to fatty tissues (T1 = 250 msec, T2 = 80 msec). It is this ability to control image contrast that makes MRI so uniquely qualified to noninvasively study musculoskeletal trauma and dysfunction.

Figure 6

Figure 7

Figure 8

Figure 9

For all of our subjects, 4 mm thick contiguous slices of data were collected inferior and superior to the posterior arch of the atlas using a 1.5-Tesla superconducting magnet (SignaTM; General Electric Medical Systems, Milwaukee, WI) with TR = 2000 msec, TE = 25 msec, a 12 cm field of view, a 128 x 256 matrix, NEX = 2, and a total scan time approximately equal to 8.5 minutes. Figure 6 shows image data from one subject that was considered to be representative of the control group. The boundaries of muscles are well defined and the muscles are of uniform intensity. Figure 7 shows image data from one of the chronic pain subjects. There is a significant difference between the appearance of rectus capitis posterior major (RCPMA) muscles between the two groups. In general, muscles from the chronic pain group have an appearance that is characteristic of skeletal muscle that has died and been replaced with fatty tissue. The intensity of tissue infiltrating the RCPMI muscles in chronic pain patients in proton density-weighted (balanced) and in T2-weighted images (TR = 2000 msec., TE = 80 msec.) follows the intensity of structures that are known to be fat.\

Figure 8 shows a histogram of 200 samples taken from an image of the spinal cord of a control subject. These data are typical of spinal cord data from both control and chronic pain subjects, and show that the distribution of pixel intensity values is, for all practical purposes, Gaussian. The average intensity of a sample of pixel values taken from within the spinal cord of each subject, from eight consecutive slice levels, was used to calculate an individual normalization factor that permitted the comparison of muscle intensity values between subjects. Samples taken from the RCPMI muscles of control subjects had the same Gaussian distribution as the samples from the spinal cord. However, the histogram of pixel intensity samples taken from the RCPMI muscles of chronic pain subjects (See Figure 9) was found to be skewed towards higher intensity values as a result of an increased percentage of high intensity pixels (ie. fatty tissue).

The two most probable causes of this pathology are disuse atrophy and neurogenic atrophy. While suboccipital muscles can be functionally classified as extensors, their small size, relative to larger muscles that act in parallel with them, minimizes their contribution to motion. Consequently, if the atrophy were due to disuse then we would expect to see atrophy of the larger, parrallel muscles that are mainly responsible for motion. We have not seen atrophy in any of the parallel muscles. Ruling out myopathic disease, local trauma, and disuse atrophy, progressive wasting of skeletal muscle, accompanied by increased excitability of its fibers, is an indication that there is motor axonal loss consistant with Wallerian degeneration. Disruption of the neural pathway can result from mechanical trauma to the nerve such as occurs from a tractional injury, or as a result of compression of the nerve at the point that it penetrates a muscle.

The C1 dorsal ramus, arising from the C1 spinal nerve as it crosses the superior aspect of the posterior arch of the atlas, branches to innervate the suboccipital muscles. The C1 dorsal ramus passes dorsolaterally through the suboccipital plexus of veins, describing an upward arch to enter the suboccipital triangle where it divides to form a branch for the RCPMI muscles. Most commonly the suboccipital muscles are innervated on their dorsal surfaces, however the branch to the RCPMI muscles may innervate its muscle ventrally or by penetrating the RCPMA muscle. Anatomic variations such as this, both intra- and interindividually, are commonly reported. At this potential site of entrapment (See Figure 10), the nerve root to the RCPMI muscles is vulnerable to traction during whiplash-type neck distortions. The presence of abnormal spontaneous electrical activity (fibrillation potentials and positive sharp waves), when a muscle is relaxed and an EMG needle electrode is not being moved, is an indication of abnormal nerve and/or muscle membrane stability that most commonly accompanies lower motor neuron trauma that results in denervation of skeletal muscle. As a rule, no abnormal spontaneous activity develops in skeletal muscle weakened by disuse atrophy. Consequently, abnormal spontaneous activity, if reproducible in at least two muscle sites, indicates the existence of an abnormality that is associated with denervated muscle and certain primary muscle diseases. We have collected EMG data from three chronic pain subjects.

Figure 10.   Innervation of suboccipital muscles by C1 dorsal ramus. Note the course of the C1 nerve going through the RCPMA muscle prior to innervation of the RCPMI muscle. The white arrow shows the region of potential entrapment.

Positive sharp waves were seen in 4 areas of the RCPMI muscle on the right side of one of the chronic pain subjects. There were between 15-25 acceptable insertions observed on this person suggesting that approximately 20% of all insertions would show abnormal spontaneous activity. No fibrillation potentials were seen (See Figure 11). On the left side of the same patient, no spontaneous activity was seen. However, rapid repeated recruitment of the rectus capitis posterior minor muscle with a firing rate between 25-50 hertz in one motor unit was observed and documented (See Figure 12). The amplitude of the motor units were a maximum of 1-1.5 k and did not appear to be a long duration (approximately 10 msec). The second chronic pain subject exhibited markedly reduced insertional activity. The third chronic pain subject was not able to tolerate the procedure.

Figure 11.   Positive sharp waves recorded from the right RCPMI muscle. Gain: 50 microvolts/division. Sweep: 10 milliseconds/division.

Figure 12.   Recruitment from the left RCPMI muscle. The arrows point out one motor unit firing at between 25-50 Hertz. Gain: 100 microvolts/division. Sweep: 10 milliseconds/division.

Data were also collected from a 49 year old male control subject with no history of whiplash or neck pain. The MRI of his rectus capitis posterior minor muscles was normal. This subject did not show any evidence of positive sharp waves, fibrillation potentials, spontaneous activity or abnormal recruitment of the RCPMI muscles. At least 20 areas of the right and left RCPMI muscles were explored for abnormal insertional activity and spontaneous activity. Recruitment did not show any evidence for rapid firing motor units. We were able to isolate motor units that were approximately 1,000 mv in size.

Detection of a Cervical Myodural Bridge

We have recently described a previously unreported connection between the rectus capitis posterior minor (RCPMI) muscle and the spinal dura (See Figure 13). Nowhere in Gray's Anatomy is any functional relationship described between the RCPMI muscle and the dura mater. The existence of a cervical myodural bridge is of special interest to the osteopathic profession because it provides a direct physical link between the musculoskeletal system and the dura mater.

Figure 13.   Line drawing of a hemisected head showing the region displayed in Figure 14.

A midline sagittal section was performed on a head and neck specimen obtained from a fresh unembalmed human adult male cadaver. The RCPMI muscle was immediately visible arising from the posterior arch of the atlas and ascending to be inserted into the surface of the occipital bone from the inferior nuchal line to the foramen magnum (See Figure 14). The influence of the RCPMI muscle on the dura mater was artificially produced in the hemisected specimen. Artificially functioning the muscle produced obvious movement of the spinal dura between the occiput and the atlas, and resultant fluid movement was observed to the level of the pons and cerebellum (See Figures 15 and 16).

Figure 14.   Photograph of fresh hemisected cadaveric specimen showing:
1)   posterior border of foramen magnum;
2)   posterior arch of C1;
3)   posterior atlanto-occipital (PAO) membrane-spinal dura complex;
4)   connective tissue attaching the RCPMI muscle to the PAO membrane-spinal dura complex;
5)   rectus capitis posterior minor (RCPMI) muscle.

Figure 15.   Photograph of fresh hemisected cadaveric specimen showing the spinal dura at rest with no tension on the RCPMI muscle.   1) posterior border of foramen magnum;   2) posterior arch of C1.

Figure 16.   Photograph of fresh hemisected cadaveric specimen showing the effect upon the spinal dura when tension is applied to the RCPMI muscle.   1) posterior border of foramen magnum;   2) posterior arch of C1.

It has been suggested that the function of the RCPMI muscle is to provide static and dynamic proprioceptive feedback to the CNS, monitoring movement of the head and influencing movement of the surrounding musculature. We suggest that the RCPMI muscle may also act to monitor and control movement (such as folding) and tension of the spinal dura. The dura may tend to fold or pleat during extension of the cervical spine with the greatest amount of folding theoretically occurring in the area capable of the greatest degree of extension, which would be the atlanto-occipital joint. Consequently, the RCPMI may act to remove the dural folds thus protecting cerebrospinal fluid hydrodynamics (flow) during head extension, and this mechanism may be compromised when atrophy of the RCPMI muscle occurs. Hypertrophy of this connection could result in excessive tension being placed upon the dura, resulting in pain. Pathology in a muscle having direct influence on a pain sensitive structure suggests an alternative mechanism for generation of cervical headache.

In a recent article in the JAOA, researchers describe the effects of placing a physicians hands on the suboccipital region of the cervical spine and performing a simple circular kneading to the region. This soft tissue manipulative technique is similar to the occipitoatlantal technique of Sutherland. The study found that simply placing the physicians hands under the head caused vasodilation to occur in the subject's finger. A larger increase in pulse amplitude was observed when manipulation was applied. Variations in digital pulse amplitude can be used as a relatively direct and immediate index of vasomotor tone of the dermal arterioles. The authors suggest that the this sympathetic response may occur as a result of a perturbation of the cerebrospinal fluid resulting from mechanical pressure.

Future Work

In order to more accurately quantify the extent of atrophy, we will be collecting additional data using a contiguous slice, multi-echo, multi-planar spin-echo protocol (TR=2500, TE=15/80, NEX=1.0, 192x256, 16mm FOV). We plan on reformatting the image set to obtain a new set of images that will be approximately perpendicular to the long axis of a specified muscle. This will enable us to collect pixel intensity data along the long axis of the muscle from a region surrounding the very center of the muscle, thus reducing the possibility of an operator specified region of interest that might produce significant error due to using images that have "cut" the muscle at an oblique angle. Figure 17 shows a sequential series of MR images from a cervical spine data set of a control subject, reformatted for analysis of the right, obliquus capitis inferior muscle.

We also plan on collecting images to see if the spinal dura folds doing extension of the cervical spine in individuals who have atrophy of the RCPMI muscles.

Figure 17.   Sequential MR images from the cervical spine of a control subject reformatted for analysis of the right, obliquus capitis inferior muscle.


Chronic pain syndromes are costly to both the individual and to society. Even apart from the personal implications of spending 20-40 years in disabling pain, the financial burden of chronic pain is estimated to cost our country over 60 billion dollars annually, with most of this expense attributed to a small group of patients who are considered to be disabled. Chronic pain syndromes are often difficult to treat because a clear cause for the pain is often absent. For this reason, detection of fatty infiltration in suboccipital muscles of individuals being treated for chronic head and neck pain is a significant finding. Equally important is the discovery of a connection between the RCPMI muscle and the spinal dura. The ability to identify atrophic and hypertrophic changes in muscles of the upper cervical spine from MRI data at an early point in the disease process could enhance a physician's ability to initiate appropriate treatment that would prevent the condition from progressing from the acute to the chronic stage.

Based upon preliminary data, we conclude that some individuals suffering from chronic head and neck pain have significant infiltration of a fatty type of tissue into suboccipital muscles, accompanied by EMG abnormalities compatible with denervated muscles. The presence of positive sharp waves in EMG recordings strongly suggests that there are dennervated muscle cells in the RCPMI muscle. High frequency recruitment in the contralateral muscle also suggests decreased innervation of the RCPMI muscle. These findings give support to the hypothesis that neck trauma can cause peripheral mononeuropathy resulting in dennervation atrophy in skeletal muscle and associated clinical symptoms of chronic pain. While the anatomic and physiologic basis for the chronic pain is far from certain, we propose the following possibilities:

  • During flexion/extension/rotation of the atlanto-axial-occipital motion segment as a consequence of whiplash-type neck distortions, there is stretching and/or contraction of the rectus capitis posterior major (RCPMA) muscles that causes traumatic constriction and/or entrapment of the Cl dorsal ramus

  • Damage to the Cl dorsal ramus may result in deafferentation pain. It has been reported that low back operations sometimes cause lesions to back muscle innervation, with corresponding dennervation atrophy, and that failure of paravertebral musculature has the potential to cause spinal instability that may result in low back pain.

  • Atrophy of the RCPMI muscles may result in decreased proprioceptive activity, resulting in dizziness and/or balance problems. Functional weakness of RCPMI muscles may also result in increased infolding of the spinal dura during movement of the head and neck which may result in pain.

  • Entrapment of the Cl dorsal ramus may result in ectopic pain generators that are hyper-sensitive to motion and touch. Abnormal impulses generated in peripheral nerve at the level of Cl are conducted to the trigeminocervical nucleus where they may be interpreted by the brain as painful stimuli originating from muscles, joints, and ligaments of the head and neck.

Return to HEADACHE

Return to ChiroZINE ARTICLES