There are several procedures that fall under the rubric of “electrodiagnosis”. These include electromyography, nerve conduction studies (including late potentials) and evoked potentials.


Electromyography (EMG) refers to the electrical detection of signals arising from the depolarization of skeletal muscle. These signals may be detected from skin surface electrodes or from needles placed directly within the muscle. These two types of recordings are used for different purposes, with needle recording used to detect the behavior of individual muscle fibers and motor units while surface recordings are used to detect overall muscle activity in particular positions or actions. Surface EMG is not a common clinical procedure, though it may be used in rehabilitation. Needle electromyography is used to determine whether there is damage to nerve fibers to individual muscles.

Needle Electromyography

Needle electromyography (EMG) is designed to investigate the amplitude and morphology of the electrical signal within skeletal muscle. There are specific findings that appear as a result of diseases of muscle and due to denervation of muscles. Some of these findings are seen spontaneously when simply recording from a needle placed in the muscle. Some findings appear when the needle is moved in the muscle (insertional activity). Some appear during voluntary contraction of the muscle.

A normal muscle is electrically silent when recording from a needle electrode. Movement of the needle normally elicits a brief burst of depolarization from muscle fibers (called insertional activity). This burst of activity ends immediately upon termination of the movement, with restoration of electrical silence. The only place within the muscle that is not electrically silent is the motor end-plate. There are two types of electrical activity that can be seen in the motor end-plate at rest, miniature end-plate potentials and end-plate spikes. These can be distinguished from abnormal resting discharges, but this distinction requires some care. This presents a particular challenge for beginning electromyographers, since end plate spikes can be misinterpreted as evidence of denervation or of increased insertional activity and membrane instability. This is the reason that it is recommended that electromyographers take care to avoid the end plate region. The usual electromyographic test will examine at least ten locations within any single muscle before making a determination as to the normality of insertional activity or presence of abnormal activity at rest.

After the resting electromyographic activity and insertional activity is assessed, the patient is asked to voluntarily contract the muscle. Contraction takes place by activating motor neurons to the muscle, each of which is connected to many muscle fibers scattered throughout the muscle (termed a motor unit). The electrical signal that is recorded as a “motor unit potential” (MUP) arises from the integration of the electrical signals arising from the discharge of the several muscle fibers within recording distance of the tip of the needle (typically 1-3mm) that are attached to the same motor neuron (Figure 1A). The amplitude of the MUP is dependent on the density of the muscle fibers attached to that one motor neuron (also to the proximity of the MUP). This is remarkably uniform for most clinically tested muscles, the amplitude roughly being between 200 and 2000 microvolts. Additionally, MUPs usually have only one or two upward peaks. As the strength of contraction is slowly increased, motor units are recruited in a very orderly sequence. Each active motor unit increases its firing frequency to a defined level (usually around 10 cycles/second or hertz, at which point an additional motor unit is recruited. This process is quite orderly and can be quantified (recruitment pattern). Delayed recruitment (i.e., excessive firing rate of individual units prior to the recruitment of an additional unit) is a reflection of loss of motor units within the muscle. The final step in the EMG assessment of a muscle (a step which may not be necessary if everything else has been normal) involves maximal contraction of the muscle. During such contraction, the electrical activity should fully obscure the baseline (termed a full interference pattern). Incomplete interference pattern is considered to be a reflection of loss of motor units in a muscle, though it can also be seen with diminished effort.

Needle EMG evaluates the integrity of the motor unit, i.e., the motor neuron, motor axon and the muscle to which it is attached. Muscle diseases can produce some membrane instability if the disease is very active. This can result in the appearance of "fibrillation potentials" that represent the contraction of individual muscle fibers. As a rule, these contractions are much too small to be seen clinically. However, fibrillations are seen more commonly in diseases of the nerves. Muscle disease changes the motor unit, as and in some case can be associated with fibrillations due to damage of the distal motor axon. Due the fact that muscle fibers are “sick” in myopathy, the MUPs tend to be of low amplitude short duration. During even minimal contraction a greater number of these sick muscle fibers are needed to maintain the force of contraction, so “early recruitment” of motor units is seen (more motor units firing at higher rate than expected for the force).

Damage to motor axons (either at the level of the anterior horn cell, the motor root or the peripheral nerve) results in a series of quantifiable changes in the EMG. It is noteworthy that these changes are triggered by actual disruption of the motor axon and develop in an orderly sequence that can help determine the timing of the injury. These changes are not seen with damage to the myelin of the motor axon (assuming that the axon, itself, is undamaged). This is interesting because damage to myelin can result in complete block of motor conduction and even produce complete paralysis of the muscle without any of the changes that are associated with denervation. Additionally, damage to the central nervous system above the level of the motor neuron (such as by cervical spinal cord trauma or stroke) can result in complete paralysis without any abnormality on needle EMG except incomplete (or absent) interference pattern.

A series of events take place in the individual, denervated muscle fibers that can be detected as abnormal electrical signals. First of all, over the period of a week or two, the denervated muscle fiber becomes progressively more mechanically irritable. Therefore, electrical discharges provoked by movement of the needle can outlast the actual movement by more than a second. This is termed “increased insertional activity.” Although this finding is not particularly specific, it does indicate that the muscle is excessively irritable. Muscle fibers also become chemically sensitive to their microenvironment and their membranes can also become unstable enough to produce spontaneously activity. This is recorded as depolarization of individual muscle fibers. The spontaneous depolarizations of the individual fibers appear as fibrillation potentials (Figure 1B) and positive sharp waves (Figure 1C). These do not occur in normal muscles since the normal muscle fibers are only responsive to the activation of their motor unit by neuromuscular transmission. Typically, it takes more than a week for such potentials to develop and they will disappear with complete degeneration of the denervated muscle fiber.. Needle EMG is very sensitive for the detection of these signals and they most often reflect denervation, although they may also occur in severe muscle disease or injury. The finding of fibrillations and positive sharp waves is the most reliable and objective test that there is for damage to motor axons to the muscle after one week at least up to 12 months after the damage. If there is ongoing damage such as in Amyotrophic Lateral Sclerosis one can see ongoing denervation. Unfortunately, the finding of fibrillations and positive sharp waves is often termed “acute denervation”, although “acute” in this case refers to weeks and months.

Reinnervation of muscle is an ongoing process, occurring whenever a muscle is partially denervated. This process typically involves the development of sprouts from adjacent, unaffected motor nerve fibers that ultimately contact at least some of the denervated muscle fibers. These reinnervated muscle fibers cluster right in the area of other, normally innervated muscle fibers. This process results in the development of clumps of reinnervated muscle fibers attached to individual motor neurons (remember, the normal motor unit innervates muscle fibers scattered throughout the muscle). Typically these motor units become significantly larger both in amplitude and duration, since the needle is likely to be recording from more muscle fibers in this clump. Also, the MUPs often become more irregular (termed “polyphasic”) (Figure 2). This process takes months to develop and indicates the presence of chronic denervation. It should be noted that the needle study is much less sensitive to the process of reinnervation than it is to the findings of fibrillations and positive sharp waves that are seen with recent denervation.

The typical needle EMG examination requires sampling several muscles. Its ability to localize a lesion depends on sampling muscles innervated by the same nerve but different nerve roots, muscles innervated by the same nerve root but different nerves and muscles innervated at different locations along the course of the nerves. Paraspinal muscles can be very useful in this regard because nerve root damage will tend to produce abnormalities in these muscles as well as within the muscles of the limbs (helping to distinguish a radiculopathy from a plexopathy or peripheral neuropathy, for example). Sometimes precise localization can be difficult due to the overlap in innervation of the various nerve root levels. Usually MUPs and recruitment patterns are not assessed in the paraspinal muscles.

Nerve Conduction Velocity (NCV) Studies

Nerve conduction studies can test sensory or motor nerve fibers and can determine both the speed of conduction as well as the amplitude of the electrical signal evoked following stimulation of a nerve. They can detect areas of focal nerve damage.

Motor Conduction Studies

Motor conduction studies are performed by stimulating a motor nerve while recording the response from its target muscles (Figure 3). It is important to note that the electrical signal that is being recorded following motor nerve stimulation (called the compound muscle action potential - CMAP) is actually generated by the muscle, and therefore it is quite large. When motor nerve fibers are stimulated close to the muscle, the amount of time before the muscle starts depolarizing is called the “terminal latency”. The term “latency” in electrodiagnosis is used to define the time between a stimulus and the appearance of a response. In the case of “terminal latency,” this value includes both the amount of time that it takes the nerve to conduct from the point of stimulation to the motor end plate area and the amount of time for the neuromuscular junction transmission to activate the muscle. Strictly speaking, the terminal latency does not directly measure nerve conduction (because it includes the neuromuscular junction activation phase also) but it is a reasonable reflection of nerve conduction over this segment of the nerve in the absence of uncommon neuromuscular diseases. There are tables of normal for the terminal latencies of defined lengths for each of the major motor nerves of the limb. Abnormal prolongation of this value is often of benefit in the detection of distal entrapment neuropathies. Once a terminal latency has been recorded, the motor conduction velocity can be determined by stimulation of another, more proximal site along the motor nerve. The computation of motor nerve conduction velocity requires knowing the distance between the two simulation sites and the difference in the terminal latencies recorded from the more distal and more proximal sites. Dividing the distance by the time gives the nerve conduction velocity over the segment in between the stimuli.

Sensory Conduction Studies

Sensory conduction velocity is an easier measure to compute, but is more technically difficult to record (Figure 4). This test can be done in either an orthdromic (i.e. distal stimulation and proximal recording) or antidromic (i.e. proximal stimulation and distal recording) direction. Sensory nerves that can be recorded are: radial, median, ulnar, sural nerve and superficial peroneal nerve. The recording is made directly from the sensory nerve (the evoked response is called the sensory nerve action potential - SNAP) and therefore is quite small (about a thousand times smaller than the CMAP). The distance between the site of stimulation and recording is divided by the latency (i.e., the amount of time from the electrical stimulus to the SNAP) to determine the sensory nerve conduction velocity over the segment. Of course, the SNAP is quite small in amplitude, and recordings must be done in a rather meticulous fashion to avoid artifact. If the extremity is too cold the SNAP may not be recordable.

NCV Limitations

The sites from which nerves can be directly stimulated and from which the nerve or appropriate muscles can be recorded limit sensory and motor nerve conduction studies. For example, this makes the technique poorly suited to the investigation of nerve root problems since it is difficult to directly stimulate nerve roots in patients and similarly challenging to record from individual nerve roots. This is due to several factors, but mostly due to the deep location of the roots and the multiple surrounding structures. Other common technical problems in nerve conduction studies include difficulties locating the nerves and in measuring the course of a nerve (particularly for those nerves that follow a winding or bending course).

Measuring NCV results

The results of nerve conduction studies are compared to tables of normal and also to the values in an unaffected limb of the same individual. There are normal values for both sensory and motor conduction (as well as for terminal latency). For example, a good rule of thumb is that motor nerve conduction should be at least 40 meters/second in the lower limb, while sensory conduction should be at least 40 meters/second. Normal aging can slow the conduction velocity as can low temperature of a limb. In the very elderly it may be very difficult to record some the sural SNAP. There are tables that can be used to adjust normal values with extremes of age. For the F response there are tables factoring in height (see below).

The two values that are most important in a nerve conduction study are the speed of conduction and the amplitude of response. The speed is a reflection of the diameter of the axons and, most importantly, the thickness of the myelin sheath. Most of the conditions that damage nerves result in at least some injury to the myelin covering the axons. During recovery from focal neuropathy a thinner and less well-developed myelin sheath is produced, slowing conduction. Of course, this slowing would be greatest in the area of the damage. Additionally, other conditions such as Charcot Marie Tooth Disease or Guillian Barre Syndrome preferentially damage the myelin of the largest, fastest conducting fibers. This causes slowing as manifest by decreased conduction velocity. Actual blockage of conduction can occur due to damage to the myelin of 3-4 internode segments. When remyelination occurs conduction velocity is still decreased to the shorter internode distance (see diagram).

Axonal neuropathies can occur in toxic neuropathies. In these situations the amplitude of the CMAP and SNAP are much more affected than velocity. Diabetic distal symmetrical neuropathy, the most common neuropathy, has features of both demyelination and axonal damage.

Evoked Potentials

Evoked potentials are responses in the nervous system to stimulation of a sensory pathway. Clinically, this includes stimulation of a sensory nerve in the limb (somatosensory evoked potentials - SSEPs), the visual system (visual evoked potentials) or the auditory system (brain stem auditory evoked potentials). These techniques have the potential for evaluating the integrity of the pathways of sensory transmission all the way from the point of peripheral activation through the cerebral cortex.

The procedure for recording evoked potentials requires placement of low-impedance surface electrodes over several portions of the nervous system, followed by repeated activation of the sensory pathway. The minute electrical responses that are evoked by stimulation (for SSEPs, this usually consists of electrical stimulation of sensory or mixed nerves in the limbs) are recorded and averaged over many trials. This averaging eliminates background “noise” and the normal ongoing electrical nervous system activity that is often much larger than the signal evoked by the stimulus. In the case of SSEPs, usually over at least 256 stimuli are needed in order to obtain reliable, reproducible responses. Damage to the sensory pathway decreases the speed of conduction (much as was described in the section on nerve conduction studies), although diminished amplitude (which normally has a higher degree of inherent variability) may also be seen.

Somatosensory evoked potentials

SSEPs are produced by activation of the large diameter peripheral nerve sensory fibers (Figure 5). These nerve fibers include many that are conveying sensation from muscles as well as those from touch and pressure receptors in the skin and deeper tissues. Pain fibers contribute little (if anything) to the normal, clinical evoked potential. This limits the utility of the procedure for investigation pain physiology or for detecting damage to pain pathways.

As described previously, repair of damaged myelin results in an axon that conducts more slowly than before the damage. Just as with nerve conduction studies, decreased amplitude of evoked signals may also reflect damage. However, since amplitude is a significantly more variable in evoked potential testing, only large differences in the amplitude of SSEPs are significant.

Most clinical SSEPs are evoked by stimulation of large-diameter mixed nerves of the periphery (such as the median, ulnar, peroneal or tibial). These nerves are composed of sensory nerve fibers from many nerve roots, limiting the ability to identify damage to a single nerve root. While there has been some discussion of the value of mixed nerve SSEPs in the identification of radiculopathy (such as by using the fibular [peroneal] nerve SSEP for the L5 nerve root) most investigators have not found this to be of particular value. SSEPs are used predominately in intraoperative monitoring during spinal surgery and instrumentation. If SEP abnormalities occur during the surgery the surgeon is alerted and changes in the operative procedure are implemented. In addition SSEPs can be used to assess the prognosis of patients suffering severe anoxic brain injury.

Late Potentials

Late potentials are electrodiagnostically-elicited responses in muscle that appear more than 10-20 milliseconds after stimulation of motor nerves. They have been termed “late-potentials” because they take substantially longer to appear than the direct responses to stimulation of motor nerves (described in the section on nerve conduction studies as the CMAP). There are two distinct types of late responses, the H-reflex (Figure 6) and the F-response (Figure 7).


The first type of “late response” is called the H-reflex, named in honor of Hoffmann, who first described this response in 1918. The pathway for this reflex and the significance of abnormalities is easiest to understand by recognizing that it is basically the electrophysiologic equivalent of the muscle stretch reflex. The H-reflex is most commonly tested by electrical stimulation of the tibial nerve, with recordings from the gastrocnemius/soleus muscle complex (i.e., the triceps surae) (Figure 6). Therefore, this response utilizes the same neural pathway as the ankle jerk reflex.

Understanding of the H-reflex it is aided by some knowledge of the technical details of the procedure. Electrical stimulation will depolarize the largest, most heavily myelinated nerve fibers at a lower stimulus intensity than is required to activate other smaller nerve fibers. Since the largest nerve fibers in a peripheral nerve are those arising from muscle stretch receptors, there should be a stimulus intensity that activates muscle stretch afferent nerve fibers without directly activating many motor nerve axons (which are slightly smaller in diameter). When muscle stretch sensory fibers are stimulated (whether it be by electrical impulse or by tapping the tendon of the muscle), a monosynaptic reflex contraction will be elicited in the muscle. Because this response must traverse the sensory axon all the way back to the spinal cord before synapsing on the motor neuron, and since the motor response must then traverse the length of the motor axon to reach the triceps surae muscle, this reflex takes a long time (at least in electrodiagnostic terms). That is where the designation of “late potential” comes from. Theoretically, this reflex can be elicited from virtually any muscle. However, in practical terms, only the triceps surae muscle produces H-reflexes that are reliable enough to be clinically useful. Therefore, when a clinical electrodiagnostic procedure reports an H-reflex, the test has evaluated the integrity of the reflex arc from the tibial nerve through the spinal cord and back to the triceps surae. Damage to any portion of the reflex arc, including the sciatic nerve or the S1 sensory or motor nerve root, can result in loss or slowing of the reflex response. Additionally, the amplitude of response (expressed as a ratio of the reflex response to the maximum direct motor response - the so-called H/Mmax) is also reliable enough to be of diagnostic value. Since the H-reflex is mediated primarily over the S1 nerve root (just like the ankle jerk reflex), it is a sensitive test for S1 radiculopathy. However, once the reflex arc has been damaged, it often does not return to normal (making the test less useful in investigating the question of recurrent radiculopathy). While the H-reflex may be viewed as an electrical test of the ankle jerk, or Achilles’ tendon reflex, there are some differences that should be noted. For example, as opposed to the clinical ankle jerk, the H-reflex and be precisely quantified (both in terms of latency and amplitude) and, therefore, may be a more useful index to follow with time or treatment. Additionally, the H-reflex can be elicited from many patients even when the ankle jerk can not be elicited due to age.

With the notable exception of the triceps surae muscle, the H reflex is very difficult to elicit. This limits the H-reflex to being a sensitive specific and quantitative test of sciatic nerve and S1 nerve root function. This may be of utility in investigating patients with suspected S1 radiculopathy.


The second type of late potential is the F-response. This is a response that occurs in muscles during a motor nerve conduction study long after the initial contraction of the muscle (the CMAP, see above), While the CMAP usually appears within several milliseconds (depending on how close the stimulus point is to the muscle), depending on the stimulus site, another response can be normally recorded in the muscle approximately 25 -55 milliseconds later (Figure 7). Since this response was first recorded in foot muscles, it came to be known as the F-response. Over time it was found that this late response was not a reflex in the usual definition.. The electrical impulse is trasmitted proximally along the motor axon from the site of initiation of the action potential. When this antidromic (opposite to the normal direction of conduction) depolarization reaches the motor neurons in the spinal cord, a percentage of these motor neurons are activated a second time. This results in an orthodromic electrical signal being conducted in the normal (orthodromic) direction from the spinal cord to the muscles innervated by the nerve. This second, later activation produces a small muscle contraction that is termed the F-response. Because the number of motor neurons that are re-activated is somewhat unpredictable, the amplitude of this signal is variable and, therefore, amplitude measurements are usually not used. However, delay in the F-response indicates some slowing of conduction of the motor axon. Since the F-response traverses more proximal portions of the motor axons (twice, in fact) it may be useful in the investigation of proximal nerve pathology such as root pathology seen in radiculopathy, Guillian Barre Syndrome, or Chronic Inflammatory Demyelinating Polyradiculopathy (CIDP).

The F–response is also very helpful in the confirmation of demyelinative peripheral neuropathies. In these neuropathies the F-responses may be quite prolonged.

Jump to: