New Matthew C.Kiernan,MBBS,PhD,FRACP Jon A.Jacobson,MD … · 2008. 9. 10. · American Association...

��American Association of Neuromuscular & Electrodiagnostic Medicine

Course H2005 AANEM COURSE H

AANEM 52nd Annual Scientific MeetingMonterey, California

NEW DIRECTIONS IN

NEUROPHYSIOLOGIC

ASSESSMENT OF NERVE

AND MUSCLE

Matthew C. Kiernan, MBBS, PhD, FRACPSeward B. Rutkove, MD

Jon A. Jacobson, MDTimothy J. Doherty, MD, PhD, FRCPC

Daniel W. Stashuk, PhD

2005 COURSE HAANEM 52nd Annual Scientific Meeting

Monterey, California

Copyright © September 2005American Association of Neuromuscular & Electrodiagnostic Medicine

421 First Avenue SW, Suite 300 EastRochester, MN 55902

PRINTED BY JOHNSON PRINTING COMPANY, INC.

Matthew C. Kiernan, MBBS, PhD, FRACPSeward B. Rutkove, MD

Jon A. Jacobson, MDTimothy J. Doherty, MD, PhD, FRCPC


New Directions in Neurophysiologic Assessment of Nerve and Muscle

��


Faculty

ii

Matthew Kiernan, PhD, FRACP

Senior Lecturer

Department of Neurology

Prince of Wales Medical Clinical School

University of New South Wales

Sydney, Australia

Dr. Kiernan is a senior lecturer in neurology at the Prince of Wales ClinicalSchool Faculty of Medicine at the University of New South Wales and asenior scientist at the Prince of Wales Medical Resarch Institute in Sydney,Australia. He is also a consultant neurologist at the Institute ofNeurological Sciences at the Prince of Wales and Prince Henry Hospitals.Dr. Kiernan is a board member of the Australian Brain Foundation, allo-cating funds to research and medical education for the treatment and pre-vention of neurological disorders. He is a member of the steering commit-tee developing the Australian Motor Neurone Disease Registry. His currentprojects and interests include the pathophysiology of uremic neuropathy,nerve excitability in spinal cord injury patients, nocturnal hypoventilationin motor neuron disease, and mechanisms of chemotherapy induced neu-rotoxicity.

Seward B. Rutkove, MD

Associate Professor of Neurology

Harvard Medical School

Department of Neurology

Beth Israel Deaconess Medical Center

Boston, Massachusetts

Dr. Rutkove is currently an associate professor of neurology at HarvardMedical School and the director of the Division of Neuromuscular Diseasein the Department of Neurology at Beth Israel Deaconess Medical Center.He has held a longstanding interest in novel neurophysiologic techniquesand enhanching current methodologies. His other interests include thetreatment of myositis, development of impedance technologies in the eval-uation of neuromuscular disease, and studying the effect of temperature onnerve and muscle function. Dr. Rutkove is also the author of AANEMMonograph #14: The Effects of Temperature on NeuromuscularElectrophysiology. He has been the recipient of multiple National Institutesof Health (NIH) research grants and a member of the BioengineeringNIH study sections.

Jon A. Jacobson, MD

Associate Professor

Department of Radiology

University of Michigan

Ann Arbor, Michigan

Dr. Jacobson earned his medical degree from Wayne State UniversitySchool of Medicine in Detroit, Michigan. He then performed a residencyin diagnostic radiology at Henry Ford Hospital and a fellowship in mus-culoskeletal radiology at the University of California, San Diego. He is cur-rently an associate professor of radiology at the University of Michigan andthe director of the division of musculoskeletal imaging at the University ofMichigan Hospitals. Dr. Jacobson has done editorial work for many jour-nals, including the American Journal of Roentgenology, Arthritis Care andResearch, Journal of Ultrasound in Medicine, RadioGraphics, and SkeletalRadiology, among others. In 2003, he won the Excellence in TeachingAward from the University of Michigan.

Timothy J. Doherty, MD, PhD, FRCPC*

Assistant Professor

Departments of Clinical Neurological Sciences and RehabilitationMedicine

University of Western Ontario

London, Ontario, Canada

Dr. Doherty is an assistant professor in the Departments of ClinicalNeurological Sciences and Rehabilitation Medicine at the University ofWestern Ontario in London, Ontario. In March 2005, he was named theCanada Research Chair in Neuromuscular Function in Health, Aging, andDisease. He is a consultant physiatrist and clinical neurophysiologist atLondon Health Sciences Centre, and is the current President of theCanadian Association of Physical Medicine and Rehabilitation. Dr.Doherty is the author of over 50 peer-reviewed papers. His research focus-es on the examination of the motor system and motor units in health,aging, and disease. He and his collaborators have developed a number ofelectrophysiological methods for examining the numbers and properties ofhuman motor units, including EMG decomposition analysis. These meth-ods are being applied to the study of aging, motor neuron disease, andhereditary neuropathy.

Course Chair: Timothy J. Doherty, MD, PhD, FRCPC

The ideas and opinions expressed in this publication are solely those of the specific authors and do not necessarily represent those of the AANEM.

Dr. Doherty receives research support from Neuroscan, Inc. All other faculty had nothing to disclose.

iii


Contents

Faculty ii

Objectives iii

Course Committee iv

Excitability Testing in the Clinical Setting 1Matthew C. Kiernan, MBBS, PhD, FRACP

Electrical Impedance Myography: What High-frequency Alternating Current Can Tell Us About Muscle 7Seward B. Rutkove, MD

Ultrasound Imaging of Peripheral Nerve and Muscle 17Jon A. Jacobson, MD

Decomposition-based Quantitative EMG: From Theory to Application 25Timothy J. Doherty, MD, PhD, FRCPCDaniel W. Stashuk, PhD

CME Self-Assessment Test 37

Evaluation 41

Member Benefit Recommendations 43

Future Meeting Recommendations 45

O B J E C T I V E S —This course is designed to highlight a number of new methods for assessment of the neuromuscular system. As a resultof taking part in this course participants will (1) understand the basic concepts involved in excitability testing peripheral nerve, (2) learnthe potential role of electrical impedance myography in assessment of neuromuscular disorders, (3) learn how diagnostic ultrasound can aidin the assessment of disorders of nerve, muscle and tendons, and (4) understand the basic concepts of EMG signal decomposition and theirapplication.

P R E R E Q U I S I T E —This course is designed as an educational opportunity for residents, fellows, and practicing clinical EDX physiciansat an early point in their career, or for more senior EDX practitioners who are seeking a pragmatic review of basic clinical and EDX prin-ciples. It is open only to persons with an MD, DO, DVM, DDS, or foreign equivalent degree.

AC C R E D I TAT I O N S TAT E M E N T —The AANEM is accredited by the Accreditation Council for Continuing Medical Education toprovide continuing medical education (CME) for physicians.

CME C R E D I T —The AANEM designates attendance at this course for a maximum of 3.25 hours in category 1 credit towards theAMA Physician’s Recognition Award. This educational event is approved as an Accredited Group Learning Activity under Section 1 of theFramework of Continuing Professional Development (CPD) options for the Maintenance of Certification Program of the Royal Collegeof Physicians and Surgeons of Canada. Each physician should claim only those hours of credit he/she actually spent in the activity. TheAmerican Medical Association has determined that non-US licensed physicians who participate in this CME activity are eligible for AMAPMR category 1 credit. CME for this course is available 9/05 - 9/08.

Please be aware that some of the medical devices or pharmaceuticals discussed in this handout may not be cleared by the FDA or cleared by the FDA for the spe-cific use described by the authors and are “off-label” (i.e., a use not described on the product’s label). “Off-label” devices or pharmaceuticals may be used if, in thejudgement of the treating physician, such use is medically indicated to treat a patient’s condition. Information regarding the FDA clearance status of a particulardevice or pharmaceutical may be obtained by reading the product’s package labeling, by contacting a sales representative or legal counsel of the manufacturer of thedevice or pharmaceutical, or by contacting the FDA at 1-800-638-2041.

iv

Thomas Hyatt Brannagan, III, MDNew York, New York

Timothy J. Doherty, MD, PhD, FRCPCLondon, Ontario, Canada

Kimberly S. Kenton, MDMaywood, Illinois

Dale J. Lange, MDNew York, New York

Subhadra Nori, MDBronx, New York

Jeremy M. Shefner, MD, PhDSyracuse, New York

T. Darrell Thomas, MDKnoxville, Tennessee

Bryan Tsao, MDShaker Heights, Ohio

2004-2005 AANEM PRESIDENT

Gary Goldberg, MDPittsburgh, Pennsylvania

2004-2005 AANEM COURSE COMMITTEE

Kathleen D. Kennelly, MD, PhDJacksonville, Florida

INTRODUCTION

Clinical neurophysiology plays a critical role in the diagnosis,classification, and prognosis of neuropathies. In the case ofdemyelinating neuropathies, for instance, typical abnormali-ties include slowed nerve conduction velocities (CVs), pro-longed distal latencies, prolonged or absent H reflexes and F-waves, multifocal conduction block (CB), and varyingdegrees of denervation (Table 1). These abnormalities reflectunderlying pathological changes, particularly segmentaldemyelination and axonal damage. However, the degree ofconduction slowing documented using standard neurophysi-ological investigation does not correlate well with clinical dis-ability, and even when patients have fully recovered, CV mayremain permanently slow. The abnormalities of nerveexcitability that underlie conduction slowing, CB, andectopic impulse activity are not adequately explored by rou-tine nerve conduction studies (NCSs). Using demyelinatingneuropathy as an illustrative example, this manuscript willdiscuss new nerve excitability techniques complementary tostandard neurophysiological investigation that may providegreater insight into patient symptomatology and diseasepathogenesis.

CONVENTIONAL NERVE CONDUCTION STUDIES AND NERVEEXCITABILITY

In the last 30 years there have been few significant changes tothe neurophysiological investigation of patients with suspect-

ed demyelinating neuropathies. Motor and sensory NCSs, incombination with electromyography, have remained themethod of choice for the clinician investigating nerve func-tion in such patients. While routine NCSs can document thepresence of a neuropathy, they may not provide furtherinsight into disease pathophysiology. Measurements of actionpotential amplitude and latency are limited indices of func-tion, providing information relating to the number of con-ducting fibers, and the CV of the fastest fibers. These data donot always correlate well with a patient’s clinical status: on theone hand, even quite large changes in CV may be asympto-matic while, on the other hand, routine NCSs may be nor-mal when there is clinical weakness. Conduction velocity is anonspecific indicator of pathophysiology: it may bedecreased by cooling, membrane depolarization or hyperpo-larization, sodium channel blockade, axonal thinning,demyelination, or remyelination with short internodes.

Measurements of axonal excitability are known to be sensitiveto and capable of providing an indirect measure of restingmembrane potential.2 As such, these measures provide com-plementary information to conventional NCSs. Most tech-niques currently employed to assess nerve excitability rely onthe stimulus-based method of threshold tracking. Withthreshold-tracking, changes in the intensity of the stimuluscurrent required to generate a test potential of fixed ampli-tude are measured online by a computer, which in turnadjusts stimulus intensity to keep the amplitude of the subse-quent test potential constant. “Threshold” in this contextindicates the stimulus current required to produce a target

Excitability Testing in the Clinical Setting

Matthew C. Kiernan, MBBS, PhD, FRACP

Prince of Wales Medical Research Institute University of New South Wales and Institute of Neurological Sciences

Prince of Wales HospitalSydney, Australia

nerve action potential (e.g., 40% of maximum), and can beadjusted online by computer (“tracked”) during differentmaneuvers to follow changes in nerve excitability. For exam-ple, when axons are hyperpolarized, the test potential will besmaller, and stimulus intensity will be gradually increased bythe computer until the test potential has returned to its tar-get size, usually chosen to be around 40-50% of maximum(because this is on the fast rising phase of the S-shaped stim-ulus-response curve for the compound potential).Measurement of threshold depends on, and therefore pro-vides, an indirect measure of resting membrane potential.Resting membrane potential is in turn determined by a com-plex network of ion channels (mainly persistent sodium[Na+] channels, slow and fast potassium [K+] channels) andthe activity of the Na+/K+ pump.2

IMPULSE CONDUCTION IN DEMYELINATING NEUROPATHIES

Acute demyelination lowers the safety margin for impulseconduction, such that axons can become sensitive to shifts inmembrane potential, even when those shifts occur throughnormal physiological mechanisms.1 In critically conductingaxons, impulse conduction can be impaired by the effects ofheating and activity and probably by any mechanism thatproduces a significant shift in membrane potential, whetherdepolarizing or hyperpolarizing.

Raising temperature by 0.5o C can be sufficient to precipitateconduction failure in a critically conducting axon, and warm-ing commonly accentuates the deficit in, for example, multi-ple sclerosis, even to the point that warming is sometimesused as a provocative test. Conduction failure occurs becausewarming speeds up channel gating, affecting both activationand inactivation, and this decreases the time integral of theNa+ current at the node of Ranvier. In a critically conductingaxon, the duration of the driving current at the blockingnode can reach 1.0 ms,4 and CB can be precipitated orrelieved by maneuvers that manipulate the time course of thedriving current, such as changing temperature or the admin-istration of agents that interfere with Na+ channel inactiva-tion.

Activity hyperpolarizes axons. With brief high-frequencytrains (less than 10-20 impulses at 100-200 Hz), this is large-ly due to activation of a nodal slow K+ conductance.10 Theresulting hyperpolarization can increase threshold by approx-imately 40%, more than sufficient to jeopardize conductionin impaired axons, but the return to control excitabilityoccurs over 100-150 ms, such that this mechanism mightdisrupt the discharge pattern and limit the discharge rate, butit could not produce conduction failure by itself.

When an axon conducts long impulse trains, particularly at ahigh frequency, there is an accumulation of Na+ ions within

2 Excitability Testing in the Clinical Setting AANEM Course

Table 1 Electrodiagnosis of demyelinating neuropathy

Demyelinating Neuropathy

Neurophysiological features: demyelination +/- axonal degeneration

Acute Chronic

Delayed/absent H reflex, F waves Motor conduction slowing (>1nerve)

Decreased SNAP amplitude Partial conduction block/Dispersion

Prolonged distal latency Prolonged distal latency (>1 nerve)

Conduction slowing Prolonged/absent F waves (>1 nerve)

Conduction block

Increased temporal dispersion Supportive: Absent H reflexSensory slowing

EMG: Spontaneous activity (PSWs, fibs, occasional myokymia)Reduced recruitment

MMN

Conduction block in >1 motor nerves at uncommon sites of focal neuropathy

Normal sensory conduction velocity for segments with conduction block

Normal sensory amplitudes with distal stimulation of nerves with conduction block

EMG = electromyogram; MMN = multifocal motor neuropathy; PSW = positive sharp waves; SNAP = sensory nerve action potential.

the axon, and this activates the Na+/K+ pump to restore ionicbalance (Table 2). The stoichiometry of the pump is suchthat 3 Na+ ions are extruded in exchange for 2 K+ ions, andthe resulting imbalance in charge results in axonal hyperpo-larization. This activity-dependent hyperpolarization hasbeen demonstrated in human sensory and motor axons and,importantly, it can be produced by natural activity.13 Theextent and duration of the hyperpolarization depends on thedischarge rate and train length, and can result in an increasein threshold of approximately 40% that takes many tens ofminutes to decay to control excitability. Importantly, volun-tary contractions lasting as little as 15 seconds can increasethe threshold of motor axons by 10-15% and this increasedecays over some 5-10 minutes.13 Such changes are likely tobe clinically relevant: a conservative estimate of the safetymargin for impulse conduction in a series of patients withchronic inflammatory demyelinating polyneuropathy(CIDP) suggested that significant conduction failure ofapproximately 14% would occur if the axons hyperpolar-ized.3 For the same impulse load, the extent of the activity-dependent hyperpolarization seems to be greater for motoraxons than for sensory axons.9,13 An important factor in thisdifference is probably the difference in the hyperpolarization-activated cation conductance (IH).

In patients with inflammatory demyelinating polyneu-ropathies, normal activity-dependent hyperpolarization canprecipitate conduction failure at sites of impaired function.This was first demonstrated in multifocal motor neuropathy(MMN)5 and subsequently in CIDP.3 There is nothing spe-cial about activity: any process that produces sufficient hyper-polarization will produce clinically significant CB at patho-logical sites in these disorders, provided that a sufficientnumber of axons are critically conducting. The release ofischemia results in a post-ischemic hyperpolarization, andthis too can precipitate conduction failure. Paradoxically, inCIDP, conduction failure may also occur during ischemiaduring the depolarizing shift in membrane potential. Thisprobably occurs because depolarization inactivates transientNa+ channels, thus decreasing the availability of functioningchannels in an axon that is critically dependent on the size ofthe Na+ current. It is also possible that ischemia produces anischemic metabolite that blocks Na+ channels, a mechanismthat would further limit the number of Na+ channels avail-able for the action current.

The important message is that critically conducting axons aredelicately poised. Conduction may block if membranepotential is too far from threshold (i.e., the axon is hyperpo-larized) or if the Na+ current becomes inadequate (because ofheating or because of a limitation on the number of function-ing Na+ channels).8 Significant changes in activity or signifi-cant shifts in membrane potential, whether depolarizing orhyperpolarizing, may be sufficient to produce a transientworsening of symptoms. Clinical fluctuations are well docu-mented in multiple sclerosis; they are as likely in demyelinat-ing polyneuropathies if a sufficient number of axons can onlyjust maintain conduction.

MULTIFOCAL MOTOR NEUROPATHY AND AXONALHYPERPOLARIZATION

Multifocal motor neuropathy is a disease process character-ized by motor nerve involvement, with sparing of sensoryfunction (other than tendon jerks), producing a syndrome inaffected patients of slowly progressive muscle atrophy, weak-ness and fasciculation.11 A critical diagnostic feature inMMN is the demonstration of CB in multiple peripheralnerves on electrophysiological investigation.12 Conductionblock involves only motor axons, with sensory conductionspared across the lesion.

The presence of “positive” symptoms and signs such as mus-cle cramp, myokymia, and fasciculation in the context of pre-dominantly negative features such as depressed tendon jerks,muscle atrophy, and weakness remains unexplained in thesepatients. Nor has the mechanism of the CB itself been eluci-dated. Limited information is available from pathological

AANEM Course New Directions in Neurophysiologic Assessment of Nerve and Muscle 3

Table 2 Mechanism of activity-dependent conduction block

Impulse Train

accumulation Na+ inside, K+ outside axon

stimulation of “electrogenic” Na+ /K+ pump

x3 Na+ out for x2 K+ in

Axonal hyperpolarization

(5 treshold of motor axons by ~40% for 10-15 min after MVC for 1m)

Normal axons Impaired axons

safety margin 5:1 [500%] critically lowered margin

secure conduction conduction block

MVC = maximal voluntary contraction

5

5

5

5

5 5

studies, some of which provide evidence favoring demyelina-tion.

Axonal properties remote from foci of CB are normal, find-ings that support the contention that MMN is not a general-ized disease with a focal presentation. However, excitabilitystudies have also been undertaken in MMN patients just dis-tal to the site of CB in affected nerves, with strikingly abnor-mal findings.7 The most significant alterations in excitabilityproperties in the MMN patients were: (1) the reduction inminimum slope of the current/threshold (I/V) relationship(indicating a reduction in input conductance); (2) the “fan-ning-out” of threshold electrotonus; and (3) the increase insuperexcitability recorded during the recovery cycle. All theseexcitability parameters, highly abnormal in MMN patients,depend on the resting conductance of the paranodal andinternodal axon membrane.

One way that the resting conductance of the axonal mem-brane can be reduced is by means of membrane hyperpolar-ization. To explore the possibility that the nerve distal to CBin these patients was hyperpolarized, the results from patientswere compared to those from recordings of polarized andischemic nerves in healthy subjects.6 The patient axonsbehaved as if hyperpolarized or post-ischemic. However,subexcitability was preserved in the MMN patients, suggest-ing that their axons were in a post-ischemic state. Whereasboth hyperpolarization and post-ischemia increase superex-citability, late subexcitability is decreased by hyperpolarizingcurrents. However, it remains unchanged when the nerve is

post-ischemic, and unchanged or even increased in MMNpatients, despite the changes consistent with hyperpolariza-tion in the other excitability parameters.

These comparisons suggested that the nerve distal to the siteof CB in these patients with MMN was behaving as if hyper-polarized, most closely resembling a post-ischemic nerve. Fora critical test of whether hyperpolarization was responsiblefor the abnormal membrane properties, a depolarizing cur-rent of 0.5 mA was applied to the nerves of the MMNpatients, to test whether depolarization could reverse theabnormalities. All excitability abnormalities were improvedby applied depolarization.

It is notable that, in these MMN patients, the hyperpolariza-tion was not merely long-lasting (as in a post-ischemic state)but represented a stable steady state. Excitability recordingstaken 10 weeks apart showed almost identical excitabilityabnormalities. Persistent overactivity of the Na+/K+ pumpwould require a persistent intra-axonal source of Na+ ions todrive the pump. These Na+ ions could not be continuouslyentering the axon at the recording site, otherwise there wouldbe a net membrane depolarization.

Longitudinal diffusion or transport along the axon couldprovide an intra-axonal source of Na+ ions (Figure 1). Giventhat the lesion in these patients is focal, Na+/K+ pump activ-ity could be blocked by, for example edema, or alternatively,by antibodies directed against the protein components of thepump. In either case, prevention of Na+/K+ pump function

4 Excitability Testing in the Clinical Setting AANEM Course

Proximal Lesion Distal

Na+

Membrane depolarizationHyperactive Na+ pump

Figure 1 Distal hyperpolarization in multifocal motor neuropathy suggests depolarization at lesion site.

could produce a depolarizing block with intracellular accu-mulation of Na+ at the site of the lesion. Disruption of theblood nerve barrier might increase the K+ concentration ofthe endoneurial fluid, and in doing so, further aggravate anydepolarization block. If Na+ influx continued, a steady statewould be achieved only by Na+ ions moving intracellularlyalong the axon to a site where the pump was still working,and this would result in overactivity and membrane hyperpo-larization. Therefore, depolarization at the site of the lesionwould co-exist with chronic hyperpolarization on one orboth sides of this site. Such a lesion, with adjacent ischemicand post-ischemic lengths of nerve, would be likely to gener-ate ectopic activity, a hallmark of MMN, with patients expe-riencing fasciculation or myokymia in the presence of CB.

Other clinical symptoms and studies provide further supportfor this hypothesis. Cold paralysis described in monomelicamyotrophy also occurs in patients with MMN. The electro-genic Na+/K+ pump is temperature-sensitive, with slowerkinetics at lower temperatures. Cooling would exacerbate theeffect of the already compromised Na+/K+ pump function,increasing the depolarizing CB. This contrasts with theexpected effect of cooling in alleviating CB due to demyeli-nation. Similarly digitalis, a known blocker of Na+/K+ pumpactivity, paradoxically exacerbated the “fanning-out” changesseen in threshold electrotonus for a patient with MMN.Unfortunately attempts at tracking excitability changes moreproximally toward the site of the focal lesion in patients withMMN were impossible as the nerve became completely inex-citable. Finally, axonal hyperpolarization is consistent withthe selective increase in motor nerve threshold described inpatients with MMN at the site of the lesion.

SUMMARY

Conventional NCSs remain the gold standard for the electro-diagnosis of demyelinating neuropathy. Electrodiagnosticfindings reflect a combination of demyelination and axonalloss. Demyelination lowers the safety margin for impulseconduction and activity-dependent CB may be induced.Novel nerve excitability techniques provide complementary

information about membrane potential and axonal ion chan-nel function not investigated by conventional methods

REFERENCES

1. Bostock H, Grafe P. Activity-dependent excitability changes in nor-mal and demyelinated rat spinal root axons. J Physiol 1985;365:229-257.

2. Burke D, Kiernan MC, Bostock H. Excitability of human axons.Clin Neurophysiol 2001;112:1575-1585.

3. Cappelen-Smith C, Kuwabara S, Lin CS, Mogyoros I, Burke D.Activity-dependent hyperpolarization and conduction block inchronic inflammatory demyelinating polyneuropathy. Ann Neurol2000;48:826-832.

4. Inglis JT, Leeper JB, Wilson LR, Gandevia SC, Burke D. The devel-opment of conduction block in single human axons following a focalnerve injury. J Physiol 1998;513:127-133.

5. Kaji R, Bostock H, Kohara N, Murase N, Kimura J, Shibasaki H.Activity-dependent conduction block in multifocal motor neuropa-thy. Brain 2000;123:1602-1611.

6. Kiernan MC, Bostock H. Effects of membrane polarization andischaemia on the excitability properties of human motor axons. Brain2000;123:2542-2551.

7. Kiernan MC, Guglielmi JM, Kaji R, Murray NM, Bostock H.Evidence for axonal membrane hyperpolarization in multifocalmotor neuropathy with conduction block. Brain 2002;125:664-675.

8. Kiernan MC, Isbister GK, Lin CS, Burke D, Bostock H. Acutetetrodotoxin-induced neurotoxicity following ingestion of puffer fish.Ann Neurol 2005;57:339-348.

9. Kiernan MC, Lin CS, Burke D. Differences in activity-dependenthyperpolarization in human sensory and motor axons. J Physiol2004;558:341-349.

10. Miller TA, Kiernan MC, Mogyoros I, Burke D. Activity-dependentchanges in impulse conduction in a focal nerve lesion. Brain1996;119:429-437.

11. Parry GJ Clarke S. Pure motor neuropathy with multifocal conduc-tion block masquerading as motor neuron disease. Muscle Nerve1988;11:103-107.

12. Sumner A. Consensus criteria for the diagnosis of partial conductionblock and multifocal motor neuropathy. In: Kimura J, Kaji R, editors.Physiology of ALS and related diseases. Amsterdam: Elsevier; 1997. p221-227.

13. Vagg R, Mogyoros I, Kiernan MC, Burke D. Activity-dependenthyperpolarization of motor axons produced by natural activity. JPhysiol 1998;507:919-925.


6 AANEM Course

THE BEGINNINGS OF ALTERNATING CURRENT AND ITSEARLY APPLICATIONS

The earliest experiments in generating electricity, beyondthat of simply collecting static charge, were performed in thelast decade of the eighteenth century by Alessandro Volta. Heproduced direct current (DC) by placing a series of silver andzinc blocks in the form of a vertical column (his “electricpile”). Two decades later, Michael Faraday realized that move-ment of a magnet could induce the flow of electric current ina wire and in 1831 built the first DC generator. By the mid-nineteenth century, the great Scottish physicist James ClerkMaxwell identified the laws governing electricity and mag-netism and within two decades, electricity, mostly in theform of DC, was beginning to find practical uses, includingin artificial lighting (the incandescent lamp being inventedby Thomas Edison in 1879). Although DC appeared to holdgreat promise, and was promoted by Edison and GeneralElectric, its use was limited due to difficulty in transmittingit across long distances. By the 1880s it was becomingincreasingly clear that alternating current (AC) might be thebetter option for widespread power usage since it could besent over vast distances effectively through the use of step-uptransformers at the energy source and step-down transform-ers near the location of consumption. But it was not untilNikola Tesla developed an effective AC motor and improve-ments to AC generators a few years later that the use of AC

started becoming widespread. The true proof of AC was dra-matically demonstrated during the Columbian Exhibition of1893, where AC generators provided all the electricity for theentire fair.

Other uses for AC soon became apparent. The most imme-diate was the possibility of using AC for the generation ofelectromagnetic radiation. Based on Maxwell’s theories,Hertz in the mid-1880s demonstrated that an electric sparkcould create current flow in a distant copper coil. Althoughthis was a relatively crude experiment, it proved that suchtransmission was possible. Soon Tesla was producing AC athigher and higher frequencies which allowed for more effec-tive radiation of signals from the wires. The first successfulwireless transmitters and receivers were developed, and in1904 Marconi oversaw the first transmission of radio com-munication across the Atlantic Ocean.

It was soon realized that measuring how AC flowed througha wire could tell a person something about the wire’s compo-sition. Similarly, since AC (especially at higher frequencies)could also easily flow through air or other mediums, it wasappreciated that studying how the AC was affected by themedium could tell a person something about the characteris-tics of that medium. By around the turn of the twentieth cen-tury, the first studies using AC to probe the characteristics ofsubstances were initiated. These techniques were called

7

Electrical Impedance Myography: WhatHigh-frequency Alternating Current Can

Tell Us About Muscle

Seward B. Rutkove, MD

Chief, Division of Neuromuscular DiseaseBeth Israel Deaconess Medical Center

Associate Professor of NeurologyHarvard Medical SchoolBoston, Massachusetts

impedance measurements, since the term “impedance” cap-tured the two main electrical properties of a medium: itsresistance to current flow and its reactance, the oppositionto change in the direction of current flow with each cycle ofAC.

EARLY BIOLOGICAL USES OF IMPEDANCE TECHNIQUES

By the first two decades of the twentieth-century impedancemeasurements were being made on many different structures,including biological ones. Between the years of 1910-1913,R. Höber performed detailed impedance measurements onred blood cells in solution, showing a frequency-dependenceto the applied measurements and also demonstrating thatupon lysing of the red cells, the resistance to current flowdecreased. In 1926, it was demonstrated that breast tumorshad substantially lower electrical “capacity” than normalbreast tissue and thus impedance measurements could beused to assist in the diagnosis of tumors.

Although of considerable theoretical interest, electricalimpedance methods in biology continued to find only limit-ed applications until the 1950s and 1960s when it becameclear that these measurements could be used to assess bodycomposition in some fashion, since the proportion of freeand intracellular water relative to the amount of adipose tis-sue would affect impedance values. Hence empiric relation-ships could be drawn between measured impedance valuesand lean body mass. Such measurements could be obtainedby running a current at 50 kHz between the hand and footand measuring the voltage difference between these twopoints. Eventually, a minor industry developed around theevaluation of 50 kHz current flow to assist in the determina-tion of body fat content, and today devices to measure thisare commonplace in fitness centers and nutritionist’s offices.

Outside the world of nutrition, bioimpedance measurementsfound other uses in medicine. The evaluation of blood flowusing impedance plethysmography was first described in the1930s. Moreover, the technique of electrical impedancetomography (EIT) was introduced. In this procedure, electri-cal currents are applied and detected at multiple sites on thebody along an area of interest with the goal of developing apicture of the internal anatomy. Electrical impedance tomog-raphy has been used to some extent to evaluate the heart andother organs; however, given the incredible complexity of thecalculations involved to identify the various components ofsuch a system (the so-called “inverse problem”) this field hasnever met with great success. With the advent of both com-puted tomography and magnetic resonance imaging, therehas been little commercial impetus to pursue this work. Theuse of impedance measurements has found some commercialutility in evaluation of breast tumors, and a United States

Food and Drug Administration-approved device is nowbeing used for patient care at several sites in the United Statesand Europe.

IMPEDANCE MEASUREMENTS OF MUSCLE—WHY THEY MAKESENSE

Electrical impedance myography (EIM) is a term first intro-duced in 2003 to describe an electrical impedance-basedmethodology for the assessment of skeletal muscle. The setupfor the most basic of these techniques, linear EIM, is demon-strated in Figure 1. The technique has at its roots past workin electrical impedance but the approach is quite different,with the focus being on relatively small regions of tissue (e.g.,several inches of the thigh) rather than entire limbs or theentire body; in fact, the technique was initially referred to as“localized bioimpedance analysis.”

The application of impedance methodologies to the assess-ment of skeletal muscle makes excellent practical sense for thefollowing eight reasons: (1) skeletal muscle has a high watercontent and relatively low resistivity; (2) skeletal musclemakes up the largest component of the limbs; (3) skeletalmuscles in human beings are relatively large; (4) many skele-tal muscles are in close proximity to the outside world; (5)skeletal muscle, in both health and disease, is dense, relative-ly homogeneous, and predictable; (6) skeletal muscle isanistropic; (7) skeletal muscle’s contractile properties addanother potential dimension to impedance measurements;and (8) disease changes in skeletal muscle are important.These will be discussed in detail.

Skeletal Muscle Has a High Water Content and RelativelyLow Resistivity

Impedance methods in biological systems mainly evaluatemembrane health and structure. However, for the electricalcurrent to appropriately survey the membranes in a specificregion of interest, it is critical that the current preferentiallyflow to that region. Electricity follows the path of least resist-ance and little current will flow through areas with higherresistivity (the inherent electrical resistance of the tissue).Muscle’s resistivity is substantially lower than that of bone,skin, or subcutaneous fat, the other major components of alimb. Current will therefore preferentially flow through mus-cle.

Skeletal Muscle Makes Up the Largest Component of theLimbs

From a volumetric standpoint, there can be no question that,except in the most obese individuals, muscle is the majorconstituent of the arms and legs. While there are, in fact,

8 Electrical Impedance Myography: What High-frequency Alternating Current Can Tell Us About Muscle AANEM Course

some components of a limb that have lower resistivity thanmuscle (namely blood), the importance of these othercomponents is relatively minor since they make up only asmall fraction of the overall limb volume.

Skeletal Muscles in Human Beings are Relatively Large

While it may be self-evident that skeletal muscles in a humanbeing are relatively large, it simplifies the measurementprocess considerably since electrodes can be easily affixed tothe skin overlying several areas of a muscle or group of mus-cles.

Many Skeletal Muscles are in Close Proximity to the OutsideWorld

While some muscles are buried deep within the body, such asthe psoas, many lie directly beneath the skin. Electrical cur-rent can thus be directed through a region of specific interestand electrodes placed in a variety of configurations over thatarea. While impedance methods have been used with some

success for evaluation of cardiac structure, as has been noted,the simple fact that the heart is buried deep within a bodycavity makes it far less amenable to study.

Skeletal Muscle, in Both Health and Disease, is Dense,Relatively Homogeneous, and Predictable

Except in the most advanced neuromuscular diseases, muscletissue still consists of muscle fibers that are in relatively closejuxtaposition with one another. Also, the structure is relative-ly homogeneous with one major tissue type (muscle fibers)throughout; this situation is distinctly different from say thebrain or kidney where probing the tissue with impedancemethodologies is considerably more challenging given thevariety of tissues present.

Skeletal Muscle is Anisotropic

Anisotropy refers to the ability of a tissue to conduct electri-cal current preferentially in one direction versus another. Thebundled fiber structure of muscle ensures that electrical cur-


LaptopComputer

VoltageElectrode Array

Current Electrodes

Computer Controlled/Manual ElectrodeSwitching Box

BioimpedanceDevice

Figure 1 50 kHz, linear-EIM being performed on the left quadriceps.

EIM = electrical impedence myography.

rent will preferentially flow parallel to the fibers rather thanacross them. Whereas in some sense this may complicate themeasurement of impedance, it also opens up a whole newwindow into the evaluation of muscle since changes in theanisotropy may help evaluate disease status.

Skeletal Muscle’s Contractile Properties Add AnotherPotential Dimension to Impedance Measurements

Unlike many tissues in the body, muscle is not static. It cancontract and relax. The entire process of muscle contractionis complex, but results in a substantial change in the structureof the tissue which can be easily measured with impedancemethods.

Disease Changes in Skeletal Muscle are ImpedanceFriendly

Many of the changes that occur in disorders that affect mus-cle are potentially accessible to impedance measurements.Muscle fiber atrophy, variation in fiber size, the presence offat and connective tissue in the endomysium, changes inmembrane structure, muscle edema/inflammation, andchanges in the contractile ability of the tissue may all con-tribute to relatively diverse impedance signatures dependingon the disease state.

THE BASIC METHODS OF ELECTRICAL IMPEDANCEMYOGRAPHY

Electrical impedance myography is still under development,and the final techniques that emerge and their nomenclatureremain to be determined. However, a collection of variationson the basic technique of linear-EIM has been establishedthat have the potential for the assessment of muscle. Thebasic setup, shown in Figure 1 and Figure 2, demonstrates asimplified electrical schematic.

Linear, Single-Frequency Electrical Impedance Myography

Linear, single-frequency EIM, as suggested, is the most basicform of the technique: current is applied at a distance at onlya single frequency of 50 kHz and the measurements are madelinearly along the axis of the limb. This simple technique maybe best for following a disease in an individual sequentiallyover time.

Linear, Multifrequency Electrical Impedance Myography

Linear, multifrequency-EIM represents a simple extension ofthe linear technique by gathering impedance measurements

at multiple frequencies, usually ranging from 1 kHz to 3MHz. These data have demonstrated a fascinating range ofbehavior in different neuromuscular diseases and may haveimportant implications for disease diagnosis.

Multidirectional Electrical Impedance Myography

Multidirectional-EIM represents a variation in the techniquewhich is geared toward the evaluation of muscle anisotropy.Rather than applying current at a distance, current is appliednear the voltage electrodes and the voltage difference meas-ured at different angles relative to muscle fiber direction. Thiscan be performed at one or many frequencies.

Dynamic Electrical Impedance Myography

Dynamic-EIM consists of performing one of the aforemen-tioned forms of EIM during muscle contraction. The musclecontraction can be voluntary or electrically induced (by stim-ulation of the nerve supplying the muscle[s] of interest).

BASIC ELECTRICAL IMPEDANCE MYOGRAPHY PARAMETERS

In EIM, muscle tissue is modeled as a complex circuit (Figure2). The circuit model representing each element (of two resis-tors and capacitor, shown at the bottom of the figure) datesback to the 1920s. The basic concept is that by measuring thevoltage drop at any two points in the circuit and knowing theapplied current, the overall impedance from this tissue can beobtained and from that, one can calculate its resistance (R)and reactance (X). Nonetheless, both values, R and X, havecertain limitations, in part being dependent on the muscleshape and size. In order to help eliminate these effects, oneobtains the phase (ρ) of the muscle. The phase is defined asarctan (X/R). A separate value for θ can obtained betweeneach pair of voltage electrodes; these values can then be aver-aged and for this, the term θavg is used.

Another parameter that can be calculated is the resistivity ofthe muscle, designated ρ; this variable is less straightforwardto obtain than X or R. The resistivity, as noted earlier, is ameasure of the ability of a muscle to conduct an applied elec-trical current which can provide novel information on themakeup of the tissue. For example, increased connective tis-sue and reduced water content that occur in myopathic stateswill lead to increased muscle resistivity. In order to calculatethis value, however, one needs to know the approximate vol-ume of the muscle tissue underlying the electrodes. This isachieved by making girth measurements of the limbs and tak-


ing ultrasound measurements to calculate the size of the boneand the skin-subcutaneous fat layers.

APPLICATIONS OF ELECTRICAL IMPEDANCE MYOGRAPHY INHEALTH AND DISEASE

Normality, Muscle Strength, and Aging

Like many neurophysiologic parameters, muscle impedanceparameters also demonstrate an age-dependence with valuesdecreasing prominently over age 70 years. Figure 3 demon-strates this characteristic decline. It is anticipated that whilethis age-dependence may complicate analysis of the data, justas it does for other standard neurophysiologic methods, italso has the potential to provide interesting information onsarcopenia of the elderly. However, physicians are currently inthe process of defining age- and sex-dependent lower limitsof normal for different muscles, a key element to makingEIM a practicable technique.

Neurogenic and Myopathic Disease Progression

The most straightforward and easily implemented applica-tion for EIM is the quantification of neuromuscular diseaseprogression and remission. Standard electrophysiologic test-ing has proven itself useful in the actual diagnosis of disordersand differentiating nerve disease from muscle disease, butgenerally does not serve as good a role in quantifying thedegree of abnormality and the progression or remission of adisorder in a single person. For this reason, other techniqueshave been sought, such as the motor unit number estimation(MUNE) for the assessment of disease progression in motorneuron disease.

Electrical impedance myography is ideally suited for this pur-pose in myopathic and neurogenic disease alike. Even themost simple, quickly applied version of single-frequency, lin-ear EIM, can provide rapid information in this regard. Forexample, Figure 4 shows the change in phase in a group ofpatients with amyotrophic lateral sclerosis over time.

A simple explanation for this reduction in phase with diseaseprogression is as follows: regardless of the cause, as musclefibers atrophy, the capacitance of the muscle tissue decreases.In a network of resistors and capacitors (as shown in Figure2), the reactance of a single element,


Figure 2 Simplified circuit schematic/model for electricalimpedance myography.

Figure 3 50 kHz linear electrical impedance myography data ina normal population.

will behave according to the following equation:

Accordingly, as the capacitance of each individual celldecreases, the reactance of the entire network will go down as

well. This makes intuitive sense since if there are no capaci-tors in the system, the reactance would be zero (i.e., it wouldpurely resistive). The decreased reactance leads to a low θ,since the two are related via the equation θ = arctan(X/R),and, assuming all the segments of tissue studied are affected,θavg will decrease as well.

Disease Diagnosis

One of the more ambitious goals of this author’s work is todetermine whether the technique can be used to noninvasive-ly characterize the type of alterations present in the muscle.There is reason to anticipate that this may be possible, sincedifferent conditions produce different pathological and mor-phological changes in the muscle. For example, in acute myo-pathic disease, such as inflammatory or toxic myopathy, there


Figure 4 Example of how linear electrical impedance myography can be useful in the noninvasive assessment of disease progression inpatients with amyotrophic lateral sclerosis.

is substantial muscle edema with associated inflammatoryinfiltrates. In contrast, in chronic myopathic or dystrophicdisease, a variety of alterations will be present including a mixof muscle fiber hypertrophy and atrophy with substantialquantities of endomysial connective tissue. In acute neuro-genic disease, there will be pockets of atrophied muscle fibers(in the acute stage) and fiber type grouping in the chronicstate. The wealth of structural differences in the muscle islikely to lead to a variety of electrical signatures dependent onthe exact pathology present, perhaps best appreciated byusing different frequencies of applied current (multifrequen-cy-EIM), as shown in Figure 5. Consistent with this hypoth-esis, one early trend being observed is that in neurogenic dis-ease, reactance values fall into a relatively narrow rangeregardless of the frequency applied, whereas in myopathicdisease reactance values remain relatively normal even inpatients with more advanced disease. Continuing research

will explore the extent to which such differences can play arole in diagnosis.

Disuse and Rehabilitation

Since any cause of muscle atrophy will cause the EIM signa-ture to vary from normal, EIM has the capability of detect-ing changes due to disuse. Being able to identify and quanti-fy the presence of disuse atrophy could have a number ofimportant implications, from assisting in rehabilitation, toevaluation of muscle atrophy and sarcopenia in the elderly.Perhaps more provocatively, the technique could also be valu-able in monitoring muscle health during extended spaceflight, in part because the equipment necessary for the mostbasic assessment is small and light. In order to identify theeffects of disuse here on Earth, this author and colleagueshave been obtaining EIM data on patients pre- and post-cast-ing for orthopedic injuries. A prominent change has been


Figure 5 Example of different impedance profiles in patients with myopathic and neurogenic disease as compared to normal subjects.

ALS = amyotrophic lateral sclerosis; IBM = inclusion body myositis; NL = normal limits

found in EIM patterns that parallels the atrophy foundimmediately after cast removal and improves with therestoration of muscle bulk and strength during rehabilitation.

Novel Insights Into Contraction

Electrical impedance myography can also be performed dur-ing muscle contraction, both voluntary and stimulated. As asimple example of EIM’s potential use in muscle assessmentduring contraction, in Figure 6 reactance and resistance datais provided for the forearm of a normal individual voluntari-ly contacting against a handheld dynamometer and is com-pared to that from a patient with myotonic dystrophy type 1.Note the additional impedance dip that follows each contrac-tion, likely corresponding in some fashion to the delay inmuscle relaxation. Whereas it is anticipated that impedancemeasurements will provide unique data on contractile prop-erties of muscle, it remains to be determined if it can alsoprovide additional information on the actual depolarizationof muscle, a question currently attempting to be answered inthis author’s laboratory.

ONGOING AND FUTURE WORK

Many challenges are still faced in developing EIM into a use-ful technology. This author and colleagues are currently pur-suing several different avenues of research, including clinicalhuman work, which is mainly geared toward better identify-ing the relationships between impedance parameters andneuromuscular disease states. An additional key component

to this plan is using EIM in rat models of neuromuscular dis-ease to help confirm the association between the impedancesmeasures and the type and degree of pathological change.Another long-term goal of pursuing animal studies is thatEIM has the potential to serve as a simple method for screen-ing drugs for various mouse models of neuromuscular disease(such as the muscular dystrophy and superoxide dismutase 1-deficient mice). Learning how to perform EIM in rats pro-vides a first step along that path. This author and colleaguesalso plan to pursue work in pediatric disease since the non-invasive nature of the test may prove to be especially useful inthe assessment of children with muscle weakness. In additionto these programs, the technology is continuing to be refined,including developing pre-formed electrode arrays, improvinguser interfaces, and refining the methods used to obtain mul-tidirectional and dynamic data.

SUGGESTED READING

1. Aaron R, Huang M, Shiffman CA. Anisotropy of human muscle vianon-invasive impedance measurements. Phys Med Biol1997;42:1245-1262.

2. Aaron R, Shiffman CA. Using localized impedance measurements tostudy muscle changes in injury and disease. Ann N Y Acad Sci2000;904:171-180.

3. Gielen FL, Cruts HE, Albers BA, Boon KL, Wallinga-de Jonge W,Boom HB. Model of electrical conductivity of skeletal muscle basedon tissue structure Med Biol Eng Comput 1986;24:34-40.

4. Kushner RF. Bioelectrical impedance analysis: a review of principlesand applications. J Amer Coll Nutr 1992;11:199-209.

5. Lukaski HC. Regional bioelectrical impedance analysis: applicationsin health and medicine. Acta Diabetol 2003;40:S196-S199.


Figure 6 Example of impedance curves in a normal subject (A) and a patient with DM-1 performing isometric contractions. Unlike thesecurves, the force curves (not shown) show the same morphology in both subjects.

6. McAdams ET, Jossinet J. Tissue impedance: a historical overview.Physiol Meas 1995;16:A1-A13.

7. Rutkove SB, Aaron R, Shiffman CA. Localized bioimpedance analy-sis in neuromuscular disease. Muscle Nerve 2002;25:390-397.

8. Rutkove SB, Esper GJ, Lee KS, Aaron R, Shiffman CA. Electricalimpedance myography in the detection of radiculopathy. MuscleNerve (in press).

9. Shiffman CA, Aaron R, Amoss V, Therrien J, Coomler K. Resistivityand phase in localized BIA. Phys Med Biol 1999;44:2409-2429.

10. Shiffman CA, Aaron R, Rutkove SB. Electrical impedance of muscleduring isometric contraction. Physiol Meas 2003;24:213-234.


16 AANEM Course

INTRODUCTION

There is a growing interest in ultrasound imaging of the mus-culoskeletal system including peripheral nerves. This interestcan be attributed to the increased resolution provided by newtransducers, which allows exquisite demonstration ofanatomical structures. With current transducers, structures assmall as individual peripheral nerve fascicles can be visual-ized.21 Increasing interest in ultrasound may also be second-ary to the fact that ultrasound equipment is becoming morecompact and more affordable, although currently image res-olution is compromised. Ultrasound imaging has severalunique advantages over other imaging methods, such as mag-netic resonance imaging (MRI), as it applies to peripheralnerves and the musculoskeletal system.6 In addition todecreased cost and improved accessibility at many centers,ultrasound, compared to MRI, can employ dynamic imagingto assess disorders that only are evident with particular jointpositioning or dynamic states, such as active muscle contrac-tion. Ultrasound can also image an entire limb, includingcomparison with the other limb if required, in a shorter peri-od of time than MRI. Imaging of structures near hardware orother ferromagnetic structures is also possible with ultra-sound without the artifact present as opposed to MRI.8

Ultrasound allows for the evaluation of patients who areunable to undergo MRI because of ferromagnetic devices inspecific locations (such as aneurysm clips), or when a patienthas claustrophobia. In fact, with regard to evaluation of therotator cuff, patients prefer ultrasound to MRI.15

One must also realize that there are several disadvantages inusing ultrasound compared to MRI. The primary difficulty isultrasound’s learning curve, which is steeper than that ofMRI. Magnetic resonance imaging uses specific sequences tohighlight pathology on images acquired in anatomical planes.This relative uniformity allows one to become more quicklyfamiliar with anatomy and pathology. In contrast, althoughstandardized protocols are described for ultrasound, under-standing anatomy and identification of pathology requiresdedicated training under supervision, and a thorough under-standing of the structure imaged in three dimensions.

ULTRASOUND BASICS

An ultrasound unit consists of a central machine containingthe electrical and computer components, and the imagescreen. The size of this unit can range from a laptop comput-er to the size of an automated teller machine. This unit isconnected to a transducer through a cable. Images areacquired by holding or moving a transducer in contact withthe skin surface over the area of interest using interveningacoustic coupling gel. Images are acquired through propertiesof the piezoelectric crystals in the transducer where electricalsignal is converted to a pulse wave. This pulse wave thenpropagates through the soft tissues. When the sound waveinteracts with various soft tissue interfaces, it can reflect backto the transducer, where the pulse wave is converted back toelectric signal used to create the ultrasound image. An ultra-

17

Ultrasound Imaging of Peripheral Nerveand Muscle

Jon A. Jacobson, MD

Associate Professor of Radiology; Director, Division of Musculoskeletal RadiologyGandikota Girish, MBBS, FRCS, FRCR

Lecturer of RadiologyDepartment of Radiology

University of Michigan Medical CenterAnn Arbor, Michigan

sound wave can interact with soft tissues in various ways,which allows discrimination between the various types of tis-sues. For example, if a sound wave hits a soft tissue interfaceof significant differences in impedance (such as soft tissue-bone surface, or at soft tissue-gas interface), a majority of thesound beam reflects back to the transducer and therefore abright or hyperechoic line is registered where the bone sur-face is located. The ultrasound beam can also be absorbed orrefracted by the soft tissues. Various artifacts may be pro-duced in this manner. Most ultrasound machines have colorand possible power Doppler options. In this mode, bloodflow is demonstrated on the image as red or blue, which isdetermined by the direction the blood flow relative to thetransducer.

By convention, the horizontal top edge of the resulting ultra-sound image represents the skin surface where the transduc-er is located. From left to right, the ultrasound image can beviewed as though the soft tissues have been sectioned alongthe long axis of the transducer. The bottom of the image rep-resents the deepest part of the imaged tissues, farthest fromthe transducer. The image that is produced may containbright echoes (called hyperechoic), intermediate echoes (isoe-choic), low echoes (hypoechoic), or no echoes (anechoic) rel-ative to the echogenicity of skeletal muscle. Normal tendonsshows a fibrillar and hyperechoic echotexture, normal muscleshows predominately hypoechoic muscle tissue with inter-vening hyperechoic fascia, and the bone surface is very hyper-echoic with posterior acoustic shadowing (Figure 1).9 A nor-mal peripheral nerve will appear speckled in cross-sectionwith the nerve fascicles appearing hypoechoic and the sur-rounding connective tissue appearing hyperechoic (Figure2).21 A fluid collection will be anechoic with enhanced echoesdeep to the fluid. This is called posterior acoustic enhance-ment. An important artifact in imaging of tendon and mus-cle is termed anisotropy.4 This occurs when a tendon is notimaged perpendicularly to the sound beam, but ratherobliquely. Subsequently, a normal hyperechoic tendon willappear hypoechoic similar to an abnormal tendon.

An important aspect of peripheral nerve and muscle ultra-sound is the choice of transducer type. In general, a lineartransducer is optimal in visualizing a segment of peripheralnerve or muscle perpendicular to their intrinsic fibers. Thetransducer frequency choice is also critical. A higher frequen-cy transducer will have higher resolution, but this will be atthe expense of depth penetration. Therefore, to image a deepstructure such as around the hip, a lower frequency transduc-er will be required (5-7.5 MHz), but the resolution will bedecreased. When imaging a superficial structure, a high-fre-quency transducer will show the anatomic detail. In general,rotator cuff ultrasound is best imaged with at least 10 MHz,while superficial peripheral nerves are best seen with at least

12 MHz, although 15 or 17 MHz will show even moredetail. It is important to note that the small portable ultra-sound machines are currently somewhat limited with regardto available high-frequency transducers and therefore resolu-tion is less than that which can be achieved with a larger andmore powerful ultrasound machine.

PERIPHERAL NERVE ULTRASOUND

Normal Ultrasound Appearance

Normal peripheral nerves have a characteristic ultrasoundappearance.21 When imaged in cross-section, the nerve has aspeckled or honeycomb appearance with the individual nervefascicles appearing hypoechoic and the surrounding connec-tive tissue appearing hyperechoic (Figure 2).24 When imagedlongitudinal to the nerve fibers, the individual hypoechoicnerve fascicles can again be seen. One must be aware thatbecause the peripheral nerve has both hypoechoic and hyper-echoic areas, either may be accentuated depending on thesurrounding soft tissues. For example, when a peripheralnerve is imaged when surrounded by hypoechoic muscle tis-sue, the hyperechoic components of the nerve will be moreconspicuous. In contrast, when a peripheral nerve is imagedwhen surrounded by hyperechoic tendons, such as in thecarpal tunnel, the hypoechoic nerve fascicle components willbe more obvious.10 What is most important beyond under-standing the sonographic appearance of peripheral nerves ishaving knowledge of peripheral nerve anatomy. Without this,

18 Ultrasound Imaging of Peripheral Nerve and Muscle AANEM Course

Figure 1 Normal shoulder ultrasound. Ultrasound imagelongitudinal to the distal supraspinatus tendon showshyperechoic and fibrillar tendon (between arrows), relativelyhypoechoic deltoid muscle (D), and subacromial-subdeltoidbursa (*). Note reflective and hyperechoic surface of humerus(H) (skin surface is at top of image, lateral is at right).

a small peripheral nerve may not be recognized. It is impor-tant to use bone landmarks, as well as muscle and fascialplanes to identify peripheral nerves.

Compression Neuropathies

Common compression neuropathies of peripheral nervesinclude the median nerve in the carpal tunnel, the ulnarnerve in Guyon’s canal and the cubital tunnel, the posteriorinterosseous nerve at the supinator muscle, the posterior tib-ial nerve in the tarsal tunnel, and the digital nerve betweenthe metatarsal heads producing a Morton’s neuroma.14 Thehallmark of any compressive neuropathy is hypoechoic nerveswelling at or just proximal to the compression, with flatten-ing or compression distally. Direct pressure with the ultra-sound transducer can reproduce symptoms, which offersadditional evidence that the segment of nerve is abnormal.

Carpal tunnel syndrome (CTS) results from compression ofthe median nerve in the carpal tunnel. This can be due toprocesses that cause narrowing of the carpal tunnel (trauma)or crowding of the carpal tunnel (tenosynovitis, ganglioncyst). It may also be idiopathic, possibly due to overuse con-ditions. Imaging of the median nerve is in the transverseplane at the level of the volar wrist crease just proximal to thecarpal tunnel and pisiform bone. At this location, the hypoe-choic nerve fascicles of the median nerve are most obvious asthe nerve is surrounded by hyperechoic tendon.10 The cir-cumference of the median nerve is then measured using toolsbuilt into the ultrasound unit. It has been shown that a medi-an nerve area greater than 9 or 10 mm2 correlates with CTS(Figure 3).5,25 The compressed median nerve may also beabnormally hypoechoic, and distal evaluation will show flat-tening.

Ulnar nerve compression at the level of the elbow can occurin the cubital tunnel. By imaging in the transverse planebetween the medial epicondyle and olecranon process of theelbow, the ulnar nerve will be seen as a relatively hypoechoicstructure since it is surrounded by hyperechoic fat. Proximalswelling at this level can occur with compression, and a cross-sectional area greater than 7.5 mm2 correlates with cubitaltunnel syndrome (Figure 4).3 It is important to recognize thatin some cases the ulnar neuritis may be secondary to repeti-tive ulnar nerve dislocation and relocation. Therefore, thisshould be assessed. In fact, this is one of the advantages ofultrasound over MRI—dynamic imaging with elbow flexionand extension can be completed. With the transducer held inthe transverse plane fixed to the medial epicondyle, thepatient flexes the elbow.7 Abnormal translation of the ulnarnerve over the medial epicondyle apex indicates subluxationor dislocation. It is important to not apply too much pressurewith the ultrasound transducer, as this may inhibit the move-ment of the ulnar nerve with elbow flexion. During elbowextension, the ulnar nerve typically relocates and a snap iscommonly felt through the transducer. Ulnar nerve subluxa-tion may occur in asymptomatic individuals, so it is impor-tant to correlate this finding with swelling of the ulnar nerveand patient symptoms to determine significance. A conditioncalled “snapping triceps syndrome” has also been describedwhere the ulnar nerve dislocates over the medial epicondyleand also over the medial head of the triceps muscle.7 Thiscondition often affects patients with increased muscle bulkrelated to weight training.

Morton’s neuroma is the result of plantar digital nerveentrapment or impingement between the metatarsal heads,


Figure 2 Normal median nerve. Ultrasound images transverseto the median nerve (arrowheads) shows individual hypoechoicnerve fascicles (arrow) (T = tendons).

Figure 3 Carpal tunnel syndrome. Ultrasound imagetransverse through the median nerve at the level of the volar wristcrease shows hypoechoic swelling (arrows). Note normaltendons (T), several of which appear mildly hypoechoic due toanisotropy.

commonly between the third and fourth heads. This repeti-tive injury can cause fibrosis, which produces a swollen andhypoechoic appearance seen during ultrasoundexamination.17 Ultrasound imaging in the transverse plane ofthe digital nerve from the plantar aspect of the foot will reveala hypoechoic mass between the metatarsal heads (Figure 5).Because this appearance is nonspecific, it is important toimage the mass longitudinal to the peripheral nerve to iden-tify the affected nerve entering the neuroma. An associatedanechoic intermetatarsal bursa may also be seen. Pressurefrom the transducer and the opposing hand of the examineron the dorsum of the foot will produce symptoms typical ofa Morton’s neuroma, which is also a helpful diagnostic sign.

Nerve Trauma and Transection

Peripheral nerve trauma may be in the form of a crush injuryor transection. A crush injury may cause hypoechoic swellingof the affected nerve. In this situation, it is important to con-firm nerve fiber continuity through the affected area toexclude complete nerve transection. Peripheral nerve transec-tion may be secondary to a penetrating injury or surgery, ormay occur after a fracture of an adjacent bone. After a pene-trating injury, evaluation of the affected nerve at the site ofinjury can reveal hypoechoic swelling. In the setting of anerve transection, nerve fiber discontinuity will be identified.In addition, the nerve ends often retract away from eachother and hypoechoic swelling or neuroma formation can beseen.2 Neuroma formation after nerve transection is a normalresponse in attempted nerve healing (Figure 6).16 Direct pres-sure on the neuroma with the ultrasound transducer can elic-it symptoms. This feedback, not possible with MRI, is

helpful in determining which neuroma is causing symptoms,e.g., in patients after knee amputation (Figure 6).

Peripheral Nerve Sheath Tumors

Peripheral nerve sheath tumors consist of schwannomas (orneurilemmomas), neurofibromas, and their malignant coun-terparts termed malignant peripheral nerve sheath tumors.Using ultrasound examination, they typically appear speckledthroughout in a uniform heterogeneous manner, and may behypoechoic or isoechoic to muscle tissue (Figure 7). Posterioracoustic enhancement may occur similar to a fluid structure;however, increased flow on color or Doppler imaging willindicate the solid nature of the mass.20 Because most solidmasses are nonspecific with ultrasound imaging, it is impor-tant to identify the entering peripheral nerve to indicateprobable nerve origin. The involved peripheral nerve may beabnormally hypoechoic at the site of the nerve sheath tumor.

ULTRASOUND OF MUSCLE

Normal Ultrasound Appearance

Normal muscle appears relatively hypoechoic with interven-ing connective tissue and fascia appearing hyperechoic(Figure 1).9 Ultrasound examination shows muscle architec-ture converging toward a hyperechoic tendon. Tendonattachment on hyperechoic and shadowing bone can then beassessed when viewing the ultrasound image. It is importantwhen imaging muscle to first become oriented to the region-al anatomy. This can be accomplished using a lower frequency


Figure 4 Cubital tunnel syndrome. Ultrasound imagetransverse to the ulnar nerve proximal to the cubital tunnel showshypoechoic swelling (arrows) of the ulnar nerve. Note medialepicondyle of the humerus (H) and triceps (T).

Figure 5 Morton’s neuroma. Ultrasound image transverse tothe metatarsal shafts shows hypoechoic neuroma (arrows)between the metatarsal heads (M) (plantar aspect is at top ofimage).

transducer, often with a curved surface to allow a wider fieldof view. Once the regional anatomy is determined, accurateidentification of each muscle and muscle group can be com-pleted. It is also important to use a lower frequency transduc-er in this situation to ensure adequate depth penetrationwhen evaluating for pathology. Using only a higher frequen-cy transducer may cause one to overlook deeper pathology; itis suggested that evaluation of muscle begins from the bonesurface. Once the entire area is screened for pathology, thena higher frequency transducer can be used to optimally assessthe more superficial structures.

Muscle and Tendon Injury

Muscle and tendon injury may be categorized into acute andchronic. There are two common mechanisms of acute injury:direct blow/compression and indirect strain injury (fromexcessive contraction or stretch). Each of these mechanismsproduces pathology at a different location. The strain injurytypically involves the musculotendinous junction, and com-monly involves those muscle-tendon units that cross twojoints such as the hamstrings and medial gastrocnemius(Figure 8).22 In children, this stretch injury can cause boneavulsion at the tendon attachment. In contrast, the directblow or compression injury typically involves the musclebelly (Figure 9). Muscle injury can be divided into categoriesbased on severity: grade 1 is a mild strain, grade 2 (moderatestrain) is a partial tear, and grade 3 (severe strain) is a com-plete tear. Ultrasound of grade 1 injuries show abnormalhyperechoic hemorrhage and edema or may appear normal.23

With grades 2 and 3, there is often a mixed echogenic appear-ance from hemorrhage, showing both hypoechoic and hyper-

echoic areas (Figures 8 and 9).23 The key difference betweengrade 2 and 3 injuries is the presence of complete fiber dis-ruption and resulting total muscle or tendon retraction,which would indicate full-thickness or grade 3 injury.

Chronic injuries to muscle and tendon most commonly takethe form of tendinosis. This term, or tendinopathy, is pre-ferred over the term “tendinitis” as studies have demonstrat-ed no significant inflammatory cells in this condition.1,11,12

Common sites of tendinosis include the supraspinatus of therotator cuff, the common flexor and extensor tendons of theelbow (producing what is called epicondylitis), the patellartendon (also called jumper’s knee), and the Achilles’ tendon.


Figure 6 Transection neuroma of sciatic nerve after aboveknee amputation. Ultrasound image shows hypoechoictransection neuroma (arrows) in continuity with the sciatic nerve(arrowheads).

Figure 7 Schwannoma. Ultrasound image shows peripheralnerve (arrowheads) entering the heterogeneous well-definedschwannoma (arrows). Note posterior acoustic enhancement(open arrow).

Figure 8 Medial gastrocnemius tear. Ultrasound image showshypoechoic tear with fiber disruption involving the distal medialgastrocnemius (MG). Note intact plantaris tendon (arrowheads)(S = soleus muscle).

The ultrasound image of tendinosis shows hypoechoicswelling of the involved tendon (Figure 10). The presence ofanechoic areas would suggest coexisting partial-thickness tear,which can be common in the patellar and Achilles’ tendons.

Inflammatory Conditions

There are a number of acute inflammatory conditions thatcause myositis, which includes dermatomyositis, polymyosi-tis, and muscle infection. There is significant overlap in theultrasound appearances of these conditions because all showincreased echogenicity of the involved muscle and possiblehyperemia (Figure 11).18 In the setting of infection, abscesswill appear as a mixed echogenicity or anechoic area withposterior acoustic enhancement (Figure 12).13 The role ofultrasound in these conditions is not only to identify thelocation and extent of the abnormality, but also to guide per-cutaneous biopsy or aspiration for diagnosis. In chronic


Figure 10 Patellar tendinosis. Ultrasound image shows hypoechoic thickening of the proximal patellar tendon (arrows). Note normaldistal patellar tendon (arrowheads) (P = patella).

Figure 9 Triceps muscle, partial tear. Ultrasound image showspartial disruption of the triceps muscle (T) characterized byanechoic hemorrhage (arrows).

inflammatory conditions, a hyperechoic muscle of decreasedsize indicates atrophy.19

CONCLUSION

With ultrasound imaging, high-resolution transducers areable to show anatomy and pathology of peripheral nerves andmuscles. With regard to peripheral nerves, individual nervefascicles can be demonstrated. Ultrasound imaging is alsoeffective in the diagnosis of peripheral nerve entrapment dis-orders, nerve injury, and peripheral nerve sheath tumors.With regard to muscles, ultrasound imaging is able to imagemuscle groups from origin to insertion, demonstrating bothacute and chronic muscle and tendon abnormalities.Inflammatory conditions of muscle can also be identifiedwith ultrasound imaging, which is helpful in guiding diag-nostic biopsy. Inherent advantages of using ultrasound toimage peripheral nerve and muscle include portability, rela-tive low cost, improved accessibility, high resolution, and theability to image structures dynamically.

REFERENCES

1. Astrom M, Gentz CF, Nilsson P, Rausing A, Sjoberg S, Westlin N.Imaging in chronic achilles tendinopathy: a comparison of ultra-sonography, magnetic resonance imaging and surgical findings in 27histologically verified cases. Skeletal Radiol 1996;25:615-620.

2. Bodner G, Buchberger W, Schocke M, Bale R, Huber B, Harpf C,Gassner E, Jaschke W. Radial nerve palsy associated with humeralshaft fracture: evaluation with US--initial experience. Radiology2001;219:811-816.

3. Chiou HJ, Chou YH, Cheng SP, Hsu CC, Chan RC, Tiu CM, TengMM, Chang CY. Cubital tunnel syndrome: diagnosis by high-resolu-tion ultrasonography. J Ultrasound Med 1998;17:643-648.

4. Crass JR, van de Vegte GL, Harkavy LA. Tendon echogenicity: exvivo study. Radiology 1988;167:499-501.

5. Duncan I, Sullivan P, Lomas F. Sonography in the diagnosis of carpaltunnel syndrome. AJR Am J Roentgenol 1999;173:681-684.

6. Jacobson JA. Musculoskeletal sonography and MR imaging. A rolefor both imaging methods. Radiol Clin North Am 1999;37:713-735.

7. Jacobson JA, Jebson PJ, Jeffers AW, Fessell DP, Hayes CW. Ulnarnerve dislocation and snapping triceps syndrome: diagnosis withdynamic sonography--report of three cases. Radiology2001;220:601-605.

8. Jacobson JA, Lax MJ. Musculoskeletal sonography of the postopera-tive orthopedic patient. Semin Musculoskelet Radiol 2002;6:67-77.


Figure 12 Abscess. Ultrasound image shows predominatelyhypoechoic deltoid muscle abscess (arrows) with posterioracoustic enhancement (arrowheads).

Figure 11 Inflammatory myositis. Ultrasound imagetransverse to sartorius muscle shows abnormal hyperechoicthickening (arrows) (V = femoral artery/vein).

9. Jacobson JA, van Holsbeeck MT. Musculoskeletal ultrasonography.Orthop Clin North Am 1998;29:135-167.

10. Jamadar DA, Jacobson JA, Hayes CW. Sonographic evaluation of themedian nerve at the wrist. J Ultrasound Med 2001;20:1011-1014.

11. Khan KM, Bonar F, Desmond PM, Cook JL, Young DA, VisentiniPJ, Fehrmann MW, Kiss ZS, O’Brien PA, Harcourt PR, Dowling RJ,O’Sullivan RM, Crichton KJ, Tress BM, Wark JD. Patellar tendinosis(jumper’s knee): findings at histopathologic examination, US, andMR imaging. Victorian Institute of Sport Tendon Study Group.Radiology 1996;200:821-827.

12. Kjellin I, Ho CP, Cervilla V, Haghighi P, Kerr R, Vangness CT,Friedman RJ, Trudell D, Resnick D. Alterations in the supraspinatustendon at MR imaging: correlation with histopathologic findings incadavers. Radiology 1991;181:837-841.

13. Loyer EM, DuBrow RA, David CL, Coan JD, Eftekhari F. Imagingof superficial soft-tissue infections: sonographic findings in cases ofcellulitis and abscess. AJR Am J Roentgenol 1996;166:149-152.

14. Martinoli C, Bianchi S, Gandolfo N, Valle M, Simonetti S, DerchiLE. US of nerve entrapments in osteofibrous tunnels of the upperand lower limbs. Radiographics 2000;20:S199-S217.

15. Middleton WD, Payne WT, Teefey SA, Hildebolt CF, Rubin DA,Yamaguchi K. Sonography and MRI of the shoulder: comparison ofpatient satisfaction. AJR Am J Roentgenol 2004;183:1449-1452.

16. Provost N, Bonaldi VM, Sarazin L, Cho KH, Chhem RK.Amputation stump neuroma: ultrasound features. J Clin Ultrasound1997;25:858-859.

17. Quinn TJ, Jacobson JA, Craig JG, van Holsbeeck MT. Sonographyof Morton’s neuromas. AJR Am J Roentgenol 2000;174:1723-1728.

18. Reimers CD, Fleckenstein JL, Witt TN, Muller-Felber W, PongratzDE. Muscular ultrasound in idiopathic inflammatory myopathies ofadults. J Neurol Sci 1993;116(1):82-92.

19. Reimers K, Reimers CD, Wagner S, Paetzke I, Pongratz DE. Skeletalmuscle sonography: a correlative study of echogenicity and morphol-ogy. J Ultrasound Med 1993;12:73-77.

20. Reynolds DL Jr, Jacobson JA, Inampudi P, Jamadar DA, Ebrahim FS,Hayes CW. Sonographic characteristics of peripheral nerve sheathtumors. AJR Am J Roentgenol 2004;182:741-744.

21. Silvestri E, Martinoli C, Derchi LE, Bertolotto M, Chiaramondia M,Rosenberg I. Echotexture of peripheral nerves: correlation betweenUS and histologic findings and criteria to differentiate tendons.Radiology 1995;197:291-296.

22. Speer KP, Lohnes J, Garrett WE Jr. Radiographic imaging of musclestrain injury. Am J Sports Med 1993;21:89-96.

23. Takebayashi S, Takasawa H, Banzai Y, Miki H, Sasaki R, Itoh Y,Matsubara S. Sonographic findings in muscle strain injury: clinicaland MR imaging correlation. J Ultrasound Med 1995;14:899-905.

24. Thain LM, Downey DB. Sonography of peripheral nerves: tech-nique, anatomy, and pathology. Ultrasound Q 2002;18:225-245.


INTRODUCTION

Information obtained from electrodiagnostic (EDX) studiesis often helpful in assisting with the diagnosis, following thecourse, or indicating the responsiveness to treatment of dis-eases affecting the neuromuscular system. In conjunctionwith nerve conduction studies, analysis of motor unit actionpotentials (MUAPs) or spontaneous activity detected usingsingle fiber, concentric, or monopolar needle electrodes com-prises an important component of the EDX assessment ofmany disorders affecting the motor system. Needle elec-tromyography (EMG) analysis may, at the least, provide anindication of motor axon or motor neuron injury, may showevidence of early or chronic motor unit (MU) reinnnervationor muscle fiber injury, and may provide evidence of the lossof MUs based upon recruitment. Although typically per-formed in a qualitative manner, quantitative analysis ofMUAP size, morphology, and recruitment may be helpful insome cases to better characterize or rule out a given disorder,provide baseline values against which to compare future stud-ies, and judge the usefulness of any intervention.32

Furthermore, estimation of the numbers of MUs in a muscle

group can provide a direct assessment of the number of func-tioning MUs to assess for possible MU or motor axon loss.

In the mid 1950s, Buchthal and colleagues8,9 introduced amethod for quantitative MUAP analysis. Twenty or moreMUAPs meeting strict amplitude and rise-time criteria werecollected, MU by MU, from a minimally contracting muscle.Statistics (mean values), based on the detailed measurementof the duration, amplitude, and number of phases from thefilmed recordings of these MUAPs were compared againstnormative data. While these principles are generally accept-ed, this method is extremely time-consuming with a com-plete study often requiring hours. Therefore, based largely onthe practical disadvantages related to the duration of thestudy and the need for considerable patient cooperation this“manual” method is impractical in a typical clinical laborato-ry.

Manual, operator-dependent methods of collecting MUAPsfor quantitative analysis rely on the EDX physician’s experi-ence and “internal pattern recognition algorithm.” Manualmethods have the advantage of visually considering the

25

Decomposition-based Quantitative EMG:From Theory to Application

Timothy J. Doherty, MD, PhD, FRCPC

Assistant ProfessorDepartments of Clinical Neurological Sciences and Rehabilitation Medicine

The University of Western OntarioLondon, Ontario, Canada


Associate ProfessorDepartment of Systems Design Engineering

University of WaterlooWaterloo, Ontario, Canada

shapes and firing patterns of the MUAPs as opposed to justtheir peak amplitudes. The shapes of the MUAPs of otheractive MUs can be considered and visually compared to thepotential of interest. This ensures that the MUAPs collectedand analyzed are produced by the same single MU andexcludes inclusion of potentials from other MUs. Of course,the major drawback of manual methods is the impracticalamount of operator time and energy required to collect a suf-ficient number of MUAPs in order to provide meaningfuldata. Few EDX physicians are able to visually track morethan three to four MUAPs simultaneously in an interferencepattern. Additionally, window- or level-based triggering sys-tems are usually overwhelmed by interference patterns withmore than three to four active MUs. This being the case, thesample is typically drawn from only the earliest recruited,lowest threshold MUs. Therefore, automated methods havebeen developed that attempt to mimic this manual, operator-dependent MUAP selection process.

Following the introduction of signal triggering and delaylines in clinical EMG systems, successive discharges of theMUAP could be time-locked on the display. This alloweddirect measurement of MUAP parameters directly from thedisplay or from printouts. The use of spike-triggered averag-

ing further improved the signal-to-noise ratio and facilitatedanalysis of quantitative features of the MUAP. Computerswere later introduced into the clinical laboratory and aidedboth measurement and averaging. However, despite theadded speed of computer-assisted collection and analysis, astudy of 20 MUAPs from one muscle in an alert and cooper-ative patient still required at least 20 minutes, considered tooslow for routine work.

The advent of faster, more powerful computers enabled thedevelopment of EMG decomposition methods.33 The goal ofEMG decomposition is to extract as much information aspossible from each muscle contraction and associated epochof EMG data. In essence, decomposition methods breakdown the EMG signal interference pattern into its con-stituent MUAP trains (MUAPTs) (Figure 1). From theseMUAPTs, information relating to MUAP morphology andstability and MU activation patterns may be determined.

The goals of decomposition methods range from simpleextraction of the MUAPs of a single MU, to completedecomposition of a composite EMG signal into its con-stituent MUAPTs.2,18,21,24,27,31,33 As with manual techniques,automated techniques consider the shapes of the MUAPs,

26 Decomposition-based Quantitative EMG: From Theory to Application AANEM Course

Figure 1 Decomposition EMG is designed to break down a composite EMG signal into its constituent motor unit action potential trains.

but in addition they are faster, require much less operatorintervention, are potentially less prone to bias, and requireless patient cooperation.

The ideal EMG signal decomposition system should be ableto handle signals with the following attributes:

1. five or more MUAPTs

2. nonstationary MUAPs and background biologicalnoise

3. variable MUAP shapes

4. nonstationary MU activation patterns

5. superpositions of MUAPs

6. two or more MUs with similar MUAP shapes

7. lapses in the recruitment of particular MUs

The basic steps in EMG signal decomposition are:

1. signal acquisition

2. segmentation (detecting MUAPs)

3. feature extraction

4. clustering of detected MUAPs

5. supervised classification of detected MUAPs

DECOMPOSITION-BASED QUANTITATIVEELECTROMYOGRAPHY

Decomposition-based quantitative EMG (DQEMG)14,33 is aseries of algorithms that have been developed to provide clin-ically relevant quantitative electrophysiologic information.Like automatic decomposition EMG (ADEMG)23,24 andmulti-MUP2 analysis, DQEMG is based on the partialdecomposition of concentric or monopolar needle-detectedEMG signals and MUAP superpositions that are notresolved. Decomposition-based quantitative EMG is basedon a multi-algorithm, multi-stage decomposition processthat uses robust data-driven assignment thresholds andactively uses firing-pattern information to make classificationdecisions. Further details of the steps in this process, most ofwhich are common to all decomposition systems, will beexplained.

Basic Assumptions

Signal Acquisition

To obtain detailed, or micro, spatial, and temporal informa-tion about the fibers of each MU, signals need to be acquiredwith electrodes that have small, selective detection surfacessuch as monopolar (MN), concentric (CN), single-fiber (SF),or fine-wire electrodes. In contrast, to obtain informationregarding the overall size and muscle fiber spatial distributionof each MU, or macro information, signals must be acquiredwith electrodes that have large, less-selective surfaces such assurface electrodes or indwelling macro or concentric macro(conmac) electrodes.19,29,30 Therefore, it is useful to acquiresignals from two instrumentation channels during a givenvoluntary contraction. One channel typically acquires amicro signal detected by a standard MN, CN, or SF electrodewith an appropriate bandpass and sampling rate. A secondchannel acquires a macro signal detected by an overlying sur-face or indwelling macro or conmac electrode with an appro-priate bandpass and sampling rate. The micro signal isdecomposed into its MUAPTs and the macro signal is thenanalyzed in conjunction with the results of the micro signaldecomposition. Additional signals, such as force, could alsobe acquired.

Segmentation of the Composite Micro-electromyographySignal

The first step in the decomposition of an acquired micro-EMG signal is to attempt to detect all of the MUAPs gener-ated by MUs active during signal acquisition. However, inpractice, the MUAPs of MUs without fibers close to thedetection surface will be of low amplitude and will be com-posed of primarily low-frequency components. Therefore, itis difficult to consistently assign such MUAPs to their correctMUAPT and it is easy to miss occurrences of such MUAPswhen they occur in close temporal proximity to largerMUAPs. Consequently, it is necessary to select and consis-tently detect only those MUAPs that can be subsequentlysuccessfully assigned. Several different methods have beendevised to segment the signal into sections that contain sig-nificant MUAPs. All of these methods are based on definingsome detection threshold, which is based on a statistic com-puted using the composite signal. When the signal character-istics produce a statistic value above the threshold value, a sig-nificant signal segment (putative MUAP) is detected that canbe processed and further analyzed.

One consistent method of selecting segments containingMUAPs that can be consistently correctly assigned is to firstfilter or pre-process the micro-EMG signal. Bandpass filter-


ing shortens the duration of the MUAPs, which reduces theirtemporal overlap and reduces the number of superimposedwaveforms. It also reduces the amplitude of similarly shapedMUAPs of different MUs that do not have fibers close to theelectrode (“distant” MUAPs). This filtering removes nondis-criminative, low-frequency information and allows for dis-crimination between MUAPs of different MUs by reducingthe variability of the MUAP shapes from a given MU.

Most preprocessing and threshold determination schemesinvolve using the slope of the composite signal in some way.Nandedkar and colleagues27 use 10% of the maximum low-pass filtered first derivative of the composite signal as athreshold. Stålberg and colleagues31 use an absolute thresholdcriteria consisting of an amplitude of greater than 50 µV anda slope of greater then 0.3 V/s. All of these methods directlyor indirectly use the first derivative of the signal and a detec-tion threshold in an attempt to only detect significantMUAPs of MUs with at least some of their fibers close to thedetection electrode. These MUAPs would typically corre-spond with those considered “close” or “sharp” when per-forming routine EMG analysis.

In similar fashion, DQEMG processes a filtered signal anduses an amplitude threshold (e.g., 50 µV) and slope criteria(≥ 0.3 V/s) to detect fixed-length sections containing candi-date MUAP occurrences. The detected MUAPs are then rep-resented by the data samples of the filtered micro signal.

Clustering of Detected Motor Unit Action Potentials

Following the detection stage, the MUAPs must be groupedor clustered into groups of responses of similar shape. Eachcluster has a prototypical shape that is called the cluster meanor MUAP template. The exact number of clusters—that is,the number of active MUs with fibers sufficiently close to thedetection surface of the electrode—is unknown a priori.Within the context of EMG signal decomposition, a cluster-ing algorithm is usually applied to only a portion of thedetected MUAPs. It has two main objectives. One is to esti-mate the number of MUs contributing significant MUAPs tothe composite EMG signal (i.e., the number of possibleMUAPTs). The other is to assign as many MUAPs as possi-ble to their correct MUAPT so that the prototypical, or meanMUAP shape, of each contributing MU can be optimallydetermined. However, every MUAP assigned to the wrongMUAPT increases the probability of errors and thus, subse-quent supervised classification results will be more successfulif based on accurate clustering results. Therefore, it is moreimportant that the clustering results are accurate and that

superimposed MUAPs are ignored than it is for all detectedMUAPs to be assigned (especially for clinical applicationswhere MUAP morphology is of more importance than accu-rate determination of MU activation patterns). As such, aclustering algorithm does not need to classify all detectedMUAPs. It should be conservative and try to minimize thenumber of erroneous classifications. In addition, because atlater stages of the decomposition it is easier to merge twotrains into one than it is to split one erroneous train into two,overestimation of the number of MUAPTs is preferred tounderestimation. The pattern recognition algorithmsemployed at this stage employ feature extraction to representthe shape of detected MUAPs and assign them to groups.These features mathematically represent the shape of theMUAPs and are typically based on the actual sampled voltagevalues or morphologic statistics such as peak-to-peak voltage,number of phases, duration, and slope. As well, some algo-rithms are dependent on information in the frequencydomain such as those obtained with Fourier transformationcoefficients and wavelet analysis.

Supervised Classification of Detected Motor Unit ActionPotentials

Recall that the goal of clustering is to conservatively deter-mine the number of active MUs and provide initial classifica-tion. Once estimates of the number of active MUs and theshapes of their prototypical MUAPs are available, the com-plete set of detected MUAPs can be further classified usingsupervised classification techniques. For supervised classifica-tion, the MUAPs are represented as they are for clustering.

For clinically applicable decomposition methods where thegoal is usually the determination of quantitative data relatingto MUAP amplitude, duration, and morphology, the super-vised classification algorithms are typically conservative withthe emphasis on accuracy regarding assignment to a classrather than completeness of the decomposition. If, for clini-cal or investigative purposes, the goal is the accurate determi-nation of recruitment and firing time data, then less conser-vatism is appropriate and resolution of superpositions is nec-essary.

Resolving Superimposed Motor Unit Action Potentials

During muscle contraction, recruited MUs fire asynchro-nously and at variable firing rates. When two or more MUsdischarge at the same time, or in close temporal relationshipto one another, the detected potential is the algebraic summa-tion of the individual MUAPs. This response is termed a


superimposed MUAP. If complete decomposition is one ofthe objectives of a given method, then superpositions need tobe resolved into their component MUAPs.

Discovering Temporal Relationships Between Motor UnitAction Potential Trains

Following clustering, supervised classification, and superposi-tion resolution, a MUAPT is produced. If each discharge ofan MU is independent of previous and future firings andidentically distributed, its MUAPT can be modeled as arenewal stochastic point process. Such modeling representsthe discharge times or the intervals between discharge timesof MUAPTs as random variables. The random variables rep-resenting the activity of two MUAPTs can be either inde-pendent or dependent. If they are independent, the firingtimes of one train have no affect on and are not related to thefiring times of the other train. On the other hand, if thetrains are dependent the firing times of one train are affectedor related to the firing times of the other train. Three kindsof dependent behavior defined as linked, exclusive, or synchro-nized are possible. Linked trains fire with a definite, constantinterval relative to each other, such as a main MUAP spikeand its satellite. Exclusive trains never fire together. If twotrains are exclusive they may have been created by the sameMU and erroneously separated into two trains by the classi-fication algorithms. Synchronized trains have behavior some-where between linked and exclusive trains and probably rep-resent a true biological tendency of MUs to have dependentfiring behavior. This information can be used to detect erro-neous, multiply detected trains and resolve such errors.33

HOW CAN DECOMPOSITION RESULTS BE USED?

The resulting EDX data from the decomposition of an EMGsignal can be used to assist in the diagnosis, treatment, andfurther study of neuromuscular disorders. In addition, infor-mation pertaining to the motor control of skeletal musclesand the effects of disease, aging, and fatigue can be obtained.Following are descriptions of some of these specific uses.

QUANTITATIVE NEEDLE (MICRO) ELECTROMYOGRAPHY

In a normal muscle the fibers of each MU are randomly dis-tributed throughout a roughly circular area of approximately5-10 mm in diameter called the MU territory. The morpho-logical makeup of the MUs of a muscle changes with disease.With myopathic processes, the numbers of fibers in MUs areoften reduced, as are the sizes of the fibers. With neuropath-ic disorders, the number of MUs is reduced but the sizes of

the remaining MUs are typically increased, at least in thechronic phase. This means that the MUs have larger numbersof fibers and that the fibers are no longer randomly spatiallydistributed. These morphological changes in the structure ofMUs result in changes in the shapes of their detectedMUAPs. Characterizing these MUAP changes can be used toaid in the diagnosis and treatment of some neuromusculardisorders. Various statistics of the MUAP shapes, calculatedfrom a representative sample of MUAPs detected in a muscleof interest, are used to quantify their characteristics.

As described earlier, decomposition methods provideMUAPTs that in turn represent individual firings from agroup of concurrently active MUs (Figure 2). From eachMUAPT the prototypical MUAP shape can be estimated.Decomposition-based quantitative EMG uses isolatedMUAPs and mode estimation over a 50 ms interval centeredon each MUAP’s peak slope, to calculate the prototypicalMUAP. An isolated MUAP is defined as one that occurs iso-lated in time by at least 3 ms from any other detected MUAPand with a shape suitably close statistically to that of the pro-totypical MUAP. Isolated MUAPs are used to better estimatethe spike portion of the prototypical MUAP. Given the pro-totypical shape, its duration, peak-to-peak voltage, numberof phases and turns, area, and area-to-amplitude ratio are cal-culated using standard algorithms. When using DQEMG, a25 µV threshold is used to define a turn and a phase musthave an amplitude of at least 20 µV and duration of at least240 ms. The values of MUAP parameters along with the pro-totypical MUAP waveform for each MUAPT extracted froma composite EMG signal are presented in a clinical summary(Figure 3). Markers identifying the onset, end, and peak val-ues can be manually adjusted if necessary. These parametersfrom the sample of MUAPs collected from a given musclestudied are then summarized statistically and graphically.

MACRO MOTOR UNIT ACTION POTENTIALS

Stålberg first introduced the concept of ensemble or spiketriggered averaging, using individual MU firing occurrencesas triggers, to extract macro-MUAPs.29,30 His original macroelectrode was a 15 mm length of cannula centered around asingle-fiber detection surface. Decomposition EMG andspecifically DQEMG simply extends Stålberg’s original idea.The firing times of multiple MUs are represented in theMUAPTs obtained following the decomposition of a microsignal. For each MUAPT the MU firing times are used astriggers for locating 100 ms epochs in the macro detected sig-nal (detected using surface or indwelling needle electrodes).Each located interval is ensemble averaged to extract theMU’s macro-MUAP. The duration, peak-to-peak voltage,


and negative peak area of the macro MUAP are calculatedand can be displayed along with the macro-MUAP waveformin a clinical summary (Figure 4).

The size parameters of the macro-MUAPs, such as peak-to-peak voltage or area are related to the overall size of the con-tributing MU. This is certainly true when using indwellingmacro electrodes such as Stålberg’s original “macro” electrodeor a conmac electrode. When using surface electrodes, thedepth of the MU in the muscle can confound these relation-ships. Additionally, amplitude or area of a mean surface-detected MUAP can be determined and compared with thatof the compound muscle action potential in order to derive aMUNE in suitable muscles (e.g., thenar, hypothenar, bicepsbrachii, extensor digitorum brevis, tibialis anterior).

Firing Rates and Recruitment Patterns

Potential limitations of the specific decomposition algorithmmust be taken into account when considering firing timedata. If the decomposition algorithm resolved superimposedMUAPs and/or a manual editing was completed, the firingpattern information may be assumed to be complete and cor-rect and can therefore be used for detailed analysis. If this isnot the case, information regarding only the average behaviorof the MU is available. While average information might besuitable for clinical use it is more likely that detailed informa-tion will be required for investigative purposes. The contrac-tile force at the time of signal acquisition is another impor-tant consideration when interpreting firing pattern informa-tion. Were MUs being recruited or becoming inactive? Were


Figure 2 A single motor unit action potential (MUAP), extracted by decomposition-based quantitative electromyography (DQEMG), isshown centered in the raster plot along with two other identifiable MUAPs. The data along the left are the time occurrences (in seconds) forthe extracted MUAP within the contraction, while the data along the right are inter-discharge intervals (in milliseconds) between eachidentified occurrence of the MUAP.

firing rates changing to create a varying force profile? If theforce of a contraction is also measured more detailed infor-mation can be obtained.

With DQEMG, for each MUAPT the temporal pattern of itsdischarges are characterized by a histogram and estimation ofthe mean interdischarge-interval (IDI) and the standard devi-ation (STD) and coefficient of variation of the IDIs (Figure4). Because there are erroneous and missed MU firings pres-ent in the trains, a technique that recognizes and ignoreserroneous IDIs is used. An MU’s average firing rate is calcu-lated as simply the inverse of its mean IDI. An MU’s instan-taneous firing rate, at each of its discharge times, is estimat-ed as the inverse of a Hamming weighted average of 10 validIDIs centered around the current discharge time. A valid IDIhas a value within 3 STDs of the mean IDI. The identifica-tion rate is calculated as the percentage of expected firingsdetected. The number of expected firings is estimated as theproduct of the mean firing rate times the length of time thatthe MU was active. The times of an MU’s discharges can be

displayed along with its instantaneous firing rate, a histogramof its IDIs, and calculated firing pattern statistics in a decom-position summary (Figure 4).

APPLICATION OF DECOMPOSITION ELECTROMYOGRAPHY

Normative Data

It is inappropriate to compare data obtained with manualmethods with that obtained using decomposition EMG-based methods. Similarly, while many of the previouslydescribed decomposition techniques share certain features,the differences in the algorithms require the collection of spe-cific normative data. Therefore, normative data has been col-lected and published for a number of clinically applicablemethods.2,14,24,27,27 Interestingly, despite significant differ-ences in the algorithms, there is reasonable similarity whencomparing, for example, quantitative data for the bicepsbrachii muscle (Table 1).


Figure 3 Clinical summary from decomposition-based quantitative electromyography illustrating extracted motor unit action potential(MUAP) (micro template), macro (S-MUAP) and motor unit (MU) activation pattern data for a single contraction from which six MUs wereextracted.

Clinical Application of Decomposition Analysis

There are relatively few reports describing the clinical appli-cation of decomposition EMG. Dorfman and colleagues17

used ADEMG to compare MU firing rates and firing ratevariability in patients with neuropathic, myopathic, or cen-tral disorders. Based on the results of decomposition analysis,they reported increased MUAP amplitudes and firing rates,and firing rate variability in patients with motor neuron dis-eases. In contrast, patients with myopathies (primarilyinflammatory) were found to have MUAPs with reducedamplitude, duration, and turns. Firing rates, especially atthreshold force, were often dramatically increased. Forpatients with central disorders, the MUAP morphology wassimilar to control subjects. However, mean firing rates werereduced and firing rate variability was increased.

Jongen and coworkers20 compared ADEMG with conven-tional EMG in 17 patients with idiopathic inflammatorymyopathy. Automatic decomposition EMG was carried outon only the biceps brachii (due to the availability of norma-tive data) whereas the conventional qualitative analysis wasperformed on a mean of four muscles per subject. Theauthors reported less sensitivity with ADEMG as comparedto the conventional analysis (9/17 versus 13/17 consideredprobable for myopathy respectively). By way of explanation,the authors suggest that ADEMG may have missed somesmall MUAPs, especially at higher contractile forces, and thatthe automated duration measurement may not always be per-formed correctly. On the other hand, it may be inappropriateto compare the results of ADEMG from a single muscle withthose from qualitative analysis of multiple muscles when per-formed by a nonblinded EDX physician.

Barkhaus and colleagues1 studied predominately the bicepsbrachii of 17 patients with a diagnosis of inclusion bodymyositis (IBM) with multi-MUAP analysis. Traditionally, ithas been held that IBM exhibits both myopathic and neuro-pathic features on needle examination. Quantitative data,however, were myopathic (decreased mean duration orarea:amplitude ratio) in 16 of the 17 patients studied. Macro-EMG amplitude was reduced in 3 of 17 studies with theremainder in the normal range. No subject’s mean macro-EMG values were increased as would be expected with a neu-ropathic process.

Application of Decomposition-based QuantitativeElectromyography

The authors have reported control data14 for both concentricneedle detected MUAP data and surface electrode detectedMUAP (S-MUAP) data from control subjects in the deltoid,biceps, first dorsal intersosseous, vastus medialis, and tibialisanterior muscles. There was significant variability across thesefive muscles with regard to the mean MUAP size, mean S-MUAP size, and MUAP duration. These values were in keep-


Figure 4 Micro (motor unit action potential [MUAP], indicatedby markers) and macro (S-MUAP) data for a single motor unit(ver tical scale, 200 µV/division horizontal scale, 10ms/division).

Table 1 Mean values ± standard deviation for biceps brachii concentric needle motor unit potential data from decomposition analysisacross four different methods

Amplitude (µV) Duration Phases Firing Rate

Doherty and Stashuk (n = 290) 327 ± 187 0.8 ± 1.5 2.5 ± 0.2 12.3 ± 1.3

McGill and Dorfman (1985) 468 ± 292 9.2 ± 3.9 1.8 ± 1.0 12.2 ± 1.8

Bischoff and colleagues (1994) 436 ± 115 9.9 ± 1.4 2.6 ± 0.3 NA

Nandedkar and colleagues (1995) 364 ± 296 10.6 ± 5.0 2.1 ± 1.0 10.6 ± 2.5

ing with other published data. This study simply confirmedthat DQEMG was able to validly extract needle and surface-detected signals from a number of muscles across contrac-tions of varying intensity in healthy control subjects.

Decomposition-based Quantitative Electromyography-assisted Motor Unit Number Estimates

The authors have recently applied DQEMG to obtain motorunit number estimates (MUNEs) to study both aging anddisease. A number of methods have been developed to obtainMUNEs, all of which are based on the same principle. Thatis, the size (amplitude or area) of the maximum compoundmuscle action potential from a muscle or muscle group inresponse to supramaximal stimulation of its motor nerve isdivided by the mean size of a representative sample of S-MUAPs.10,11,28 One method of obtaining a MUNE, spike-triggered averaging, uses the needle-detected MUAP as a trig-ger to ensemble average the surface EMG single to extract S-MUAPs from which a mean S-MUAP size can be derived7

(Figure 5). Decomposition-based quantitative EMG enablesthe collection and analysis of multiple MUs from each con-traction and thus allows for more representative sampling ofthe S-MUAP population and considerably shortens the col-lection time. It has been recently demonstrated that decom-position provides a valid and reliable method forMUNE.3,4,25

Application of Decomposition-based QuantitativeElectromyography to Normal Aging

It has been well established that aging is associated with sub-stantial changes in the neuromuscular system including sig-nificant losses of MUs. The authors and others have previous-ly demonstrated MU losses on the order of 50% for intrinsichand muscles and the biceps brachii/brachialis.6,12,13,15,16

Decomposition-based quantitative EMG-assisted spike-trig-gered averaging was applied to examine the impact of agingon MU numbers in the tibialis anterior of old and very oldmen. The hypothesis was that further MU loss, beyond thatpreviously noted in those over 60 years, might explain theprogressive losses of strength and muscle mass present in thevery old. The authors compared strength measures andMUNEs in a group of young (mean age 27±3 yrs), old (66±3yrs), and very old (82±4 yrs) men. Motor unit number esti-mates were significantly reduced in the older men (90) com-pared to the young (150), and further losses were observed inthe very old (59). However, despite the significant reductionin MUNE after 65 years, strength was maintained and wasnot significantly reduced until after age 80.26

Application of Decomposition-based QuantitativeElectromyography to the Study of Neuromuscular Disease

The authors have completed a pilot project that usedDQEMG to examine S-MUAP sizes, and MUNE for boththe hypothenar muscles and biceps brachii/brachilais in asmall group of patients with amyotrophic lateral sclerosis(ALS).5 As expected the mean S-MUAP size was significant-ly increased in the patients with ALS in comparison to con-trol subjects for both muscle groups. In turn, the MUNEswere substantially reduced in the ALS patients (e.g.,hypothenar: controls 215, ALS 73). From a practical stand-point, it was possible to obtain standard quantitative MUAPdata on 20 or more MUs and MUNEs in this study in about15 minutes for each muscle group.

Recently, DQEMG was applied as a method to study MUnumbers in hereditary neuropathy. The authors examined a


Figure 5 Illustration of motor unit number estimation fromdecomposition-based quantitative electromyography in thebiceps/brachialis group in a young control subject.

MUNE = motor unit number estimate; SMUAP = surface electrode detected motorunit action potential

Results are from 20 valid

large cohort of patients with Charcot-Marie-Tooth (CMT)-X, and obtained DQEMG assisted MUNEs in thehypothenar and biceps/brachialis muscle groups. It was theauthors’ hypothesis that CMT-X may present with greaterMU losses in the more distal muscles, in comparison toCMT1A. This was supported by preliminary data thatshowed greater MU losses observed in the hand muscles(mean 27 CMT-X , control lower limit of normal > 100)than the more proximal biceps/brachialis (mean 181, con-trols LLN > 150). The ratio of the distal to proximal MUNEswas 0.15. This ratio was less than previously observed for agroup of CMT1A patients, suggesting that CMT-X mayexhibit an accentuated distal MU/motor axon loss in com-parison to CMT1A.22

SUMMARY

Decomposition EMG techniques are designed to extractquantitative data from the EMG signal through a series ofcomputer algorithms. Decomposition EMG, as well as othersimilar methods, uses pattern recognition to recognize specif-ic features including shape and the firing pattern of theMUAPs of voluntarily activated MUs to do so. Quantitativedata relating to MUAP size, duration, and morphology canbe obtained as well as recruitment and firing pattern data. AMUNE method utilizing features of DQEMG has also beendeveloped. Decomposition EMG and other decompositionmethods have been used to examine neuromuscular disordersand human aging.

REFERENCE LIST

1. Barkhaus PE, Periquet MI, Nandedkar SD. Quantitative electrophys-iologic studies in sporadic inclusion body myositis. Muscle Nerve1999;22:480-487.

2. Bischoff C, Stålberg E, Falck B, Eeg-Olofsson KE. Reference valuesof motor unit action potentials obtained with multi- MUAP analysis.Muscle Nerve 1994;17:842-851.

3. Boe SG, Stashuk DW, Brown WF, Doherty TJ. Decomposition-based quantitative electromyography: effect of force on motor unitpotentials and motor unit number estimates. Muscle Nerve2005;31:365-373.

4. Boe SG, Stashuk DW, Doherty TJ. Motor unit number estimationby decomposition-enhanced spike-triggered averaging: control data,test-retest reliability, and contractile level effects. Muscle Nerve2004;29:693-699.

5. Bromberg MB, Doherty TJ, Stashuk DW, Davis SE. Assessing motorneuron loss and collateral reinnervation in ALS using decompositionenhanced EMG. Presented at: The International Symposium onALS/MND; November 16-18, 2003; Milan, Italy.

6. Brown WF. A method for estimating the number of motor units inthenar muscles and the changes in motor unit count with ageing. JNeurol Neurosurg Psychiatry 1972;35:845-852.

7. Brown WF, Strong MJ, Snow R. Methods for estimating numbers ofmotor units in biceps-brachialis muscles and losses of motor unitswith aging. Muscle Nerve 1988;11:423-432.

8. Buchthal F, Pinelli P, Rosenfalck P. Action potential parameters innormal human muscle and their physiologic determinants. ActaPhysiol Scand 1954;32:219-229.

9. Buchthal F, Rosenfalck P. Action potential parameters in differenthuman muscles. Acta Physiologica Scandinavica 1955;30:125-131.

10. Doherty T, Simmons Z, O’Connell B, Felice KJ, Conwit R, ChanKM, Komori T, Brown T, Stashuk DW, Brown WF. Methods for esti-mating the numbers of motor units in human muscles. J ClinNeurophysiol 1995;12:565-584.

11. Doherty TJ, Brown WF. A method for the longitudinal study ofhuman thenar motor units. Muscle Nerve 1994;17:1029-1036.

12. Doherty TJ, Brown WF. Motor unit number estimation: methodsand applications. In: Brown WF, Bolton CF, Aminoff MJ, editors.Neuromuscular function and disease: basic, clinical, and electrodiag-nostic aspects. Philadelphia: WB Saunders; 2002. p 274-290.

13. Doherty TJ, Brown WF. The estimated numbers and relative sizes ofthenar motor units as selected by multiple point stimulation in youngand older adults. Muscle Nerve 1993;16:355-366.

14. Doherty TJ, Stashuk DW. Decomposition-based quantitative elec-tromyography: methods and initial normative data in five muscles.Muscle Nerve 2003;28:204-211.

15. Doherty TJ, Vandervoort AA, Brown WF. Effects of ageing on themotor unit: a brief review. Can J Appl Physiol 1993;18:331-358.

16. Doherty TJ, Vandervoort AA, Taylor AW, Brown WF. Effects ofmotor unit losses on strength in older men and women. J ApplPhysiol 1993;74:868-874.

17. Dorfman LJ, Howard JE, McGill KC. Motor unit firing rates and fir-ing rate variability in the detection of neuromuscular disorders.Electroencephalogr Clin Neurophysiol 1989;73:215-224.

18. Dorfman LJ, McGill KC. AAEE Minimonograph #29: Automaticquantitative electromyography. Muscle Nerve 1988;11:804-818.

19. Jabre JF. Concentric macro electromyography. Muscle Nerve1991;14:820-825.

20. Jongen PJ, Vingerhoets HM, Roeleveld K, Stegeman DF. Automaticdecomposition electromyography in idiopathic inflammatorymyopathies. J Neurol 1996;243:79-85.


21. LeFever RS, De Luca CJ. A procedure for decomposing the myoelec-tric signal into its constsituent action potentials - Part I: Techniquetheory and implementation. IEEE Trans Biomed Eng 1982;29:149-157.

22. Lewis RA, Li J, Fuerst DR, Shy ME, Krajewski K. Motor unit num-ber estimate of distal and proximal muscles in Charcot-Marie-Toothdisease. Muscle Nerve 2003;28:161-167.

23. McGill KC, Cummins KL, Dorfman LJ. Automatic decompositionof the clinical electromyogram. IEEE Trans Biomed Eng1985;32:470-477.

24. McGill KC, Dorfman LJ. Automatic decomposition electromyogra-phy (ADEMG): validation and normative data in brachial biceps.Electroencephalogr Clin Neurophysiol 1985;61:453-461.

25. McNeil CJ, Doherty TJ, Stashuk DW, Rice CL. The effect of con-traction intensity on motor unit number estimates of the tibialis ante-rior. Clin Neurophysiol 2005;116:1342-1347.

26. McNeil CJ, Doherty TJ, Stashuk DW, Rice CL. Motor unit numberestimates in the tibialis anterior muscle of young, old, and very oldmen. Muscle Nerve 2005;31:461-467.

27. Nandedkar SD, Barkhaus PE, Charles A. Multi-motor unit actionpotential analysis (MMA). Muscle Nerve 1995;18:1155-1166.

28. Shefner JM. Motor unit number estimation in human neurologicaldiseases and animal models. Clin Neurophysiol 2001;112:955-964.

29. Stålberg E. Macro EMG, a new recording technique. J NeurolNeurosurg Psychiatry 1980;43:475-482.

30. Stålberg E. Macro EMG. Muscle Nerve 1983;6:619-630.31. Stålberg E, Falck B, Sonoo M, Stålberg S, Astrom M. Multi-MUAP

EMG analysis--a two year experience in daily clinical work.Electroencephalogr Clin Neurophysiol 1995;97:145-154.

32. Stålberg E, Nandedkar SD, Sanders DB, Falck B. Quantitative motorunit potential analysis. J Clin Neurophysiol 1996;13:401-422.

33. Stashuk DW. Decomposition and quantitative analysis of clinicalelectromyographic signals. Med Eng Phys 1999;21:389-404.


��421 First Avenue SW, Suite 300 East

Rochester, MN 55902(507) 288-0100 / Fax: (507) 288-1225

[email protected]

��

Date post:	15-Oct-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

New Matthew C.Kiernan,MBBS,PhD,FRACP Jon A.Jacobson,MD … · 2008. 9. 10. · American Association...

Documents