Combined use of surface electromyography and 31P-NMR spectroscopy for the study of muscle disorders.

Surface electromyography (EMG) and phosphorous magnetic resonance ([sup.31]P-NMR) spectroscopy are powerful tools for assessing the electrical and chemical components of muscle fatigue, respectively. The techniques are unique in providing noninvasive access to muscle electrophysiology and cell energetics in situ. Only recently have the technical barriers to combining these techniques been overcome.[1]

Muscles transduce chemical fuels into mechanical tension. The propagation of the myoelectrical signal along the muscle membrane plays a key role in precipitating a chain of events that result in the conversion of chemical energy into the mechanical energy of muscle contraction. The ability to sustain this process over time is counteracted by the presence of muscle fatigue. Although no universal definition of fatigue exists, broadly defined, fatigue can be considered as the collection of time-dependent processes that result in the progressive impairment of the force-generating capacity of muscles during activity. Of the numerous physiological events and energy transduction processes considered as fatigue factors, muscle excitation and metabolite accumulation are two of the most common processes linked to impairment of muscle activation and force generation.[2,3] Muscle excitation is reduced during fatigue by the slowing of the myoelectric signal across the muscle membrane.[4] The accumulation of metabolites such as [H.sup.+] and [K.sup.+] ions have been directly implicated in reducing the ability of muscles to sustain a force.[3] These peripheral mechanisms of fatigue (ie, mechanisms peripheral to the neuromuscular junction) can be distinguished by surface EMG and [sup.31]P-NMR spectroscopy. The ability of these procedures to provide a means of evaluating muscle performance is further enhanced when they are used together.

This article provides an introductory description of these two methods and briefly discusses the technical considerations of combining them for human use. A recent example is provided in which the techniques were combined to study the underlying mechanisms of fatigue in patients with fibromyalgia.[1,5]

Surface Elecmyography

The surface EMG signal is an extracellular recording of propagated muscle fiber action potentials that are present within the detection volume of an electrode located at the skin surface. The evoluation of surface EMG techniques for assessing muscle function has been described for the past century.[6,7] Information from the surface EMG signal can be divided into categories describing the temporal aspects of movement, the forces exerted by muscles, and muscle fatigue.[6] Different elements of the EMG signal are quantified to extract the signal information necessary for accomplishing each objective.

Early attempts at establishing a relationship between the EMG signal and muscle fatigue relied primarily on measuring changes in the amplitude of the signal.[6,7] The increase observed in the EMG signal amplitude during a sustained muscle contraction is thought to be the result of the synchronization and recruitment of motor units to compensate for the loss of muscle fiber contractility during fatigue.[2,8] The ratio of the integrated EMG signal to the muscle force has been used as a fatigue index to quantify this phenomenon.[8] The increase in this ratio is postulated to be largely a manifestation of impairment of excitation-contraction coupling mechanisms.[2,8] The practical usefulness of this ratio has been questioned because of the contradictory findings that have resulted from its use and the presence of factors unrelated to fatigue (eg, electrode location) that can influence its measurement.[7]

More recently, a second technique has come into use that is based on the phenomenon of the surface EMG signal undergoing a change in shape during a sustained contraction.[7,9-11] The change in shape can be the result of a decrease in the propagation velocity of muscle fiber action potentials. [H.sup.+] ion accumulation at the muscle membrane from the anaerobic production of lactic acid decreases the excitability of the membrane, resulting in slowing of signal propagation.[4,12] The EMG signal can also undergo a more fundamental alteration of its shape associated with changes in [Na.sup.+] -[K.sup.+] ion balance at the muscle membrane,[4] recruitment[13] and synchronization[2] of motor units, and changes in the length of the depolarized zone (defined as the length of the muscle membrane that is undergoing depolarization).[14] Because EMG signals are inherently noisy (ie, they appear erratic) due to the random occurrences and shapes of action potentials, statistical signal analysis has been the most appropriate technique for obtaining a quantitative description of the EMG signal waveform and its change during fatigue.

The change in EMG signal shape during fatigue can be measured as a shift in the frequency content of the signal toward the low end of its power spectrum (Fig. 1). The power spectrum describes the relationship between signal amplitude and signal frequency. The median frequency, or midpower point of the spectrum, is monitored during the period of muscle contraction to describe this spectral shift (Fig. 1). The median frequency can therefore be considered as a variable that reflects the changes in the EMG signal that result from the metabolic and electrophysiological events associated with fatigue.

The renewed interest in EMG spectral measurements for evaluating muscle function can be attributed primarily to the technical advances that have facilitated its computation[9] and successful research applications to ergonomics and clinical medicine.[6,9-11,15] The most effective EMG instruments for monitoring spectral variables have relied on either analog circuits with adjustable bandpass filters that limit the frequency range of the signal[16] or digital systems.[17,18] The high versatility of digital systems and the low cost of computer memory and processing speed make this latter method the preferred choice at this time. Previous applications of EMG spectral techniques have included measurement of muscle fatigue associated with occupational tasks,[15] muscle impairment associated with lower back pain[19] or other neuromuscular disorders,[20,21] and the adaptation of muscle to exercise or disuse.[22] Still in the developmental stage, the technique is currently not routinely used to make clinical or ergonomic decisions.

Phosphorous Magnetic Resonance Spectroscopy

During relatively low-intensity steady-state exercise, muscles rely predominantly on oxidative metabolism for fuel until the capacity of the mitochondria or the circulatory supply of substrates and oxygen to replenish adenosine triphosphate (ATP) are exceeded. When muscle energy demands exceed the rate of ATP synthesis via oxidative phosphorylation, ATP resynthesis via anaerobic glycogenolysis and phosphocreatine (PCr) breakdown increases. The extent and time course of change in inorganic phosphate ([P.sub.i]), PCr, ATP, and pH can be assessed by [sup.31]P-NMR spectroscopy. Human studies using this procedure have demonstrated that vigorous exercise causes a reduction of tissue PCr with a concomitant accumulation of [P.sub.i] and [H.sup.+] ion.[8,23-27] With ischemia or extremely vigorous exercise, the change in high-energy phosphates occurs within a few seconds.[8,25,28] If PCr is depleted, the normally steady concentration of ATP is also reduced.[28] The proven capability of [sup.31]P-NMR spectroscopy to not only obtain relative measures of energy phosphates at rest but also monitor metabolic changes during exercise and recovery makes this one of the most important recent developments for the study of muscle energetics.

The principle underlying nuclear magnetic resonance (NMR) spectroscopy is that atoms having odd atomic mass or odd atomic charge number, such as hydrogen ([sup.1]H), phosphorous ([sup.31]P), and carbon ([sup.13]C), contain nuclei that spin and create a magnetic moment or vector. When these nuclei are in the presence of an external static magnetic field, such as a whole-body magnet, they precess (rotate) much like a spinning top and align themselves with or against the external field. This phenomemon is similar to the way in which a suspended bar magnet aligns itself to the earth's magnetic field. The angular frequency of this precession is different for different nuclei and is referred to as the Lamour frequency.[26] A second electromagnetic field is then briefly pulsed at right angles to the larger external field to perturb the orientation and subsequent energy level of the nuclei. The frequency of this electromagnetic pulse is typically set at or near the Larmour frequency and is referred to as the nuclear magnetic resonance frequency, which is different for each element or chemical being studied.[26] As the nuclei return to their original orientations, they emit energy that is measured by a radio-frequency receiver as oscillating voltages. This energy, called free induction decay (FID), is transformed into a spectrum by Fourier analysis Fig. 2). The spectral peaks are separable because of differences in chemical bonding between nuclei that create slightly different local magnetic fields. The area of each spectral peak is proportional to the concentration of the nuclei associated with that frequency.

Recent technological advances have permitted construction of magnets large enough to accommodate the entire body or intact human limbs while maintaining adequate field strength and homogeneity to conduct [sup.31]P-NMR studies. Muscle localization and specificity are provided by placing a surface coil turned to the resonant frequency of phosphorous on the skin over the muscle belly to be studied. The coil acts as a transceiver. The coil not only excites the phosphorous nuclei to generate the FID, but it also acts as a radio receiver to record the FID signal, which is then converted to spectral peaks by Fourier analysis. The size and orientation of the coil determine its detection volume and the intensity of the FID. Compared with the EMG signal, the NMR signal is very weak, with poor signal-to-noise characteristics. In practice, the NMR signal can be improved by averaging multiple spectra, enabling the user to more easily recognize and distinguish signal peaks. The sensitivity of the technique for [sup.31]P requires a phosphorous concentration of approximately 1 mN to obtain reasonable signal-to-noise characteristics within several minutes of accumulation and averaging of spectra.[29]

The need for spectral averaging limits the time resolution of the technique for metabolic change. For this reason, most studies using NMR have been performed either on resting muscle or during exercises designed to induce "steady-state" levels of fatigue in which muscle energetics are in equilibrium. Several other limitations of the technique must also be recognized for safety reasons as well as for reduction of artifact. For example, ferromagnetic substances should not be introduced into the magnet; therefore, subjects with metal implants or pacemakers cannot be studied. Also, the usable bores of NMR magnets are confining and limit the type of exercise that can be performed or the size of the instrumentation needed to measure mechanical performance. The muscle under study must not move during testing, and it is recommended that the muscle be positioned at the bore's center, where the field is most homogeneous. Localization of the signal can be further enhanced by combining the imaging capability of NMR with the spectroscopic part of the study.[28,30-32] With this innovation, changes in metabolite concentration can be quantified relative to muscle mass or cross-sectional area.[30]

[sup.31]P-NMR spectroscopy has proven to be a valuable tool in studying localized muscular fatigue during isometric or nonisomenic exercise and following recovery.[1,8,24-26,33-35] High-energy phosphate metabolism and glycolytic activity have been related to the physiological and subjective manifestations of fatigue.[8,23-25,27,30,33,35] Other studies have investigated the control mechanisms for muscular glycolysis,[23,26,28,36] the control of PCr resynthesis after exercise,[8,23-25,33-35] and the reaction rates of glycolytic metabolism.[28,31,32,36] Although still under development as a diagnostic tool, numerous research studies have been reported in which [sup.31]P-NMR spectroscopy has been used to assess muscle disorders.[28,29,31,32] The technique has been applied to the study of metabolic myopathies,[28,31,36-38] muscle enzyme deficiencies,[36,39-41] muscular dystrophies,[32,42] exercise-induced muscle injury,[37,43] chronic fatigue syndrome,[44] and fibromyalgia.[5,25] Other clinically relevant studies based on [sup.31]P-NMR have measured the effects of endurance training on muscle metabolism[34,45,46] and distinguished muscle fiber types without the need for a biopsy.[47]

Surface Electromyography and [sup.31]P-NMR Combined

The assessment of muscle fatigue by the combination of surface EMG with [sup.31]P-NMR spectroscopy was attempted by our group to describe the relationship between metabolic and electrophysiological manifestations of fatigue.[1] Prior to our successful completion of this experiment, only one other report[8] had described the use of both measurement techniques in the same study. This earlier study had evaluated the relative contributions of myoelectric signal propagation failure, phosphate depletion, [H.sup.+] accumulation, and excitation-contraction coupling failure to muscle fatigue and recovery. The adductor pollicis muscle was studied by measuring integrated EMG activity, muscle force, M-wave characteristics, and [sup.31]P-NMR variables. The investigators overcame the technical problem of simultaneously recording EMG and [sup.31]P-NMR signals by detecting each modality separately during the baseline and recovery phases of the protocol. There were no EMG recordings during the exercise phase when the [sup.31]P-NMR signals were being detected.

There are a number of technical and procedural issues that should be recognized before successfully implementing a combined EMG and [sup.31]P-NMR signal experiment. To take full advantage of the ability of both techniques to monitor changes in muscle performance during activity, an appropriate exercise protocol must be defined. The primary considerations are that the exercise and recovery periods be long enough to allow for enough data points of averaged [sup.31]P-NMR spectra to quantify metabolic behavior and that the intensity of the exercise be controlled so that the rate of change of metabolism is resolvable by the [sup.31]P-NMR procedure. One of the fundamental issues to be addressed in the future development of clinical [sup.31]P-NMR spectroscopy is the identification of optimal metabolic "stress tests."[28,30] Physical therapists have the clinical experience and physiological training to make a contribution to this endeavor.

Most EMG protocols for assessing muscle fatigue by spectral analysis have relied on high-intensity, short-duration isometric contractions sustained at a constant muscle force.[7,9-11] The primary reasons for this limitation is that changes in muscle length can affect EMG spectral measurement[48] and that variable forces may introduce "nonstationarities" changes in the EMG signal variance that violate preconditions for implementing spectra analysis.[9,10] Our recommendation for modifying this approach for [sup.31]P-NMR studies is to use isometric contractions at a specified duty cycle. The exercise "stress test" should be performed for at least several minutes with a duty cycle and force level that demand a significant utilization of anaerobic metabolic pathways for ATP production. An alternative to this protocol is to select a relatively low target force level that can be maintained for a long period of time. A "staircase approach" has also been successfully implemented in which the target force starts at a relatively low level and is increased by a fixed amount after an adequate time period to maintain a "steady-state" level of fatigue at each force level.[27,43,45] Identifying optimum exercise specifications for a particular muscle group or application may ultimately require a preliminary study or a trial-and-error approach.

Specially designed and fabricated equipment is needed to implement exercise protocols and record mechanical performance in an NMR magnet. The simplest implementation of these requirements for [sup.31]P-NMR studies is to use a pulley system within the bore of a whole-body magnet that is connected either to adjustable weights or springs located outside of the magnet bore or to a dynamometer. This setup is commonly used for studies of eccentric and concentric contractions[27,28,39-41,45] and could easily be adapted to studies of isometric contractions by replacing the weights or springs with stiff force transducers. More versatile isometric systems have been developed to measure joint torque from instrumentation that can be located within the bore of the magnet.[1,8,27,49] Timing information and force feedback to the subject can be provided either by verbal instructions to the subject using a speaker system or through more elaborate light-emitting diode displays.[1]

Electromyographic instrumentation must be modified to be used in conjunction with [sup.31]P-NMR measurement, particularly when active EMG electrodes are used. Active electrodes provide signal conditioning at the electrode site, which provides differential amplification, filtering, and impedance buffering. Impedance buffering effectively eliminates the need for skin preparations to reduce skin resistance by designing amplifier circuits with high-input impedances or resistances relative to skin resistance.[6] The fact that these circuit modifications are implemented at the detection site of the electrode rather than at the, opposite end of the wire leads (such as at the preamplification stage) significantly limits motion artifact caused by movement of the wire leads.[6,7,50] Passive electrodes (such as the standard silver-silver chloride electrode) do not contain electronic circuitry at the detection site and are susceptible to movement artifact from wire leads, particularly in [sup.31]P-NMR applications, in which long EMG electrode leads are required and the subject is performing an exercise.[6]

Electrode wires can also act as transmitters, bringing unwanted electrical noise from outside of the magnet to within the magnet and further reducing an already low [sup.31]P-NMR signal-to-noise ratio. Nonferrous electronic components are available to construct active electrodes for use with [sup.31]P-NMR spectroscopy. These components, however, must be specifically designed to limit the direct effects of the radio frequency pulses from the NMR surface coil on the EMG signal ground. This artifact appears as a high-frequency interference on the EMG signal (item A in Fig. 3). The radio frequency pulse can also cause a secondary effect on the EMG signal by saturating the capacitors in the EMG electrode amplifier and obliterating a good part of the EMG signal (item B in Fig. 3).

In our laboratory during preliminary studies, we found that the EMG artifact from the [sup.31]P-NMR procedure was dependent on the orientation and distance of the EMG electrode with respect to the radio frequency transceiver. this finding suggested that shielding and filtering of the instrumentation could reduce the artifact. We accomplished this by a threefold process: (1) The electronic circuitry of the EMG electrode was placed in a doubly shielded box of copper mesh with shields connected separately to ground, (2) all connections outside of the magnet were passed through filters that filtered high-frequency noise outside of the EMG signal frequency range and that specifically eliminated the radio frequency signal detected by the surface coil, and (3) the EMG signal ground wires were shielded along their full length. As a result of these modifications, the amplitude of the signal artifact was reduced from 1 mV to 20 [mu]V (duration of 1 millisecond) for an electrode orientation of 180 degrees and 1-cm distance with respect to the coil. The EMG artifacts could be reduced first by software during postprocessing of the signal, although we did not consider this reduction necessary. Another approach to this problem would be to gate the collection of [sup.31]P-NMR signals with the collection of the EMG signals so that the EMG instrumentation is grounded during the 100-millisecond time periods in which the pulse is active. Gating techniques have been described for interleaving electrically elicited contractions with [sup.31]P-NMR acquisition in human fatigue studies.[33]

Implementation of the Technique

Results from a study combining the techniques of EMG spectral measurement and [sup.31]P-NMR spectroscopy are briefly described.[5] The purpose of the study was to determine whether patients with fibromyalgia have impaired muscle metabolism or membrane conductance.

Fibromyalgia affects some 3 to 6 million patients (mostly women 20-40 years of age) and is characterized by widespread soft tissue pain and tenderness.[51] Fatigue following minimal activity is a prominent complaint in 90% of patients with fibromyalgia.[51] Although routine laboratory test results are normal in these patients, there have been inconclusive reports of abnormal muscle metabolite levels.[27,30,51] We hoped to overcome the limitations of previous studies by assessing muscle performance during exercise and recovery rather than only at rest. We also included a control group that had similar levels of aerobic fitness and activity as the patients with fibromyalgia.

Thirteen patients with fibromyalgia and an equal number of control subjects were tested. Results for this report are limited to the tibialis anterior muscle, one of two muscles (tibialis anterior and upper trapezius) investigated in the study. Surface EMG and [sup.31]P-NMR measurements were recorded concurrently while subjects were positioned supine in a wholebody NMR magnet (1.5 T, 60-cm bore). [sup.31]P-NMR measurements were collected first at rest prior to exercise, then during consecutive isometric exercises at 60% of maximal voluntary contraction (MVC) (4.5 minutes' duration) and again at 50% of MVC (4.5 minutes' duration), followed by a recovery period of 9 minutes' duration in which the exercise regimen was not conducted.

The isometric exercises were performed intermittently according to a duty-cycle pattern of 6 seconds of contraction and 3 seconds of rest. The EMG signals were recorded continuously during the exercise period and intermittently during the recovery period for 5-second contractions at 1-or 2-minute intervals. The positioning of the EMG electrode and NMR surface coil is pictured in Figure 4. The figure also shows a custom-made device that was attached to the test leg via a modified brace to measure ankle torque during isometric conditions. Subjects were given visual feedback of the torque produced at the ankle, the target force level, and the timing information needed to comply with the protocol by means of a light-emitting diode display made visible to the subject using a mirror.

A typical series of [sup.31]P-NMR spectra for the different phases of the exercise protocol are displayed in Figure 5 for one of the patients with fibromyalgia. A marked increase in [P.sub.i] and a concomitant decrease in PCr during the exercise is shown, followed by a return to baseline level by the end of the recovery period. The ratio of PCr to Pi provides a useful index of the metabolic state of the muscle.[26,30-32] The mean values of this ratio for patients with fibromyalgia and control subjects were compared for specific periods of the protocol (Fig. 6). The results demonstrated that the energetic state of the muscle was the same for both groups during and following the exercise protocol. Similarly, changes in intracellular pH were also indistinguishable for the two groups tested (Fig. 7).

Typical changes in the EMG median frequency for one of the patients with fibromyalgia during and following exercise are depicted in Figure 8. This variable behaves similarly to the phosphate metabolites and intracellular pH just described. They decreased rapidly during the 60% MVC phase, reached a steady-state plateau during the 50% MVC phase, and then recovered to baseline by the end of the rest period. Calculations of the average rate of change of the median frequency for the 60% MVC contractions (a measure of myoelectric fatigue) resulted in no significant differences between the patients with fibromyalgia and the control group (P=.27).

In summary, these data showed no appreciable differences between patients with fibromyalgia and sedentary controls for muscle phosphate metabolites and EMG spectral measurements during an exercise designed to elicit muscle fatigue. The combined techniques of EMG signal analysis and [sup.31]P-NMR spectroscopy could not identify abnormalities in muscle metabolism or signal propagation in these patients during a fatigue-inducing exercise protocol.

Neither technique alone could have provided as complete a description of these fatigue correlates during exercise. There are concerns, however, that must be resolved before these procedures can be used routinely for research or clinical purposes. The time needed to prepare subjects for testing (2-3 hours) and the costs of conducting these experiments are obvious limitations. These limitations should be mitigated with further technological advances. More importantly, however, is the need for further basic studies to better understand the underlying fatigue processes. The negative findings of our study do not necessarily imply that defects in muscle metabolism or signal propagation do not occur in patients with fibromyalgia or that these changes could not have been detected if other muscles were monitored, or if other fatigue protocols or analyses were performed. Tests using different protocols or involving patients with other suspected muscle disorders or clinical problems may provide data that can be used to optimize the approach of combining these two assessment procedures. Until these preliminary goals are achieved, it is too speculative to discuss the implications for the physical therapy or medical management of these or other patients.

Acknowledgments

Many people contributed to this effort. I wish to acknowledge Robert Simms, Mirko Hrovat, Gary Skrinar, and Carlo De Luca for their scientific collaboration on the fibromyalgia study; Don Gilmore and Mike Tanghe for their technical support; Jennifer Anderson for her statistical expertise; Erwin Boer, Arthur Kerstens, Mark Hurkmans, and Steven Le Poole for their valuable assistance; and the Department of Radiology of Brigham and Women's Hospital, the Neuromuscular Research Center of Boston University, and the Multipurpose Arthritis Center of Boston University for their invaluable support throughout this study.

References

[1] Roy SH, Boer ER, Gilmore D, Hrovat M. Concurrent measurement of muscle fatigue by EMG and [sup.31]P-NMR spectroscopy. In: Anderson PA, Hobart DJ, Danoff JV, eds. Electromyographical, Kinesiology. Amsterdam, the Netherlands: Elsevier Science Publishers BV; 1991: 279. [2] Bigland-Ritchie B, Furbush F, Woods J. Fatigue of intermittent, submaximal voluntary contractions: central and peripheral factors. J Appl Physiol. 1986;61:421-429. [3] Vollestad NK, Sejersted OM. Biochemical correlates of fatigue: a brief review. Eur J Appl Physiol. 1988;57:336-347. [4] Juel C. Muscle action potential propagation volocity changes during activity. Muscle Nerve. 1988;11:714-719. [5] Simms RW, Roy SH, Skrinar G, et al. [sup.31]P-NMR spectroscopy of muscle in fibromyalgia syndrome patients and sedentary controls. In: Proceedings of the 55th Annual Scientific Meeting of the American College of Rheumatology. 1991. [6] Basmajian JV, De Luca CJ. Muscles Alive. Baltimore, Md: Williams & Wilkins; 1985. [7] De Luca CJ. Myoelectric manifestation of localized muscular fatigue in humans. CRC Critical Reviews in Biomedical Engineering 1985;11:251-279. [8] Miller RG, Giannini D, Milner-Brown HS, et al. Effects of fatiguing exercise on high-energy phosphates, force, and EMG: evidence for three phases of recovery. Muscle Nerve. 1987; 10:810-821. [9] Merletti R, De Luca CJ. New techniques in surface electromyography. In: Desmedt JE, ed. Computer-Aided Electromyograpby and Expert Systems. Amsterdam, the Netherlands: Elsevier Science Publishers BV; 1989:115-124. [10] Merletti R, Knaflitz M, De Luca CJ. Myoelectric manifestations of fatigue in voluntary and electrically elicited contractions. J Appl Physiol. 1990;69:1810-1820. [11] Roy SH, De Luca CJ. Evolving characteristics of the median frequency of the EMG signal. In: Desmedt JE, ed. Computer-Aided Electromyography and Expert Systems Amsterdam, the Netherlands: Elsevier Science Publishers BV; 1989:205-221. [12] Brody LR, Pollock M, Roy SH, et al. pH-induced effects on median frequency and conduction velocity of the myoelectric signal. J Appl Physiol. 1991;71:1878-1885. [13] Solomonow M, Baten C, Smit J, et al. Electromyogram power spectra frequencies associated with motor unit recruitment strategies. J Appl Physiol. 1990;68:1177-1185. [14] Gydikov A, Gantchev N. Velocity of spreading of the excitation along the muscle fibres of human motor units measured by superficial electrodes. Electromyogr Clin Neurophysiol. 1989;29:131-138. [15] Hagg GM. Comparison of different estimators of electromyographic spectral shifts during work when applied on short test contractions. Med Biol Eng Comput. 1991;29:511-516. [16] Gilmore LD, De Luca CJ. Muscle fatigue monitor: second generation. IEEE Trans Biomed Eng. 1985;32:75-78. [17] Kramer CGS, Hagg T, Kemp B. Real-time measurement of muscle fatigue related changes in surface EMG. Med Biol Eng Comput. 1987;25:627-630. [18] Merletti R, Biey D, Biey M, et al. On-line monitoring of the median frequency of the surface EMG power spectrum. IEEE Trans Biomed Eng. 1985;32:1-7. [19] Roy SH, De Luca CJ, Casavant DA, Lumbar muscle fatigue and chronic lower back pain. Spine. 1989;14:992-1001. [20] Inbar GF, Noujaim AE. On surface EMG spectral characterization and its application to diagnostic classification. IEEE Trans Biomed Eng. 1984;31:597-604. [21] Ronager J, Christensen H, Fuglsang-Frederiksen A. Power spectrum analysis of the EMG pattern in normal and diseased muscles. J Neurol Sci. 1989;94:283-294. [22] Thompson DA, Biedermann H, Stevenson JM, MacLean AW. Changes in paraspinal electromyographic spectral analysis with exercise: two studies. J Electromyogr Kinesiol. 1992;2: 179-186. [23] Dawson MJ, Gadian DG, Wilkie DR. Muscular fatigue investigated by phosphorus nuclear magnetic resonance. Nature. 1978;274:861-864. [24] Dawson MJ, Gadian DG, Wilkie DR. Contraction and recovery of living muscles studied by 3 1P nuclear magnetic resonance. J Physiol (Lond). 1977;267:703-735. [25] Miller RG, Boska RS, Moussavi RS, et al. [sup.31]P nuclear magnetic resonance studies of high-energy phosphates and pH in human muscle fatigue. J Clin Invest. 1988;81:1190-1196. [26] Sapega AA, Sokolow DP, Graham TJ, Chance B. Phosphorus nuclear magnetic resonance: a non-invasive technique for the study of muscle bioenergetics during exercise. Med Sci Sports Exerc. 1987;19:410-420. [27] Wilson JR, McCully KK, Mancini DM, et al. Relationship of muscular fatigue to pH and diprotonated Pi in humans: a [sup.31]P-NMR study. J Appl Physiol. 1988;64:2333-2339. [28] Bottomley PA. Human in vivo NMR spectroscopy in diagnostic mediciine: clinical tool or research probe? Radiology. 1989;170:115. [29] Shulman GI, Alger JR, Prichard JW, Shulman RG. Nuclear magnetic resonance spectroscopy in diagnostic and investigative medicine. J Clin Invest. 1984;74:1127-1131. [30] Kushmerick MJ. Muscle energy metabolism, nuclear magnetic resonance spectroscopy, and their potential in the study of fibromyalgia. J Rheumatol. 1989; 16:40-46. [31] Radda GK. The use of NMR spectroscopy for the understanding of disease. Science. 1986;233:640-645. [32] Weiner MW. The promise of magnetic resonance spectroscopy for medical diagnosis. Invest Radiol. 1988;23:253-261. [33] Helpern JA, Kao W, Gross B, et al. Interleaved 31P-NMR with transcutaneous nerve stimulation (TNS): a method of monitoring compliance-independent skeletal muscle metabolic response to exercise. Magn Reson Med. 1989;10:50-56. [34] Kuno S, Akisada M, Mitwumori F. Phosphorus-31 nuclear magnetic resonance study on the effects of endurance training in rat skeletal muscle. Eur J Appl Physiol. 1992;65: 197-201. [35] Le Rumeur E, Le Moyec L, Toulouse P, et al. Muscle fatigue unrelated to phosphocreatine and pH: an "in vivo" [sup.31]P-NMR spectroscopy study. Muscle Nerve. 1990; 13:438-444. [36] Duboc D, Jehenson P, Dinh ST, et al. Phosphorus NMR spectroscopy study of muscular enzyme deficiencies involving glycogenolysis and glycolysis. Neurology. 1987;37:663-671. [37] de Kerviler E, Leroy-Willig A, Jehenson P, et al. Exercise-induced muscle modifications: study of healthy subjects and patients with metabolic myopathies with MR imaging and P-31 spectroscopy. Radiology. 1991;181:259-264. [38] Itoh M, Iio M, Kawai M, et al. [sup.31]P-NMR spectroscopy of myopathies: clinical application of whole-body MR. Radiat Med. 1986;4: 41-45. [39] Argov Z, Bank WJ, Maris J, Chance B. Muscle energy metabolism in McArdle's syndrome by in vivo phosphorus magnetic resonance spectroscopy. Neurology. 1987;37:1720-1724. [40] Chance B, Eleff S, Bank W, et al. [sup.31]P-NMR studies of control of mitochondrial function in phosphofructokinase-deficient human skeletal muscle. Proc Natl Acad Sci USA 1982;79:7714-7718. [41] Lewis SF, Haller RG, Cook JD, Nunnally RL. Muscle fatigue in McArdle's disease studied by [sup.31]P-NMR: effect of glucose infusion. J Appl Physiol. 1985;59:1991-1994. [42] Younkin DP, Berman P, Sladky J, et al. [sup.31]P-NMR studies in Duchenne muscular dystrophy: age-related metabolic changes. Neurology. 1987;37:165-169. [43] McCully KK, Argov Z, Boden BP, et al. Detection of muscle injury in humans with [sup.31]P magnetic resonance spectroscopy. Muscle Nerve. 1988;2:212-216. [44] Lewis SF, Haller RG. Physiologic measurement exercise and fatigue with special reference to chronic fatigue syndrome. Rev Infect Dis. 1991;13(suppl 1):S98-S108. [45] McCully KK, Boden BP, Tuchler M, et al. Wrist flexor muscles of elite rowers measured with magnetic resonance spectroscopy. J Appl Physiol. 1989;67:926-932. [46] Laurent D, Bernus G, Alonso J, et al. Effect of training on the calf muscle energy metabolism: a [sup.31]P-NMR study on four elite downhill skiers challenged with a standardized exercise protocol. Intl J Sports Med. 1992; 13:313-318. [47] Kushmerick MJ, Moerland TS, Wiseman RW. Mammalian skeletal muscle fibers distinguished by contents of phosphocreatine, ATP, and P,. Proc Natl Acad Sci USA. 1992;89:752 1-7525. [48] Inbar GF, Allin J, Kranz H. Surface EMG spectral changes with muscle length. Med Biol Eng Comput 1987;25:683-M. [49] Reutenauer H, Payen JF, Laurent D, et al. Developing a simplified ergometer to study the human calf muscles through [sup.31]P-NMR spectroscopy. Magn Reson Med Biol. 1989;2:205-214. [50] De Luca CJ, Le Fever RS, Stulen FB. Pasteless electrode for clinical use. Med Biol Eng Comput. 1979;17:387-390. [51] Fan PT, Blanton ME. Clinical features and diagnosis of fibromyalgia. J Musculoskel Med. 1922;9:24-42.

SH Roy, ScD, PT, is Research Assistant Professor, Neuromuscular Research Center, Boston University 44 Cummington St, Boston, Mass 02215 (USA).

This study was approved by the Charles River Institutional Review Board, Boston University.

This research was supported by NIH Grant AR 20613 and the Liberty Mutual Insurance Company Inc.