| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
INNOVATIVE METHODOLOGY
1Center for Sensory-Motor Interaction (SMI), Department of Health Science and Technology, Aalborg University, Aalborg, Denmark; 2Biomedical Engineering Department, Indiana University-Purdue University, Indianapolis, Indianapolis, Indiana; 3Department of Medical Engineering and Neuroprosthetics, Fraunhofer Institute for Biomedical Engineering, St. Ingbert, Germany; and 4Laboratory for Biomedical Microtechnology, Department of Microsystems Engineering-IMTEK, University of Freiburg, Freiburg, Germany
Submitted 19 July 2007 ; accepted in final form 21 November 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
motor unit; electromyograph; thin-film technology; spatial sampling
In vivo identification of extracellular action potentials in multiunit EMG recordings has allowed the assessment of the discharge pattern of motoneurons from a specific area in the muscle near the recording site of a highly selective electrode. The selectivity of the recording is necessary for the identification of individual motor units from the multiunit recordings; however, there is a limit to the number of motor units that can be investigated concurrently. For this reason, most in vivo studies report results on few motor units that constitute only a small proportion of the population of active motor units from a relatively small muscle area (6). Most of the knowledge on motor unit physiology is based on the interpretation of ensembles of serially recorded single-unit activity from different sessions and subjects. Such recordings enable the development of a generalized scheme of muscle control, although some limitations remain.
The detection of signals from many locations in the muscle, which is known as spatial sampling, is a viable strategy for increasing the number of detected sources and the sampled muscle area. The loss of information determined by high selectivity in each detection location is compensated for by sampling many muscle regions. This approach has been applied in surface EMG recordings (15, 17). Multichannel needle electrodes and multiwire electrodes have also been applied (1, 3) but with detection surfaces closely spaced between each other, thus recording the same motor unit activities from slightly different locations. Moreover, the needle electrode systems are impractical in many applications, such as dynamic contractions, and are uncomfortable for the subject.
Similar issues have been faced in nerve recordings where multiple sites allow the sampling of subpopulations of fibers conveying information to and from the limb. Longitudinal intrafascicular electrodes (LIFEs) are fine-wire electrodes designed to be implanted into peripheral nerves (10, 14). Conventional LIFEs are constructed using Teflon-insulated platinum-iridium wires (9) or metallized and insulated polyaramid filaments (12). The number of wires that can be attached to the tungsten needle limits the number of recording channels. The limit is two channels for Pt-Ir electrodes (22). To expand this capability, our group has experimented with microfabricated thin-film LIFEs for nerve recordings (24). The thin-film technology allows the design of multiple detection sites with consistent site geometry on a flexible substrate that is relatively small. This technology has not yet been applied to intramuscular electrodes. This paper describes the development and test of a prototype of thin-film system for intramuscular motor unit recordings. The system is currently limited to eight channels, as a proof of principle, and was tested on animals.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
The structure of the system is shown in Fig. 1. The device is 220 µm wide, 10 µm thick, and 1.5-cm long, and it has eight circular platinum-platinum chloride recording sites, each with a diameter of 40 µm distributed along the front and back surfaces, with 1,500-µm intersite spacing. All tracks are 10-µm wide, 300-nm thick, and made of platinum. This leads to an average track resistance of 
600 
, which is small compared with the interface impedance of the small electrode contacts.
|
|
4.3 kg) under protocols approved by the Danish Committee for the Ethical Use of Animals in Research. The rabbits were anesthetized using intramuscular injections of 0.15 mg/kg midazolam (Dormicum, Alpharma, Oslo, Norway), 0.03 mg/kg fentanyl, and 1 mg/kg fluranison (combined in Hypnorm, Janssen Pharmaceutica, Beerse, Belgium), and they were maintained with regular injections of the same anesthesia every 20 min. The depth of anesthesia was assessed by monitoring the heart rate and blood oxygen saturation. A Steinmann bone pin was placed near the distal epiphysis of the left femur and was used to anchor the preparation to the experimental apparatus. The thin-film electrode system was implanted into the medial head of the gastrocnemius muscle, perpendicular to the pennation angle of the muscle fibers.
Insertion method.
The microfabricated thin-film electrode was connected to an 80-µm-diameter, 1.5-cm-long tungsten needle by several strands of polyaramid filaments that were glued to the needle using a cyanoacrylate adhesive, as shown in Figs. 2 and 3 (7, 23). A surgical incision was made in the skin directly overlying the insertion site, and the skin was retracted to expose the belly of the muscle. Under the stereomicroscope, the fiber orientation of the intended implant site was visualized. The tungsten needle, to which the thin-film structure was attached by the polyaramid filaments, was inserted through the epimysium and parymysium, and it was threaded through the endomysium, approximately perpendicular to the fiber orientation, for 8 mm before it was directed out of the muscle. Figure 2 shows schematically the entry and exit points of the needle and the position of the thin-film system inside the muscle. The needle was then pulled to draw the thin-film electrode into the muscle via the polyaramid filament that connected the electrode to the needle. The needle was pulled in a way that the eight recording sites of the thin film were centered between the entry and exit points into and out of the muscle (Fig. 2). The needle and the polyaramid filament thus served only to introduce the thin film into the muscle. The polyaramid filament was cut after the insertion, the needle was removed, and the thin-film structure was left in the muscle for signal recording. The electrode was stabilized by sewing the connector to the animal skin (Fig. 2).
|
The EMG signals were amplified and filtered using ultra-low-noise preamplifiers (AI402, Axon Instruments) attached to an eight-channel main amplifier (Cyberamp 380, Axon Instruments). The differential preamplifiers were connected to the electrodes in monopolar recording configuration with respect to a common, distant indifferent electrode. The gain and filter characteristics of this configuration were as follows: gain 5,000, high-pass filter 1 Hz (2nd-order Bessel), low-pass filter 20 kHz (1st-order Bessel).
A customized eight-channel digital tape recorder (ADAT-XT, Alesis) was used to acquire and digitally store the signals for offline analysis. All channels on the digital tape recorder were sampled simultaneously at 48 kHz. Although eight channels were available on the electrode system, only four channels were recorded at any time because of the limited number of ultra-low-noise preamplifiers. However, all detection sites of the system were tested in each experimental session by recording the responses from different groups of four channels.
In the first four experiments, four to five contractions were elicited with withdrawal reflexes or by crushing the sciatic nerve. In the last two experiments (long term), the recordings were performed three times with a separation of 20 min between each set of recordings, for a total of 40-min separation between the first and last set of measures during which the electrode system remained in place. In these long-term experiments, two contractions were elicited for each set of recordings. During the 20 min that separated consecutive sets of measures, the leg of the rabbit was passively extended and flexed (manually by the operator) in the entire range of motion at a speed of 1 cycle/s for 20 times. After this maneuver and before recording, the leg was repositioned at approximately the initial position. The aim of these recordings was to assess the stability of the system over time and following slow movements of the leg.
After production, the integrity of the thin-film system was tested under different sterilization protocols that employed the use of an ultrasonic cleaner, Liquinox (Alconox), ethanol (reagent grade, Fisher Scientific), and ultrapure water (Milli-Q, Millipore). All experimental measures described above were performed after the system was tested for resistance to these sterilization procedures. In addition, 2 h before the last two experiments the thin-film system was sterilized in an autoclave, because it is usually done for wire electrodes.
After completion of the experimental procedures, the animals were euthanized with an intravenously delivered overdose of pentobarbital sodium, in accordance with the approved procedure.
Signal analysis. The signals were analyzed for 4 s after eliciting each withdrawal reflex or nerve crush. The action potentials of the detected motor units were identified from the recordings with a decomposition algorithm previously described (11). This interactive algorithm includes a user interface for manually editing and verifying the results (11). The software displays a segment of the EMG signal, the templates of the action potentials of the identified motor units, the discharge patterns, and a close-up of the signal for resolving missed discharges and superimpositions. Accuracy of the automatic decomposition was achieved by inspection of the identified discharge patterns. Although the decomposition program can handle multichannel signals, in this study signal decomposition was applied to each monopolar channel independently. The decomposition results from the different channels were then merged by automatically identifying the motor units detected at more than one detection site based on the estimated discharge pattern. Discharge patterns with more than 90% discharges closer than 1 ms were considered to belong to the same motor unit detected on different channels. Motor unit action potentials identified in subsequent recordings in the same experimental session were compared by cross-correlation and merged to indicate the responses to the sequential stimuli. The procedures for merging the single-channel decomposition results were performed with custom-made algorithms developed in Matlab version 7.0 (The Mathworks, Natick, MA).
Action potentials generated by the same motor unit were averaged and characterized by the peak-to-peak amplitude in the channel with the maximum amplitude. Furthermore, the amplitude of the action potential at the neighboring recording sites was normalized (%) to the maximal peak-to-peak amplitude. The root mean square of the noise was estimated from 1-s intervals without EMG activity. The analysis of action potentials and noise was performed in two conditions: without offline filtering and after applying a digital filter with band pass 200 Hz to 5 kHz (anticausal Butterworth filter of order 4, implemented in Matlab). In the first condition, the signals were only filtered in the bandwidth 1 Hz to 20 kHz by the analog filters in the amplifiers, whereas the second condition corresponded more closely to the filter settings usually applied for intramuscular recordings. The bandwidth 200 Hz to 5 kHz for the offline digital filter was chosen as that minimally affecting the action potential energy, as assessed by spectral analysis of the identified action potentials.
The signals recorded at the detection sites were also linearly combined to assess the effect of filtering in the spatial domain (15). Single, double, and triple derivatives were obtained. The single differential (or bipolar) detection corresponds to the difference between two signals recorded from adjacent detection sites, the double and triple differentials denote the linear summation of signals from three and four adjacent sites, respectively, with weights 1, –2, 1, and 1, –3, 3, –1. The ratio (%) between the peak-to-peak amplitude of the action potential after the derivative operation (single, double, or triple differential) and the peak-to-peak amplitude of the monopolar signal was used to quantify the effect of linearly combining signals from the detection sites. Similar quantification indexes were used to analyze signals obtained as the difference of monopolar signals recorded from the same location at the two opposite sides of the thin-film structure. Data are reported as means and SD.
|
RESULTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
1 min on different detection sites of one thin-film structure, the impedance was found to be 53.0 ± 17.5 k. It was possible to identify repetitive activation of motor units in all experiments. The detected action potentials represented the electrical activity of a group of muscle fibers close to the detection site and belonging to the same motor unit. Figure 4 shows the identification of a motor unit from a multiunit recording that followed a nerve crush. In this example, the signal was digitally filtered in the bandwidth 200 Hz to 5 kHz.
|
|
From the four short-term experiments and the first set of measures of the 2 longer experiments, 34 of the 67 motor units were identified by the decomposition program in all 4 channels concurrently, whereas the action potentials of the other 33 motor units were identified in only 2 or 3 channels, because of the selectivity of the recording. The amplitude of motor unit action potentials at sites adjacent to the site with maximum amplitude was 23.3 ± 43.2% of the maximum amplitude.
Figure 5 shows recordings obtained by linear combination of the signals detected from four detection sites. In the representative example of Fig. 5, the different recordings discriminated different subsets of motor units. The recordings from the first four experiments and the first set of measures of the longer experiments in which the signals from sites L1 to L4 were concurrently detected were further investigated. In these recordings, it was possible to identify a total of 30 motor units from the monopolar signal recorded at L2. The action potentials of these motor units were enhanced or attenuated when performing linear combinations of the signals, as for the representative example in Fig. 5. The ratio (%) between the peak-to-peak amplitude after the linear combination and the peak-to-peak amplitude as recorded at L2 for the 30 motor units was in the range 30.1% to 125.2% (single differential), 7.2% to 122.3% (double differential), and 8.1% to 145.1% (triple differential). Thus some motor unit action potentials were attenuated (ratio <100%) and others amplified (ratio >100%) by the linear combination.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Thin-film technology. Thin-film technology has been recently used as intrafascicular electrodes in peripheral nerves (24). Because the nerve is organized in clusters of topographically related fibers (18), many detection sites allow recording of subpopulation of fibers (24). There are currently no applications of the thin-film technology to muscle electrophysiology, however. The present study demonstrates that thin film is a viable technology for intramuscular motor unit recordings. The prototype tested has a thickness of 220 µm, which is larger than that of Teflon-coated stainless steel wires typical for in vivo intramuscular EMG recordings (50–100 µm) but it can be inserted in standard small needles. As a proof of principle, the current prototype consists of eight channels, but the technology implemented allows the extension to a larger number of channels, different intersite distances, and different detection areas. For example, a two-layer thin-film system with 16 electrodes and any intersite spacing would have a width of 280 µm and thickness of 15 µm, that is only slightly larger than the current system. Moreover, this microtechnology allows the combination of the sensors with telemetry and wireless communication (20), which can have applications in myoelectric prostheses controlled by multichannel intramuscular EMG.
Larger number of channels and intersite distance would increase the sampled muscle area and the number of detected sources. For example, a 32-channel system constructed with 2,000-µm intersite spacing would sample from the muscular cross section along 6.2 cm and could be used in large muscles. The same system designed with 400-µm intersite distance would cover 1.24 cm with 32 channels, for use in smaller muscles. A larger site area would reduce the selectivity at the individual recording sites but would decrease the noise level that is directly related to the contact impedance. The flexibility of the design allows the production of systems suited for a variety of applications.
Multichannel EMG recordings. Highly selective systems combined with high-density spatial sampling have been adopted for surface EMG detection (25). Multichannel surface EMG is suitable for spatial aspects of motor unit activity, for example to estimate the fiber propagation velocity or the location of the innervation zones (25). A multichannel intramuscular system is, on the other hand, more suited for assessing the temporal characteristics of a group of motor units, i.e., the discharge patterns.
Multichannel needle electrodes and multiwire systems have been previously applied (1, 4, 21). In these systems, however, the spatial sampling is used to record the same motor unit activity from slightly different locations to improve the accuracy of action potential classification (1, 8, 21). Alternatively, it is possible to perform subsequent recordings of EMG signals from several needle detection sites (ranging from superficial to deep needle positions) by progressively increasing the insertion depth (16). This technique does not allow concomitant recording of many motor units but does provide separate scans of the muscle electrical activity varying the location of the needle. Previously proposed multichannel, intramuscular detection systems with the aim of spatial sampling (3) were based on needle technology, with the disadvantages of poor stability of the recording, discomfort for the subject, difficulty of application during strong contractions or movement, and poor flexibility in the design of the detection sites.
The thin-film system proposed in this study consists of detection surfaces that have a diameter of 40 µm with distances between detection sites that can be varied from a few hundred micrometers to several millimeters. The small detection sites result in high selectivity that can be further improved by filtering in the time domain (11) or by linear combination of signals detected at the recording sites (Fig. 5). Despite the small detection surface, the experimental tests showed that the signal-to-noise ratio was high enough for discriminating motor unit action potentials (Fig. 4). The signal-to-noise ratio was substantially improved by offline digital filtering of the signal (Table 1), as expected.
The proposed system was also tested for stability. The results on long term recordings showed that the noise level did not substantially change over 40-min intervals and with passive slow movements of the limb. Finally, the system was tested under different sterilization procedures, including autoclaving. No damage to the system was observed after sterilization. The thin-film electrode recorded signals of similar quality with or without autoclaving before the recording (compare experiments 5 and 6 with the first four experiments in Table 1).
Limitations. The study reports the construction of a single type of intramuscular thin-film structure. The design of the system is similar to that of thin-film systems for nerve recordings (24). Other design choices may have resulted in a system better optimized for muscle recordings. However, a similar design with respect to nerve recording systems allowed the application of previously tested methods for the connector and insertion procedure (Fig. 2), which was a necessary initial step.
The experimental tests were performed on animal preparations with the skin open and the muscle exposed. The system was inserted into the muscle tissue with a needle that passed through the muscle in two points (Fig. 2). This procedure can be directly applied in human studies with subcutaneous insertion of the thin-film structure (5). The thin-film systems developed for the experiments presented in the present study have been tested under different conditions of mechanical stress after production; thus insertion through the skin would not cause damage to the system. However, the insertion modality used in this study cannot be applied for recordings inside the muscle in vivo, where a different procedure for insertion would have to be implemented. A feasible solution is the construction of a thin-film system whose top part is bent at the tip of the needle, as commonly done with classic wire EMG recordings.
The thin-film system was applied during reflex muscle activity or nerve crushes, during which it is not possible to modulate the number of active motor units, in contrast to the graded activation of the motor unit pool that occurs during voluntary contractions. Ordered activation of individual motor units was not possible with the present experimental procedures. Electrical stimulation of the nerve would have allowed a gradation in the number of activated motor units, but the recorded signals would have had different characteristics with respect to asynchronous motor unit activation. The electrically stimulated signals are deterministic, quasi-periodic signals in which a compound action potential made of the synchronous contributions of a number of motor units repeats with similar shape at each stimulus. The characterization of the system was, on the contrary, based on the analysis of individual motor unit action potentials repeating in the recording with stochastic characteristics, because it also happens in voluntary contractions.
In summary, this study describes an innovative detection method for single motor unit recordings through a new technology for electrophysiological muscle investigations. The developed electrode can be sterilized for multiple uses and was proven to record signals with stable noise level for up to 40 min.
|
GRANTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
|
REFERENCES |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. Merletti and D. Farina Analysis of intramuscular electromyogram signals Phil Trans R Soc A, January 28, 2009; 367(1887): 357 - 368. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
