Electroencephalography and clinical Neurophysiologg,, 84 (1992) 473-476
473
q) 1992 Elsevier Scientific Publishers Ireland, Ltd. 0168-5597/92/$05.00
EVOPOT 92008
"Cross-talk" in recording evoked potentials Takeshi Nagaoka *, D. David Walker, Peter J. Seaba and Thoru Yamada Dicision ()f' Clinical Eh'ctrophysiology, Department of Neurolog3', Unit:ersity of h)wa, Colh'ge ~f Medicine, Iowa City, IA 52242 (U.S.A.) (Accepted for publication: 17 July 1992)
Summary
When two potentials having large amplitude differences are simultaneously recorded, the large amplitude potential contaminates the small amplitude response. The small, early potentials generated by this contamination resemble far-field potentials. Although scalp-recorded SEP was contaminated by waves similar to the peripheral potential, peak latencies and wave form were not identical. Experiments simulating the recording situation verified the presence of "cross-talk." Capacitive coupling would shift peaks and alter the wave forms. Other possible mechanisms for the cross-talk and methods of minimizing it are offered. One should be cautious interpreting the results when potentials of large amplitude differences are simultaneously recorded. Key words: Somatosensory evoked potential; Stray capacitance; Contamination
After adding a channel of peripheral nerve action potential (NAP) or compound muscle action potential (CMAP) to our somatosensory evoked potential (SEP) recording protocol, we observed substantial alteration of SEP wave forms. The scalp-recorded SEP was contaminated by waves similar to the peripheral potential. Experiments electronically simulating the recording situation verified that the source of the "cross-talk" was in the amplification system. Frequency-dependent cross-talk pointed to capacitive coupling as the mechanism of "cross-talk." This report is to alert others to the problem we have experienced.
Material and method In t,ivo simulatkm Healthy volunteers (6 males) were used to record scalp SEP elicited by peroneal nerve stimulation at the knee. Stimulus electrodes were 7 mm diameter, flatsurfaced disks, placed over the right peroneal nerve at the knee, with the cathode 2 cm proximal to the anode. Stimuli of 0.1 msec duration were delivered at a rate of 3.1/sec via a stimulus isolation unit. The intensity was adjusted to elicit modest contractions of the peroneus
Correspondence to: Thoru Yamada, M.D., Division of Clinical Electrophysiology, Department of Neurology, University of Iowa, College of Medicine, Iowa City, IA 52242 (U.S.A.). * Present address: Department of Orthopaedic Surgery, Nihon University, Tokyo, Japan.
longus and brevis muscles, approximately 3 - 4 times sensory threshold. Recording electrodes were collodion-adhered Ag-AgCI cups with impedances less than 5 k~Q. A single channel SEP was recorded from Cz referenced to the right ear lobe. Peripheral potentials were recorded using similar recording electrodes and techniques. Recording electrodes were placed at the ankle overlying the superficial peroneal nerve, with an inter-electrode distance of 5 cm. The ground electrode was a cup electrode at the forehead (Fpz) or a strap electrode around the calf 10 cm proximal to the stimulus electrodes. The leads were connected to a commercial E E G instrument via a headboard containing preamplifiers. The amplifier bandpass was the same as used in clinical SEP recordings: 15-1500 Hz (all channels). Amplification was 2 × 105 for SEP and 104 for peripheral potential recording. 1500 responses were averaged. Electrical simulation To verify the "cross-talk" phenomenon, we experimented with a circuit commonly used to measure cross-talk. A 5 mV sine wave was applied to a pair of electrode inputs selected by one channel of the E E G instrument. Another pair of electrode inputs, selected by a second channel, were connected to ground by via 5 kS2 resistors (Fig. 1). The same electrode inputs that had been used in the SEP recordings were used. We measured the cross-talk appearing in the second channel while incrementing the sine-wave frequency from 10 to 1000 Hz. The amplification for sine wave input channel was 3 × 10 4 and 3 × 10 6 for the second (cross-talk) channel.
474
T. N A G A O K A E T A L
G2
I
~
A
CHANNEL n]
GI
B RG2 ~ 5K I R
5K
i
l ',
.... >
|
> EHANNF[
kJ
~!
[ I I
Fig. 1. Equivalent circuit for stray-capacitance coupling between inputs shown in dotted lines.
1
I t I f
II/~L
C
Results
SEPs from peroneal nerve stimulation consisted of an initial small positive peak with a latency range of 23.8-26.0 msec (mean = 25.1 msec), and subsequent large negative-positive-negative peaks. This was similar to previous peroneal SEP studies by others (Kimura et al. 1978; Vas et al. 1981; Onishi et al. 1991). Peripheral electrodes registered the polyphasic waves, an amalgam of nerve action potentials (NAP) and compound muscle action potential (CMAP) potentials, termed here the "peripheral potential." When the peripheral potential was recorded simultaneously with the scalp SEP, the SEP response showed polyphasic waves prior to the initial positive peak, with similar, but not identical, wave shapes or latencies as the peripheral potential (Fig. 2). This was observed in all subjects.
--F Z
/ 2mV
L
L
o
I
30
60ml~
Fig. 2. Peroneal SEP without recording of peripheral potentials shown in A. B is the contaminated SEP with concomitant recording of peripheral potential (C). Note the differences in wave forms and phase shift of contaminated waves from the source potential.
Amplitudes of peripheral potential recorded in the 6 subjects ranged from 0.9 to 3.6 mV (mean = 1.8 mV). These were measured from the highest peak to the lowest trough of the polyphasic waves. The ratio of cross-talk to source-potential amplitudes was roughly
A
B Attenuation
Peripheral
=r,o0
> 2mY
20 u V
+
C
Scalp SEP (C2-A
1)
-I + i
l
I
[
I
I
0
30
60
0
30
60msec
Fig. 3. The peripheral potential (A) was attenuated to approximately 1/100 (B) of the original amplitude. The contaminated waves before the attenuation (C) were diminished but not completely eliminated after the attenuation (D).
" C R O S S - T A L K " IN R E C O R D I N G EPs
475
from 1/500 to 1/1000. The ratio varied depending on the duration of source potential waves; slower waves showed less cross-talk than faster waves, as if the source potential was filtered by a high-pass characteristic. Separating or gathering the electrode wires from the scalp and from the leg did not measurably affect the cross-talk, nor did reducing the amplifier gains. Crosstalk remained when selecting electrode jacks that were widely spaced on the electrode board. Various "ground" sites were also investigated (leg and scalp) to eliminate the chance that large common mode peripheral potentials were overwhelming the amplifier. This did not affect the appearance of peripheral potentials in the cortical channels. To show that the cross-talk occurred after the electrode inputs, an attenuator was used. The different amplitude signals would still be present at the electrodes, in the wires, etc. However the peripheral potential presented to the electrode inputs would be reduced to the level of cortical potentials. The attenuation circuit had an attenuating factor of 1 : 100 with an input impedance similar to that of the electrode input. Series resistors of 10 M/2 each provide input resistance similar to the buffers 10 M ~ input impedance. Shunt resistors of 100.000 (100 K) resulted in a 1 : 102 ratio.
A
1
0
Isec
tj\,,/' 11
Chl
"
t
oo o
"°°
O° ~'~~°~'D.~7~l|°"oo°e°~°aD°ae°°°°o°°=n°°°°°°nae°°°aD= a°°ooo 5
I
Hz lOO
5oo
lOOO
Frequency of sine wave
Fig. 5. The relationship of contaminated wave amplitude with the frequency of source potential (2.5 mY). The amplitude of contaminated wave increased about linearly after the increase of source potential frequency greater than 300 Hz.
When a peripheral potential of about 2 mV was attenuated to 20/xV (1/100), the cross-talk was largely, but not completely, eliminated (Fig. 3). We then examined the electrically simulated crosstalk model. A 5 mV, 10 Hz sine wave was applied to channel 1. Channel 2, with inputs connected to ground via 5 kg2 resistors, recorded a 9/xV sine wave. The sine wave had nearly the same phase as the source potentials (Fig. 4A). When the frequency was increased to 1000 Hz, channel 2 registered a higher amplitude sine wave (23 #V), phase shifted from channel 1 (Fig. 4B). We also found that the cross-talk was much greater when unequal resistances were used to ground the channel 2 inputs; for example, 5 k£2 in " G I " ( - ) and 10 k ~ in " G 2 " ( + ) (Fig. 4C). Progressive increments of source frequency showed near-linear increase of contaminated sine wave amplitude especially when the source frequency was greater than 300 Hz (Fig. 5).
Discussion
-T
Oh2
+"
0
lOms©c
/
/ ',, j , /
t
I / k/
',
I]
~
/
C
pV
~
':'
"
-
7 ;
3mY
~1 ,/
,.
%'
\ Ch2
"\
0 [
_
.
_
.
z .
.
.
.
.
.
~
_o_ ._ ~ , 10~sec
Fig. 4. Contamination of 5 mV sine wave fed in channel l introduced in channel 2. The contaminated wave in channel 2 is greater with 1000 Hz (B) than with 10 Hz (A) sine wave. The contamination is much greater with the use of unequal registers, 5 k,¢2 in " G I " and 10 k.Q in " G 2 " (C). Note also the greater phase shift with the faster frequency sine wave.
Some capacitance exists between any two wires (Grob 1971). This "stray" capacitance is greater with closer proximity or with greater length of the wires. Special shielding or routing techniques may be employed to minimize stray capacitance. One cross-talk mechanism is the stray capacitance between electrode input wires. This would couple signals from one electrode input to another. The stray capacitance between electrode wires and the electrode impedance form a high-pass filter between one input and another. Due to small capacitances and electrode resistances, the rolloff frequency would be relatively high. A 200,000 Hz high-pass filter would attenuate a 1000 Hz signal by 200:1. This is approximately the ratio seen when a 1000 Hz signal was applied in the simulation circuit (5 mV/23/xV).
476 Initially this was considered the primary mechanism because we had been using a traditional polygraph with 15 feet of closely spaced input wires. However, using an amplifying system with buffers in the electrode box did not eliminate the cross-talk. This indicated that cross-talk occurred at the level of the input leads in the electrode box. In the above example of 23/xV coupling at 1000 Hz, we would expect to see only 2.3/xV at 100 Hz and 0.23 p,V at 10 Hz. Instead the signal was 9 #V at 10 Hz. Obviously capacitive coupling is not the only cross-talk mechanism. Other possible cross-talk mechanisms include the power supply. Although not usually a problem, power supplies are not perfect. Millivolt level signals could produce microvolt level effects. Power supply filter capacitors' smoothing effect degrades as the frequency is lowered; coupling through the power supply voltages would increase as the frequency is reduced. This effect would also produce phase shift. Another mechanism is the common mode rejection ability of tile amplifiers. Differential amplifiers depend on interference signals being " c o m m o n " to both inputs. Differences in electrode impedance and differences in capacitance between conductors (wires at various distances from each other in a bundle) would cause interference to arrive at the inputs at different levels. These differences would then be amplified and appear at the output. The common mode rejection ratio of an amplifier also degrades with frequency. Cross-talk is usually very small. It only becomes important when the amplitude ratio between signals (such ,,s peripheral to cortical) becomes as large as the cross-talk ratio. The cross-talk mechanism(s) are not merely the result of using instruments of older design but can appear in modern instruments. Manufacturers may wish to investigate the exact cause(s) and provide remedies. If protocol requires acquisition of signals with widely varying amplitude (50:1 or greater), care must be taken in interpreting the results. The degree of cross-
T. NAGAOKA ET AL. talk will vary with the impedance of the electrodes, wave shape (frequency components) of the source potentials, amplitude of the source potentials and the design of the amplification system. Federation guidelines (Barlow et al. 1978) for electroencephalographs do state a cross-talk ratio of 100:1 or greater. This ratio is adequate for E E G recording where the signals are of similar amplitude and the frequency range limited. However, the peripheral potentials/cortical potential ratio easily exceeds this minimum. Additionally, evoked potential recordings utilize a wider bandwidth where cross-talk is accentuated. A guideline for crosstalk in evoked potential amplification systems should be considered. The effect of cross-talk can be minimized by attenuating the larger amplitude signals before they arrive at the electrode junction box. Maintaining low and balanced electrode impedances, which minimizes noise and external interference, will also reduce the effect of cross-talk.
References Barlow, J.S., Kamp, A., Morton, H.B., Ripoche, A., Shipton, H. and Tchavdarov, D.B. EEG instrumentation standards (revised 1977): Report of the committee on EEG instrumentation standards of the International Federation of Societies of Electroencephalograph and Clinical Neurophysiology. Electroenceph. clin. Neurophysiol., 1978, 45: 144-150. Grob, B. Stray capacitive and inductive effects. In: Basic Electronics. McGraw-Hill, New York, 1971: 435-438. Kimura, J., Yamada, T. and Kawamura, H. Central latencies of somatosensory evoked potentials. Arch. Neurol., 1978, 35: 683688. Onishi, H., Yamada, T., Saito, T., Emori, T., Fuchigami, T., Hasegawa, A., Nagoaka, T. and Ross, M. The effect of stimulus rate upon common peroneal, posterior tibial, and sural nerve somatosensoryevoked potentials. Neurology, 1991, 41: 1972-1977. Vas, G.A., Cracco, J.B. and Cracco, R.Q. Scalp recorded short latency cortical and subcortical somatosensory evoked potentials in peroneal nerve stimulation. Electroenceph. clin. Neurophysiol., 1981, 51: 1-8.
473
q) 1992 Elsevier Scientific Publishers Ireland, Ltd. 0168-5597/92/$05.00
EVOPOT 92008
"Cross-talk" in recording evoked potentials Takeshi Nagaoka *, D. David Walker, Peter J. Seaba and Thoru Yamada Dicision ()f' Clinical Eh'ctrophysiology, Department of Neurolog3', Unit:ersity of h)wa, Colh'ge ~f Medicine, Iowa City, IA 52242 (U.S.A.) (Accepted for publication: 17 July 1992)
Summary
When two potentials having large amplitude differences are simultaneously recorded, the large amplitude potential contaminates the small amplitude response. The small, early potentials generated by this contamination resemble far-field potentials. Although scalp-recorded SEP was contaminated by waves similar to the peripheral potential, peak latencies and wave form were not identical. Experiments simulating the recording situation verified the presence of "cross-talk." Capacitive coupling would shift peaks and alter the wave forms. Other possible mechanisms for the cross-talk and methods of minimizing it are offered. One should be cautious interpreting the results when potentials of large amplitude differences are simultaneously recorded. Key words: Somatosensory evoked potential; Stray capacitance; Contamination
After adding a channel of peripheral nerve action potential (NAP) or compound muscle action potential (CMAP) to our somatosensory evoked potential (SEP) recording protocol, we observed substantial alteration of SEP wave forms. The scalp-recorded SEP was contaminated by waves similar to the peripheral potential. Experiments electronically simulating the recording situation verified that the source of the "cross-talk" was in the amplification system. Frequency-dependent cross-talk pointed to capacitive coupling as the mechanism of "cross-talk." This report is to alert others to the problem we have experienced.
Material and method In t,ivo simulatkm Healthy volunteers (6 males) were used to record scalp SEP elicited by peroneal nerve stimulation at the knee. Stimulus electrodes were 7 mm diameter, flatsurfaced disks, placed over the right peroneal nerve at the knee, with the cathode 2 cm proximal to the anode. Stimuli of 0.1 msec duration were delivered at a rate of 3.1/sec via a stimulus isolation unit. The intensity was adjusted to elicit modest contractions of the peroneus
Correspondence to: Thoru Yamada, M.D., Division of Clinical Electrophysiology, Department of Neurology, University of Iowa, College of Medicine, Iowa City, IA 52242 (U.S.A.). * Present address: Department of Orthopaedic Surgery, Nihon University, Tokyo, Japan.
longus and brevis muscles, approximately 3 - 4 times sensory threshold. Recording electrodes were collodion-adhered Ag-AgCI cups with impedances less than 5 k~Q. A single channel SEP was recorded from Cz referenced to the right ear lobe. Peripheral potentials were recorded using similar recording electrodes and techniques. Recording electrodes were placed at the ankle overlying the superficial peroneal nerve, with an inter-electrode distance of 5 cm. The ground electrode was a cup electrode at the forehead (Fpz) or a strap electrode around the calf 10 cm proximal to the stimulus electrodes. The leads were connected to a commercial E E G instrument via a headboard containing preamplifiers. The amplifier bandpass was the same as used in clinical SEP recordings: 15-1500 Hz (all channels). Amplification was 2 × 105 for SEP and 104 for peripheral potential recording. 1500 responses were averaged. Electrical simulation To verify the "cross-talk" phenomenon, we experimented with a circuit commonly used to measure cross-talk. A 5 mV sine wave was applied to a pair of electrode inputs selected by one channel of the E E G instrument. Another pair of electrode inputs, selected by a second channel, were connected to ground by via 5 kS2 resistors (Fig. 1). The same electrode inputs that had been used in the SEP recordings were used. We measured the cross-talk appearing in the second channel while incrementing the sine-wave frequency from 10 to 1000 Hz. The amplification for sine wave input channel was 3 × 10 4 and 3 × 10 6 for the second (cross-talk) channel.
474
T. N A G A O K A E T A L
G2
I
~
A
CHANNEL n]
GI
B RG2 ~ 5K I R
5K
i
l ',
.... >
|
> EHANNF[
kJ
~!
[ I I
Fig. 1. Equivalent circuit for stray-capacitance coupling between inputs shown in dotted lines.
1
I t I f
II/~L
C
Results
SEPs from peroneal nerve stimulation consisted of an initial small positive peak with a latency range of 23.8-26.0 msec (mean = 25.1 msec), and subsequent large negative-positive-negative peaks. This was similar to previous peroneal SEP studies by others (Kimura et al. 1978; Vas et al. 1981; Onishi et al. 1991). Peripheral electrodes registered the polyphasic waves, an amalgam of nerve action potentials (NAP) and compound muscle action potential (CMAP) potentials, termed here the "peripheral potential." When the peripheral potential was recorded simultaneously with the scalp SEP, the SEP response showed polyphasic waves prior to the initial positive peak, with similar, but not identical, wave shapes or latencies as the peripheral potential (Fig. 2). This was observed in all subjects.
--F Z
/ 2mV
L
L
o
I
30
60ml~
Fig. 2. Peroneal SEP without recording of peripheral potentials shown in A. B is the contaminated SEP with concomitant recording of peripheral potential (C). Note the differences in wave forms and phase shift of contaminated waves from the source potential.
Amplitudes of peripheral potential recorded in the 6 subjects ranged from 0.9 to 3.6 mV (mean = 1.8 mV). These were measured from the highest peak to the lowest trough of the polyphasic waves. The ratio of cross-talk to source-potential amplitudes was roughly
A
B Attenuation
Peripheral
=r,o0
> 2mY
20 u V
+
C
Scalp SEP (C2-A
1)
-I + i
l
I
[
I
I
0
30
60
0
30
60msec
Fig. 3. The peripheral potential (A) was attenuated to approximately 1/100 (B) of the original amplitude. The contaminated waves before the attenuation (C) were diminished but not completely eliminated after the attenuation (D).
" C R O S S - T A L K " IN R E C O R D I N G EPs
475
from 1/500 to 1/1000. The ratio varied depending on the duration of source potential waves; slower waves showed less cross-talk than faster waves, as if the source potential was filtered by a high-pass characteristic. Separating or gathering the electrode wires from the scalp and from the leg did not measurably affect the cross-talk, nor did reducing the amplifier gains. Crosstalk remained when selecting electrode jacks that were widely spaced on the electrode board. Various "ground" sites were also investigated (leg and scalp) to eliminate the chance that large common mode peripheral potentials were overwhelming the amplifier. This did not affect the appearance of peripheral potentials in the cortical channels. To show that the cross-talk occurred after the electrode inputs, an attenuator was used. The different amplitude signals would still be present at the electrodes, in the wires, etc. However the peripheral potential presented to the electrode inputs would be reduced to the level of cortical potentials. The attenuation circuit had an attenuating factor of 1 : 100 with an input impedance similar to that of the electrode input. Series resistors of 10 M/2 each provide input resistance similar to the buffers 10 M ~ input impedance. Shunt resistors of 100.000 (100 K) resulted in a 1 : 102 ratio.
A
1
0
Isec
tj\,,/' 11
Chl
"
t
oo o
"°°
O° ~'~~°~'D.~7~l|°"oo°e°~°aD°ae°°°°o°°=n°°°°°°nae°°°aD= a°°ooo 5
I
Hz lOO
5oo
lOOO
Frequency of sine wave
Fig. 5. The relationship of contaminated wave amplitude with the frequency of source potential (2.5 mY). The amplitude of contaminated wave increased about linearly after the increase of source potential frequency greater than 300 Hz.
When a peripheral potential of about 2 mV was attenuated to 20/xV (1/100), the cross-talk was largely, but not completely, eliminated (Fig. 3). We then examined the electrically simulated crosstalk model. A 5 mV, 10 Hz sine wave was applied to channel 1. Channel 2, with inputs connected to ground via 5 kg2 resistors, recorded a 9/xV sine wave. The sine wave had nearly the same phase as the source potentials (Fig. 4A). When the frequency was increased to 1000 Hz, channel 2 registered a higher amplitude sine wave (23 #V), phase shifted from channel 1 (Fig. 4B). We also found that the cross-talk was much greater when unequal resistances were used to ground the channel 2 inputs; for example, 5 k£2 in " G I " ( - ) and 10 k ~ in " G 2 " ( + ) (Fig. 4C). Progressive increments of source frequency showed near-linear increase of contaminated sine wave amplitude especially when the source frequency was greater than 300 Hz (Fig. 5).
Discussion
-T
Oh2
+"
0
lOms©c
/
/ ',, j , /
t
I / k/
',
I]
~
/
C
pV
~
':'
"
-
7 ;
3mY
~1 ,/
,.
%'
\ Ch2
"\
0 [
_
.
_
.
z .
.
.
.
.
.
~
_o_ ._ ~ , 10~sec
Fig. 4. Contamination of 5 mV sine wave fed in channel l introduced in channel 2. The contaminated wave in channel 2 is greater with 1000 Hz (B) than with 10 Hz (A) sine wave. The contamination is much greater with the use of unequal registers, 5 k,¢2 in " G I " and 10 k.Q in " G 2 " (C). Note also the greater phase shift with the faster frequency sine wave.
Some capacitance exists between any two wires (Grob 1971). This "stray" capacitance is greater with closer proximity or with greater length of the wires. Special shielding or routing techniques may be employed to minimize stray capacitance. One cross-talk mechanism is the stray capacitance between electrode input wires. This would couple signals from one electrode input to another. The stray capacitance between electrode wires and the electrode impedance form a high-pass filter between one input and another. Due to small capacitances and electrode resistances, the rolloff frequency would be relatively high. A 200,000 Hz high-pass filter would attenuate a 1000 Hz signal by 200:1. This is approximately the ratio seen when a 1000 Hz signal was applied in the simulation circuit (5 mV/23/xV).
476 Initially this was considered the primary mechanism because we had been using a traditional polygraph with 15 feet of closely spaced input wires. However, using an amplifying system with buffers in the electrode box did not eliminate the cross-talk. This indicated that cross-talk occurred at the level of the input leads in the electrode box. In the above example of 23/xV coupling at 1000 Hz, we would expect to see only 2.3/xV at 100 Hz and 0.23 p,V at 10 Hz. Instead the signal was 9 #V at 10 Hz. Obviously capacitive coupling is not the only cross-talk mechanism. Other possible cross-talk mechanisms include the power supply. Although not usually a problem, power supplies are not perfect. Millivolt level signals could produce microvolt level effects. Power supply filter capacitors' smoothing effect degrades as the frequency is lowered; coupling through the power supply voltages would increase as the frequency is reduced. This effect would also produce phase shift. Another mechanism is the common mode rejection ability of tile amplifiers. Differential amplifiers depend on interference signals being " c o m m o n " to both inputs. Differences in electrode impedance and differences in capacitance between conductors (wires at various distances from each other in a bundle) would cause interference to arrive at the inputs at different levels. These differences would then be amplified and appear at the output. The common mode rejection ratio of an amplifier also degrades with frequency. Cross-talk is usually very small. It only becomes important when the amplitude ratio between signals (such ,,s peripheral to cortical) becomes as large as the cross-talk ratio. The cross-talk mechanism(s) are not merely the result of using instruments of older design but can appear in modern instruments. Manufacturers may wish to investigate the exact cause(s) and provide remedies. If protocol requires acquisition of signals with widely varying amplitude (50:1 or greater), care must be taken in interpreting the results. The degree of cross-
T. NAGAOKA ET AL. talk will vary with the impedance of the electrodes, wave shape (frequency components) of the source potentials, amplitude of the source potentials and the design of the amplification system. Federation guidelines (Barlow et al. 1978) for electroencephalographs do state a cross-talk ratio of 100:1 or greater. This ratio is adequate for E E G recording where the signals are of similar amplitude and the frequency range limited. However, the peripheral potentials/cortical potential ratio easily exceeds this minimum. Additionally, evoked potential recordings utilize a wider bandwidth where cross-talk is accentuated. A guideline for crosstalk in evoked potential amplification systems should be considered. The effect of cross-talk can be minimized by attenuating the larger amplitude signals before they arrive at the electrode junction box. Maintaining low and balanced electrode impedances, which minimizes noise and external interference, will also reduce the effect of cross-talk.
References Barlow, J.S., Kamp, A., Morton, H.B., Ripoche, A., Shipton, H. and Tchavdarov, D.B. EEG instrumentation standards (revised 1977): Report of the committee on EEG instrumentation standards of the International Federation of Societies of Electroencephalograph and Clinical Neurophysiology. Electroenceph. clin. Neurophysiol., 1978, 45: 144-150. Grob, B. Stray capacitive and inductive effects. In: Basic Electronics. McGraw-Hill, New York, 1971: 435-438. Kimura, J., Yamada, T. and Kawamura, H. Central latencies of somatosensory evoked potentials. Arch. Neurol., 1978, 35: 683688. Onishi, H., Yamada, T., Saito, T., Emori, T., Fuchigami, T., Hasegawa, A., Nagoaka, T. and Ross, M. The effect of stimulus rate upon common peroneal, posterior tibial, and sural nerve somatosensoryevoked potentials. Neurology, 1991, 41: 1972-1977. Vas, G.A., Cracco, J.B. and Cracco, R.Q. Scalp recorded short latency cortical and subcortical somatosensory evoked potentials in peroneal nerve stimulation. Electroenceph. clin. Neurophysiol., 1981, 51: 1-8.
