Four major tasks were completed in this project: (i) Hardware design and development, (ii) control software development, (iii) reliability testing of two electrode types and (iv) pilot testing skin impedance at two APs and a nearby non-AP site.
Considerations for Hardware Design and Development
Most traditional methods for measuring skin impedance have used a current source of 1–25 μA (micro-Ampere), although Nakatani (2
), when developing his Ryodoraku method, used a moist electrode that applied a current of ~200 μA. (It should be noted, however, that the Committee on Electrocardiology recommends currents of 10 μA or less for patient-connected leads (33
)). The voltage is measured across the skin under test. Although a constant current source works well when impedances are <1 MΩ, in our previous research we found that skin impedance at APs during sleep can be of the order of 50–60 MΩ (34
). Using the traditional approach would require a constant current source capable of developing over 50 V across the skin patch, an impractical solution. This led to the development of our present constant voltage measurement system.
Since most of the impedance is determined by the resistance value, we used voltage amplifiers to accommodate the three orders of magnitude in resistance measurements (from 50 KΩ to 50 MΩ). We assume that the skin impedance is a parallel resistor and capacitor network as shown in . A block diagram of the system to measure the values of the components is also shown in . Amplifiers one and two have fixed gains, but amplifier three requires variable gain.
Block diagram of measurement system. The ADC1 and ADC3 signals are used to evaluate the resistance value. ADC2 is used to determine the capacitance value after the resistance value has been determined.
We assembled our system with off-the-shelf components and used it to develop the basic algorithm for resistance and capacitance measurements. The generator is an Agilent 33220A Function Generator, Agilent Technologies, Palo Alto, CA, USA. The basic acquisition system, an OMB Daqboard 2000, IoTech, Stamford, CT, USA provides for 16 channels of acquisition and interfaces to a personal computer. Our device collects only a single channel of data since the goal at this prototype development stage was to determine if accurate, reliable, near-continuous skin impedance data could be collected. Our final system will be fully automated for continuous data collection form eight individual channels. Near continuous data collection was achievable by moving the connector to multiple sites on a universal breadboard sequentially. For each skin site measured, only one electrode needed to be connected since the reference electrode is fixed. Three UFI2122i bioamplifiers with programmable gains (UFI, Morro Bay, CA, USA) are incorporated in the circuit as shown above. Amplifiers one and three, connected to the ADC1 and ADC3 inputs of the Daqboard, are used to determine the resistance of the load.
A square wave of period 0.5 s with 50% duty cycle is generated for the impedance measurements, yielding, which is for all practical purposes a DC measurement. Other authors have used AC measurement approaches with square waves of higher frequency (12
). An AC measurement yields a combined impedance of the resistor and capacitor at the square wave frequency; however, a disadvantage of an AC measurement system is that the signal must be continuously applied for the duration of the measurement. With a DC step measurement, the signal is applied only for the duration of the ON cycle, with no stimulation of the skin between cycles.
With our DC measurement, in the first phase only the resistance value of the skin is measured. To obtain this value, we can neglect the 50 Ω internal resistance of the pulse generator given that the amplifier input impedance of 10 MΩ and the 100 KΩ resistance Ri
are much larger than the 50 Ω. After reaching an essentially steady state (about 400 ms), the skin resistance Rskin
is given by Equation (1
) where V1
are the steady state voltages measured in the ADC1 and ADC3 of the Daqboard. V1
can be considered fixed at 5 V. Time averaging is used to reduce the measurement noise in V3
Capacitance is measured in the second phase. Amplifier two, connected to the ADC2 port of the Daqboard, is switched in. Since the skin capacitance to be measured ranges from a few to hundreds of nanoFarads (nF), we can neglect the amplifier capacitance, estimated to be on the order of 100 pF. The 10 MΩ input of the amplifier cannot, however, be neglected at high values of skin resistance and must be taken into account.
The RC (Resistance/capacitance) time constant of the equivalent circuit for capacitance measurement is given by (Rskin//Ramp2//Ramp3//Ri) Cskin, where the Ramp2 is the input resistance of amplifier two and Ramp3 is the input resistance of amplifier three and notation R1//R2 is used to denote two resistances in parallel. For medium values of skin resistance, 50 KΩ − 1 MΩ, the time constant is determined by the skin capacitance and the 100 KΩ current measurement resistance, Ri, in parallel with the skin resistance. For large values of skin resistance, the time constant is determined by the skin capacitance and the 100 KΩ Ri. Since the input resistance of the amplifier can be comparable with the skin resistance and thus, complicate the measurement of the skin resistance, it is necessary to switch the second amplifier off during the first phase of the resistance measurement.
Assuming that the rise time of the voltage measured at amplifier two is a simple exponential, the 10–90% rise time is related to the RC time constant through Equation (2
It is interesting to note that, with our constant voltage (rather than constant current) approach, the skin resistance is not a limiting factor in the calculation of the RC time constant. The time constant is Cskin*100 KΩ for large values of Rskin, such as 50 MΩ. With a 100 nF capacitance, the time constant is only 10 ms. Had a current source approach been used, the time constant would have been 5 s, for 100 nF and 50 MΩ. With a 10 ms time constant, the corresponding rise time is on the order of 22 ms. Thus, as far as the measurements are concerned, we could have used a square wave up to 10 Hz to obtain the same result. In a final implementation, we will keep the 2 Hz rate but reduce the duty cycle to 15–20%. This will allow longer than 400 ms between skin stimulations.
It should also be noted that we chose to measure the voltage across the load for the capacitance measurement. An alternative is to use the voltage across Ri since amplifier three is already connected. Then we would measure a 90–10% fall time rather than a rise time. The problem with this approach is that our amplifier bandwidth does not allow us to measure the initial voltage surge. As a result, the 100% reference level cannot be determined accurately. With a rise time measurement, the amplifier bandwidth is not a limiting factor.
Control Software Development
The control software was developed using Labview version 7.0 under Microsoft XP Professional edition running on a Pentium 4 processor. The software program acquires the data and then calculates resistance and capacitance using the equations above. It then saves and stores the data in a text file with a unique ID code. To begin impedance measurements, the operator selects the channel, the sampling frequency and the file name in which the data will be saved. The program automatically acquires data from the DAQ board. The program calculates the average value and displays a graph. The operator visually observes the waveforms on the computer screen and saves the averaged data as soon as the waveform has stabilized (~6–10 s after the current has been injected). The maximum and minimum voltage values are captured in resistance Ri and calculated as an average of the previous 5–10 resistance readings. In the next step the operator can exit or save the data to the file. The file includes the date, the average skin resistance value and the average skin capacitance value.
Preliminary testing in three separate experiments was performed on eight volunteers (aged 27–62 years). The research protocols for each experiment were conducted according to the ethical standards in the Declaration of Helsinki (37
). In Experiment no. 1, we compared the repeatability of skin impedance measurements recorded with two electrodes in each of eight volunteers over 30 min. In Experiment no. 2 capacitance and resistance measurements were tracked simultaneously in a single volunteer. In Experiment no. 3 involving four volunteers, skin impedance at two APs and a non-AP site was measured continuously over 60 min.
Reliability Testing Of Electrodes
To circumvent the problems of probe application (pressure, duration and inclination of the probe on the skin), the electrodes were positioned flatly on the skin and securely affixed with a clear adhesive wrapped circumferentially around the forearm. Electrodes remained in place for 15 min prior to recording and were then left in place, undisturbed, during 60 min recording. The pressure exerted by the undisturbed electrodes was not measured as it was presumed to remain constant during the duration of the recordings.
Two electrodes were compared for reliability of measurements on the volar surface of the forearm. The ORI electrode* developed by Orbital Research Inc. Cleveland, OH, USA for long term use in electrocardiography (ECG) is a dry, 2.4 cm diameter electrode with tiny micro anchors on the surface in contact with the skin (38
). The ECG version of these electrodes was FDA approved in December 2006 (FDA 510k#K062760). With permission from Orbital Research, Inc. we reduced the size of this electrode by grinding down the platform to a 1.1 cm diameter and coated the exposed surface with conductive epoxy (). This modification does not affect the sensing element and or skin contact portion of the device. The other electrode, a 7 mm2
stick-on Ag-AgCl gel electrode covered with conductive foil is used with the AMI (Apparatus for measuring the functioning of Meridians and their corresponding Internal organs device). Our reference electrode an Ag/AgCl conductive adhesive ECG electrode (MediTrace 530) with a diameter of 2.4 cm was placed ~8 cm proximal to the recording electrodes on the volar surface of the ipsilateral forearm. Skin resistance at two adjacent sites on the volar surface of the forearm was recorded over a 30 min period in eight volunteers. The between- and within-components of variance were computed across the eight participants using a standard method (39
). The reliability of the measurements taken with each electrode was estimated as the ratio of the between person variance to the total variance.
Dimensions of AMI and modified ORI electrodes.
Comparing Impedance at Two APs and a Non-AP Site
Two APs (LU 9 and PC 6) were compared with a nearby non-AP site (). APs were located using standard acupuncture charts, anatomical landmarks and digital palpation. The non-AP site was located at a point midway on an imaginary line from wrist to elbow that was equidistant between the two APs.
Modified ORI dry electrodes on LU 9 and PC 6 acupuncture points and a nearby non-AP site.
The skin at the selected sites was cleansed with ethyl alcohol and allowed to dry prior to electrode application. Electrodes were affixed to the skin with clear tape and left in place undisturbed for the duration of the recordings. We assumed the pressure of the electrode on the skin was constant during the recordings as the electrodes were not touched during the course of the experiments.
All volunteers were tested in the Psychophysiology Laboratory at the Helfgott Research Institute which was kept at a constant temperature of 69 and 70°F. Testing occurred between 9 am and 4 pm on 3 days.