Electrode and leakage current testing in an EEG monitor with an implantable part

Abstract

A personal wearable EEG monitor comprising an implantable electrode part with at least two electrodes (2,3) for measuring an EEG signal of a person. The electrode part comprises an electronic circuit arranged in a housing (1) with each electrode arranged external to the housing. The electrode part comprises a testing circuit for testing functionality of the electrode part. The testing circuit comprises a capacitor (9) coupled in serial connection to at least one of the electrodes, and a test signal generator for providing a test signal. The EEG monitor is adapted for analyzing the signal resulting from the signal generator for identification of faults in the electrode part. The invention further provides a method for detecting a leak current in an implanted EEG monitor part.

Claims

1. A personal wearable EEG monitor comprising an implantable electrode part with at least two electrodes adapted for measuring an EEG signal of a person, said electrode part comprising an electronic circuit arranged in a housing with each electrode arranged external to said housing, said electronic circuit being adapted for receiving an analogue EEG signal from said electrodes and being provided with an analogue to digital converter for converting the analogue EEG signal into a digital signal, said electrode part comprising a testing circuit for testing for faults in said electrode part, wherein said testing circuit comprises a capacitor coupled in serial connection to at least one of said electrodes, and a test signal generator for providing a test signal, said test signal generator being adapted for being coupled between said electrodes and said electronic circuit during a testing period, wherein said EEG monitor is adapted for analyzing the test signal from said signal generator for identification of faults in the electrode part.

2. The EEG monitor according to claim 1, wherein said test signal generator is arranged between said electrodes and an input of said analogue to digital converter, and wherein said EEG monitor is adapted for analyzing a signal at said input of the analogue to digital converter for identification of faults in the electrode part.

3. The EEG monitor according to claim 1, wherein an output, signal from said analogue to digital converter is transferred to a digital signal processing unit arranged in the electrode part.

4. The EEG monitor according to claim 1, wherein said capacitor is coupled in serial connection to the input of said analogue to digital converter.

5. The EEG monitor according to claim 1, adapted for identifying at least one of a charging and a discharging function at an input of the analogue to digital converter and thereby detecting a current leak.

6. The EEG monitor according to claim 1, wherein a resistor is arranged across an input of the analogue to digital converter.

7. The EEG monitor according to claim 1, wherein the implantable electrode part comprises a coil for an inductive coupling to a non-implantable part, said inductive coupling being adapted for transfer of data and power.

8. The EEG monitor according to claim 1, wherein said test signal generator provides a square wave test signal with frequency between 25 and 40 Hz.

9. A method for detecting a leak current in an implanted EEG monitor according to claim 1, comprising the steps of providing said test signal, subtracting an estimate of said test signal from an output of the analogue to digital converter to obtain a resulting signal, low-pass filtering the resulting signal in order to suppress EEG and noise components, transforming the low-pass filtered signal into a logarithmic domain, fitting the transformed signal to a straight line, determining an interception with y-axis of the straight line, and determining if a leak current is present based at least in part on the determined interception.

10. The method according to claim 9, wherein said low-pass filtered signal is down-sampled in order to reduce the calculation complexity.

11. The method according to claim 9, wherein said step of transforming the low-pass filtered signal comprises determining the power of the low-pass filtered signal and taking the natural logarithm of the power of the low-pass filtered signal.

12. The method according to claim 9, wherein said step of transforming the low-pass filtered signal comprises determining the absolute value of the low-pass filtered signal and taking the natural logarithm of the absolute value.

13. The EEG monitor according to claim 2, wherein said capacitor is coupled in series between said at least one electrode and an input of said analog-to-digital converter, and said test signal generator is coupled to supply said test signal to another input of said analog-to-digital converter.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) Embodiments of the invention will now be explained in further detail with reference to the figures.

(2) FIG. 1 illustrates an embodiment of an implantable part of an EEG monitor provided with a test generator.

(3) FIG. 2 illustrates an implantable part with two leak current routes.

(4) FIG. 3 illustrates the implantable part from FIG. 1 with the test generator grounded.

(5) FIG. 4 illustrates the leak current route from the implantable part in FIG. 3.

(6) FIG. 5 illustrates the implantable part from FIG. 3 in more detail and with a capacitor between electrodes and analogue to digital converter (ADC).

(7) FIG. 6 illustrates the implantable part from FIG. 5 with a resistor on the ADC input.

(8) FIG. 7 illustrates the implantable part from FIG. 6 with a possible implementation of a test circuit shown.

(9) FIG. 8 illustrates a simulation of the ADC voltage when providing a test signal while no leak current is present.

(10) FIG. 9 illustrates a simulation of the ADC voltage when providing a test signal while a leak current is present.

DETAILED DESCRIPTION OF THE INVENTION

(11) FIG. 1 shows an implantable part of an EEG monitor, or simply an implant device, with a housing 1 and with implant terminals 4, 5 connected to the electrodes 2, 3, which are in direct contact with the tissue 20. There will be some contact impedance 21 denoted R.sub.t between the tissue and the electrodes. The EEG potential 22 denoted V.sub.g will be picked up by the electrodes 2, 3. The implant housing 1 will comprise an implant electronic module 6 having an analogue to digital converter (ADC) 10, for converting the analogue EEG signal into a digital signal, and a digital signal processing and microcontroller unit (DSP) 11.

(12) The DSP 11 may prepare the EEG signal for transmission to an external non-implanted part of the EEG monitor, so that the further analysis of the EEG signal can be performed there. The DSP 11 may also perform the necessary analysis of the EEG signal in order to identify an imminent seizure or attack. The DSP 11 may also comprise a receiver and transmitter system (Rx-Tx) for communication with the surroundings.

(13) The housing 1 of the implant device also comprises a power supply 7. This may be in the form of a battery, e.g. a rechargeable battery. Other types of voltage supply could be any type of power generating means, such as an inductive coupling between a coil in the implant and a non-implanted coil arranged outside the skin and geometrically aligned to maximize the transfer of power to the implanted coil. Typically, the receiver and transmitter system will also apply such an inductively coupled pair of coils for communication between the internal, i.e. implanted, and the external parts of the EEG monitor. Preferably, the same set of coils can thus be applied both for the transfer of power and for the transfer of data.

(14) In FIG. 1 a test signal generator or voltage generator 8 is inserted between the implant terminal 5 and the ADC 10. This generator may supply a sine wave, a square wave or any other signal. The signal from the generator 8 is applied for determining if there is conductivity between the housing 1 for the implant and the electrodes. The frequency of the signal should preferably be within the bandwidth of the ADC.

(15) The generator 8 could be inserted during start-up of the monitor or at predefined intervals.

(16) FIG. 2 shows how possible leak currents may flow during monitoring of the EEG signal. The leak current will flow if there is any leak in the housing 1. Two possible leak paths 30, 31 exist.

(17) In differential measurement both leak paths 30, 31 must be present before there is a risk that a DC current can go through tissue. The risk of having two leak paths should be small.

(18) In single-ended measurements only one leak path is necessary in order to have a leak DC current go through tissue. The setup of the test voltage generator for this situation is shown in FIG. 3, where the signal generator 8 together with the lower input to the ADC 10 has been grounded by the connection 13. With this set-up there will be no leak current along the lower leak path 31 in FIG. 2.

(19) With the set-up of the test generator 8 in FIG. 3 it will be possible to see if the electrodes are broken, i.e. if there is no connection through the electrodes.

(20) FIG. 4 shows the possible leak path 30 to the implant terminal 4 for a single ended measuring system. This leak path may result in a leak current I.sub.CL through tissue. In the embodiment of FIG. 4 the terminal 5 is grounded. A leak path 14 to the lower implant terminal 5 will not result in current through the tissue surrounding the implant part.

(21) In FIG. 5 the control of the test generator 8 is shown in the case of single-ended EEG measurement. A possible leak path here could be from the positive pole of the power supply 7 and to the implant terminal 4. The frequency of the generator 8 shall be in the range of the ADC 10, i.e. below half the sample rate of the ADC. The wave form of the generator 8 can for example be square or sine.

(22) The switch 16 is controlled from the DSP unit 11. In the position of the switch 16 in FIG. 5 the test generator 8 is de-coupled or not active, and the implant will be in EEG monitoring mode. When the position of the switch 16 is changed to couple the test generator 8 into the circuit it is possible to measure if the electrode is broken and to measure if there is a leak current into the tissue surrounding the implant.

(23) In order to test for a broken electrode in the test circuit in FIG. 5, the output signal of the ADC 10 could be band pass filtered (e.g. around 30 Hz) in order to remove EEG signal. If a signal originating from the generator 8 can be seen in the output from the band pass filter (which is part of the DSP 11) then the electrode is not broken. Otherwise the electrode might be broken.

(24) In order to test for a leak current from the implant to the body tissue, the output of the ADC 10 could be low pass filtered, e.g. at a cutoff frequency at approximately 10 Hz. At the time t=0, e.g. when the implant is started up, the absolute value of the low pass filtered ADC output voltage is measured, and if this is greater than a preset threshold value (e.g. 1 mV), then there is a leak current through body tissue.

(25) The circuit of FIG. 5 also comprises a capacitor 9 in serial connection between the ADC 10 input and an EEG electrode 2. This capacitor is applied for the leak current testing. At the same time the capacitor 9 offers protection against current leak, since it will block any DC component from the electronic circuit 12 from entering the electrodes. Also the capacitor may block a DC component of from the electrodes from entering the ADC 10.

(26) In FIG. 6 a resistor 18 is arranged across the input of the ADC 10. This resistor is in serial connection with a switch 17 for switching the resistor 18 into the circuit. Switching the resistor 18 into the circuit is relevant when the test generator 8 is active. Then it is possible to estimate electrode impedance R.sub.t 21. The magnitude of the electrode impedance 21 is determined by how good the contact is between the tissue and the electrodes 2, 3.

(27) FIG. 7 shows an example of how the test generator 8 and the two switches 16, 17 of FIG. 6 might be implemented. The test generator is implemented through an AND gate 24 and a resistor 26. The AND gate is controlled by one signal line 33 providing a control signal from the DSP 11 sending a 1 (i.e. being high) when the test generator 8 needs to be active, and by one signal line 34 providing the test signal, here in the form of a square waveform.

(28) Both switches 16, 17 are implemented as transistors, which are also controlled by the signal line 33. The switch 16 is connected to the signal line 33 through a NOT gate 25 in order for this switch 16 to be open when the test is running. When the switch 16 is closed the resistor 15 is short-circuited. During testing the switch 17 receives a 1 or high through the signal line 33, and is thereby closed such that the resistor 18 is connected between the two input terminals of the ADC 10.

(29) In the following an example of how a broken electrode could be detected is given.

(30) Referring to FIG. 5, the ADC 10 input voltage V.sub.ADC is a high-pass filtered version of the sum of the EEG signal 22 and the signal from the test generator 8 assuming switch 16 is in lower position. In the present example test signal 8 is squared and has a frequency at 34.5 Hz and has an amplitude of about 1 mV. The test signal could e.g. be applied for the first second after the implanted device is powered up. Since EEG signals typically will have an amplitude of 10-100 V and a frequency range of 0.5-20 Hz the test signal will be clearly distinct from the EEG signal. This can easily be detected by the following signal processing.

(31) There are several methods to detect or calculate the 34.5 Hz signal at the ADC input. One method is to calculate the Discrete Fourier Transform (DFT) at 34.5 Hz. The numerical value of the DFT is then compared to a preset threshold, and if the DFT is above it, it is assumed that the electrode is intact.

(32) Furthermore, from the output of ADC 10 it can be checked if the DC blocking capacitor 9 is intact. Normally there will be no DC contribution at the output of ADC 10 due to the high-pass filtering established by capacitor 9 and the input impedance of ADC 10. If for instance capacitor 9 goes from normal state to a shorted state, a DC contribution at the output of the ADC 10 will appear. This is due to the half-cell potential produced by the electrode in combination with body tissue. Such a DC can easily be detected by the following signal processing and can result in a warning or an alarm to the user. Otherwise, if the capacitor 9 blocks, no test signal or EEG signal will be seen at the output if the ADC 10. This again is easy to detect by the following signal processing and can result in a warning or an alarm to the user. This analysis is not necessarily depending on the test generator 8, and therefore, constant monitoring of the DC blocking capacitor condition is possible.

(33) Reference is again made to FIG. 5, now with the existence of a current path from the positive terminal of power supply 7 to the wire that connects to terminal 4 or the terminal 4 itself. In case the power supply 7 is an inductive powered voltage supply, there will be no voltages or currents in the implant when the external power transmitting system is turned off. As a consequence no current leak will run either. At the time when an external power transmitting system is turned on, the implant device 1 starts to harvest energy from the magnetic field. At the same time a current starts to run into the electrode due to the current leakage path. The leakage current is denoted I.sub.CL. If present, this current will generate a voltage over resistor 21 and resistor 15, assuming switch 16 is in lower position. The voltage at the output of ADC 10 which originates from the current leak is denoted V.sub.CL and has the form:
V.sub.CL=.Math.e.sup.(t/)
The time constant is
=C(R.sub.t+R.sub.g+R.sub.ADC)C.Math.R.sub.ADC

(34) The approximation holds when R.sub.ADC is much larger than R.sub.g and R.sub.t. C is the capacitor 9, R.sub.g is the impedance 15 of the test generator 8, R.sub.t is the impedance 21 between tissue and electrode, and R.sub.ADC is the input impedance (not shown) of the ADC 10.

(35) When starting up from an un-powered state the typical scenario will be that switch 16 is in the lower position. In this case a is given by

(36) $=\frac{I_{\mathrm{CL}}\left(R_t+R_g\right)R_{\mathrm{ADC}}}{R_{\mathrm{ADC}}+R_t+R_g}{I}_{\mathrm{CL}}\left(R_t+R_g\right)$

(37) When the test period has elapsed, switch 16 goes from the lower position to the upper and result in an value and a time constant given by:

(38) $=\frac{\left(I_{\mathrm{CL}}.Math.R_t-I_{\mathrm{CL}}\left(R_g+R_t\right)\right).Math.R_{\mathrm{ADC}}}{R_{\mathrm{ADC}}+R_t}=\frac{-I_{\mathrm{CL}}.Math.R_g.Math.R_{\mathrm{ADC}}}{R_{\mathrm{ADC}}+R_t}-I_{\mathrm{CL}}.Math.R_g$
=C(R.sub.t=R.sub.ADC)C.Math.R.sub.ADC

(39) V.sub.CL follows a charge function when the implant is turned on and a discharge function when the test period has elapsed.

(40) FIG. 8 shows a circuit simulation of V.sub.ADC when there is no leak current. This is shown over two sequences, where the test generator is on for the first 2 seconds and subsequently turned off. The unit on the second axis is Volts. The EEG signal is also seen on the V.sub.ADC signal.

(41) FIG. 9 shows a circuit simulation of V.sub.ADC when there is a leak current. The difference between FIGS. 8 and 9 is the curve from the charging/discharging of the capacitor 9 caused by the leak current and the resistors 21 and 15.

(42) In order to decide whether there is a leak current or not, and to determine the value of I.sub.CL, an algorithm is established.

(43) The value of is proportional to the leakage current I.sub.CL. A good estimate of a is found by the value of V.sub.ADC at t=0 and at the time where the test generator is switched off. To find I.sub.CL the V.sub.ADC value at t=0 is divided by (R.sub.t+R.sub.g) or, alternatively, the V.sub.ADC value at switch off is divided by Rg. These estimates can be improved by taking the average of the two, thereby reducing the variance of the estimate. If the estimate is greater than e.g. 1 A a warning or an alarm can be given to the user, or the device can simply be turned off.

(44) A better estimate can be found by looking at the whole progression of the test sequence since more samples are taken into account. Thereby the estimate will be less influenced by EEG or noise. Several methods exist to do this kind of estimation. One method could be taking the natural logarithm in FIG. 9 whereby a straight line is achieved. and can then be determined from the slope. In practice, however, V.sub.ADC is not always above zero, and in order to avoid complex calculations of the logarithm due to negative numbers, the power or the absolute value of V.sub.ADC is calculated before calculating the logarithm. Taking the logarithm and the power of a signal is computational complex; therefore, the signal may be down-sampled by a factor of e.g. four prior to the calculation. This can be summarized in the following steps: 1. Subtract the estimated test generator signal. 2. Low-pass filter, to suppress EEG and Noise components. 3. Downsample with e.g. 4 to reduce the calculation complexity. 4. Take the power of the signal downsampled signal 5. Take the natural logarithm of the power signal. In the log domain, the tendency curve for no leak current is a horizontal line. In case of leak current the tendency curve is a line with a negative interception with the y-axis and a negative slope. 6. Fit the above signal to a straight line and find the slope. is now the interception with the y-axis and is the slope of the line 7. Determine the leak current based on . If the estimate is greater than e.g. 1 A a warning or an alarm can be given to the user, or the device can simply be turned off.

(45) Two values can be determined by the above algorithm, one for start of the test generator and the other for switching it off. Each gives an estimate of I.sub.CL. Taking the mean value of the two results in an estimate of I.sub.CL with lower uncertainty. However, for computational complexity reason, it is beneficial to calculate only the latter , since it doesn't include the test generator signal.

(46) The testing circuit further enables calculation of the impedance between the electrodes and the tissue R.sub.t. This can be done by enabling a shunt resistor 18 R.sub.shunt on the ADC input. This shunt resistor in the circuit will change the calculation of V.sub.ADC to approximately:

(47) $V_{\mathrm{ADC}}=V_8\frac{R_{\mathrm{shunt}}}{R_g+R_t+R_{\mathrm{shunt}}}$
where V.sub.g is the signal from test generator 8. Furthermore it is assumed that R.sub.ADC is much larger than R.sub.g, R.sub.t and R.sub.shunt. From this equation R.sub.t can be estimated when measuring at the frequency of the test generator, where other voltage contributions might be ignored.