Novel Ultra-long Term EEG Monitoring System

A Possible Enabler of BCI Technologies

Jonas Duun-Henriksen and Sirin W. Gangstad

HypoSafe A/S, Kgs. Lyngby, Denmark

1 OBJECTIVES

Various groups have recently sketched the future

within portable EEG equipment (Brunner et al.,

2015; Casson et al., 2010). They agree that small,

portable and convenient systems for instant and

continuous EEG monitoring are essential.

We therefore present a novel subcutaneous

monitoring system developed for unobtrusive,

continuous, ultra-long term EEG applications.

Evidence of high signal quality is provided for two

patients monitored continuously night and day for a

1 month period.

2 METHODS

The recording system is described previously in

(Duun-Henriksen et al., 2015; Elsborg et al., 2011)

and preliminary results based on the same dataset as

the current is presented at The 15

European

Congress on Clinical Neurophysiology, October

2015 (Duun-Henriksen et al., 2015).

The system consists of an implantable part and

an external part, see

Figure 1

. The implantable part of

the device consists of an insulated lead with three

embedded platinum-iridium electrodes located

30mm apart. The middle electrode is the reference to

each of the other electrodes, thus providing two

channels. The lead is fixed to the implant housing,

which contains a coil for wireless transmission of

power and brain signals to an outer device. The

outer device consists of a small shell containing a

coil and electronic components for online data

analysis. It is furthermore possible to connect a

memory unit if up to 30 days of EEG recording,

sampled at 207 Hz, should be stored. The

implantation and explantation of the subcutaneous

device is carried out under local anaesthesia and

takes approximately 15 minutes. After implantation

the electrodes do not move.

Data were collected from two healthy volunteers

for up to 45 days. The project was approved by the

regional ethical committee and the Danish Health

and Medicines Authority. All subjects have given

informed written consent before being enrolled in

the trial. The implantable device was placed with the

housing right behind the ear and the lead pointing a

little posterior to vertex. Subjects wore the system

day and night, except while taking a shower or doing

water sports.

To investigate the signal quality longitudinally,

the signal was forward and reverse filtered into

standard physiological frequency bands (δ: 0-4 Hz,

θ: 4-8 Hz, α: 8-13 Hz, β

low

: 13-20 Hz, and β

high

: 20-

30 Hz). Filters were 10

order IIR bandpass filters.

Then the average power of each frequency band in

10 seconds, non-overlapping windows were

computed. The average power values were then

further split into 30 minute intervals, from which the

percentile power for each frequency band were

extracted.

For the statistical analysis, a linear model was

fitted for each frequency band with time, day/night

and patient as independent variables. The band

power was log-transformed prior to the modelling.

Figure 1: Experimental setup. (1A) Outer device for power

and signal transmission to and from the implantable part.

(1B) External memory unit to store long-term EEG

recordings. (3) Implantable part with lead and implant

housing.

Duun-Henriksen, J. and Gangstad, S..

Novel Ultra-long Term EEG Monitoring System - {{\}it A Possible Enabler of BCI Technologies}.

Figure 2: 20

percentile power in different frequency bands during a 45 days period for subject 1. The power does not seem

to change over time for neither day nor night recordings.

Figure 3: Same as figure 2 but for subject 2. The same tendencies regarding differences in night and day as well as no trend

over time are seen for both subjects.

3 RESULTS

Figure 2 and Figure 3 show the 20th percentile of the

average power in the five standard physiological

frequency bands as described in Methods.

The figures illustrate a difference between night and

day recordings in the θ, β

low, and βhigh frequency

bands. This is also validated in the statistical

analysis with p-values<0.001. Although a clear

difference in the δ-band is not as easily seen in the

figures, the statistical analysis showed a significant

difference in this band too.

Looking at the data points as a function of time,

there is no visual trend that implies that the

distribution should change over time. For both

subjects, the variance in the θ-band is higher during

night recordings, but both the mean and the standard

deviation are consistent over time.

4 DISCUSSION

4.1 Data Quality

In a previous publication we showed that the EEG

data quality of subcutaneous measurements made 10

days after implantation of the implantable device is

comparable to the data quality of standard scalp

EEG (Duun-Henriksen et al., 2015). The present

analysis has shown that the 20

th percentile of the

average power in the five standard frequency bands

do not change over a time course of ~ 1 month. This

indicates that the good data quality documented ten

days after implantation can be expected throughout

the ultra-long term, subcutaneous EEG

measurement.

4.1.1 20th Percentile

The use of the 20th percentile as a measure of the

approximate power during a 30 min long period has

some advantages as well as pitfalls. The advantage is

that especially during day, a decent amount of

artefacts are seen. Extracting the 20

th percentile

eliminates those artefacts. Investigating the amount

of artefacts in data was out of the scope of this

extended abstract, but the number is well above the

th percentile. The pitfall is that data are not

stationary, thus eliminating states that are evident for

less than 20% of the time. Such a state is for

example the deep sleep stage 3 with pronounced

high amplitudes of delta sleep. As this state only

constitutes approximately 15% of the time for

normal adults during sleep, this will not be included

in the 20

th percentile. This is probably also the

reason that the power in the delta-band is actually

higher during day recordings than night recordings

in this comparison.

4.1.2 Beta-power Declines during Night

It was seen that for both βlow, and βhigh frequency

bands the power was significantly lower during

night recordings. Whether this is due to less β-

activity during the night or simply less muscle

activity during the day is impossible to say with the

current analysis. Further investigation is needed.

4.2 The Use of the Device for BCI

We believe that the results support the statement that

the novel device is feasible for ultra-long term EEG

monitor that can be used for applications within BCI

technologies where continuous and instantaneous

measurements of the EEG is needed.

ACKNOWLEDGEMENTS

We would like to thank The Danish Council for

Strategic Research for funding of research leading to

this paper.

DECLARATION OF INTEREST

Jonas Duun-Henriksen and Sirin W. Gangstad are

full time employed at Hypo-Safe A/S developing

and producing devices for unobtrusive subcutaneous

EEG monitoring.

REFERENCES

Brunner, C., Birbaumer, N., Blankertz, B., Guger, C.,

Kübler, A., Mattia, D., … Müller-Putz, G. R. (2015).

BNCI Horizon 2020: towards a roadmap for the BCI

community. Brain-Computer Interfaces, 1–10.

doi:10.1080/2326263X.2015.1008956

Casson, A. J., Yates, D. C., Smith, S. J. M., Duncan, J. S.,

& Rodriguez-Villegas, E. (2010). Wearable

electroencephalography. IEEE Engineering in

Medicine and Biology Magazine, 29(3), 44–56.

doi:10.1109/MEMB.2010.936545

Duun-Henriksen, J., Kjaer, T., Sørensen, J., & Juhl, C.

(2015). Ultra-long term subcutaneous recording

system for EEG surveillance. In The 15th European

Congress on Clinical Neurophysiology (p. In press).

Duun-Henriksen, J., Kjaer, T. W., Looney, D., Atkins, M.

D., Sørensen, J. A., Rose, M., … Juhl, C. B. (2015).

EEG Signal Quality of a Subcutaneous Recording

System Compared to Standard Surface Electrodes.

Journal of Sensors, 2015, 1–9. doi:10.1155/2015/

341208

Elsborg, R., Remvig, L., Beck-Nielsen, H., & Juhl, C.

(2011). Detecting hypoglycemia by using the brain as

a biosensor. In P. A. Serra (Ed.), Biosensors for

Health, Environment and Biosecurity (1st ed., pp.

273–292). Rijeka: InTech. doi:10.5772/17018