real time brain activity measurement and signal processing system using highly sensitive mi sensor

Real-time brain activity measurement and signalprocessing system using highly sensitive MI sensor Kewang Wang,1Changmei Cai,2Michiharu Yamamoto,2 and Tsuyoshi Uchiyama1 1Graduate School

Real-time brain activity measurement and signal processing system using highly sensitive MI sensor

Kewang Wang, Changmei Cai, Michiharu Yamamoto, and Tsuyoshi Uchiyama

Citation: AIP Advances 7, 056635 (2017); doi: 10.1063/1.4974528

View online: http://dx.doi.org/10.1063/1.4974528

View Table of Contents: http://aip.scitation.org/toc/adv/7/5

Published by the American Institute of Physics

Articles you may be interested in

Magnetoelectric tuning of the inverse spin-Hall effect

AIP Advances 7, 055911055911 (2017); 10.1063/1.4973845

Scaling effect of spin-torque nano-oscillators

AIP Advances 7, 056624056624 (2017); 10.1063/1.4974014

Non-monotonic probability of thermal reversal in thin-film biaxial nanomagnets with small energy barriers

AIP Advances 7, 056006056006 (2017); 10.1063/1.4974017

Long GMI sensors for the detection of repetitive deformation of a surface

AIP Advances 7, 056621056621 (2017); 10.1063/1.4973747

Real-time brain activity measurement and signal

processing system using highly sensitive MI sensor

Kewang Wang,1Changmei Cai,2Michiharu Yamamoto,2

and Tsuyoshi Uchiyama1

1Graduate School of Engineering, Nagoya University, Nagoya, Japan

2Aichi Steel Corporation, Tokai, Japan

(Presented 2 November 2016; received 23 September 2016; accepted 28 October 2016;

published online 19 January 2017)

Superconducting Quantum Interference Devices (SQUIDs) are the most used sensor

to detect the extremely weak magnetic field of brain However, the sensor heads need to be kept at very low temperature to maintain superconductivity, and that makes the devices large-scale and inconvenient In order to measure brain activ-ity in normal environment, we had constructed a measurement system based on highly sensitive Magneto-Impedance (MI) sensor, and reported the study of mea-suring Auditory Evoked Field (AEF) brain waves In this study, the system was improved, and the sensor signals can be processed in real-time to monitor brain activity We use this system to measure the alpha rhythm in the occipital region and the Event-Related Field (ERF) P300 in the frontal, the parietal and both

the temporal regions © 2017 Author(s) All article content, except where other-wise noted, is licensed under a Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/) [http://dx.doi.org/10.1063/1.4974528]

INTRODUCTION

Advance in structural and functional research of brain gave us a more profound understanding

of our brains in the past few decades and more and more programs are committed to research in this field In medical & healthcare area, analyzing the brain signals can help in diagnosing some diseases (e.g epilepsy) or neural damage.1 Furthermore, the Brain Computer Interfacing (BCI) technology based on real-time brain signal processing, also contributed in numerous fields For example, the neuroprosthetics can be used in neural repair and rehabilitation, even substituting the non-functional arm or leg of those people with disabilities.2Measuring and analyzing the brain activity in real-time could also have wide applications for healthy users, such as driver drowsiness detection or video game interaction

Human brain produces very complex and weak electrical activity every second and the elec-trical activity can cause the change of brain magnetic field Magnetoencephalography (MEG) is an established technique for measuring brain magnetic field and arrays of Superconducting Quantum Interferential Devices (SQUIDs) are currently the most widely used sensors for MEG because of the high sensitivity and accuracy However, the SQUIDs need to be kept in a liquid nitrogen or liquid helium environment to maintain superconductivity during measurement, and that makes the devices large-scale and inconvenient On the other hand, the Magneto-Impedance (MI) sensor used in this study has no need to be cooled Furthermore, the circuit is mostly based on CMOS IC It makes the structure simpler and has a lower power cost.3,4

As one of the neural oscillations, the alpha rhythm, with a 8-13 Hz frequency range, can be easily identified at a maximum amplitude over the occipital region And it usually appears in relaxed wakefulness with eyes closed, drowsiness period and REM sleep stage, and the amplitude attenuate with the eyes opening.5Based on this feature, it can be applied in drowsiness detection during driving

or observing the levels of wakefulness And also, due to the large amplitude of the alpha rhythm, it can be an influential factor during other brain activity measurements, such as Event-Related Field (ERF) P300 The ERF, recorded via MEG, or Event-Related Potential (ERP) recorded via EEG, is

056635-2 Wang et al. AIP Advances 7, 056635 (2017)

brain responses that are time-lock to some event, like a sensory stimulus or recognition of a target stimulus It is the measured brain response that is the direct result of a specific sensory, cognitive or motor event The ERF can also be distinguished by their relative latency and polarity The P300 is an important component of ERF which is usually elicited in the process of decision making It surfaces

as a positive deflection with a latency of approximately 250 to 500 ms after the onset of a stimulus.6

In application level, the P300 could be implemented in such as brain injury inspection, diagnosis of neocortical epilepsy, or the P300 speller

MATERIALS AND METHODS

Measurement system

Fig.1(a)shows the schematic diagram of the measurement system The MI sensor used in this study is a micro magnetic sensor with high sensitivity (pico-Tesla resolution) The magnetometer is based on pulse-current magneto-impedance effect, which originated from the skin effect in FeCoSiB amorphous alloy wires.7 In order to measure the extremely weak magnetic field, inside the sensor head, two MI elements provided by Aichi steel corp,(both of their length are 1 cm) were set up in series Each element includes four wires in a pickup coil for the purpose of reduction of magnetic noises As shown in Fig 1(b), the measurement element was used to measure the total magnetic field, including biomagnetic field and environmental magnetic noise The reference element was set up 3 cm behind to cancel out environmental magnetic noise such as geomagnetic field Voltage difference between those two MI elements was used as the output It is an efficient way to reduce the geomagnetic field noise For purpose of making the device more flexible, and wearable eventually, the sensor head was also fixed into a head-wear polystyrene box connected with an extended shield cable

The sensor output signals were amplified (A=1000), and a 24-bits ADC (full scale measurement range is ±10 volts) was used to convert the output signal into a digital signal In the digital signal

FIG 1 The schematic diagram of the measurement system and the structure of MI sensor head.

FIG 2 The noise spectral density of the measurement system based on MI sensor The RMS noise level of the system is about 3.7 pT/Hz -1/2

processing module, signals were sampled with a 1000Hz rate and filtered in a frequency band of 0.1-40 Hz to reduce the high frequency components Fig.2shows the noise spectral density of MI measurement system The RMS noise level of the system is about 3.7 pT/Hz-1/2within 60 seconds

of measurement

METHODS

Alpha rhythm

The subject (a 29 year old healthy male with no neurological or psychiatric problems reported) sat in a comfy wooden chair with a relaxed state of mind The sensor head was set on the middle

of the occipital region (position between O1 and O2 of the international 10-20 system) Because of thickness of the polystyrene box was 5 mm, the distance between MI sensor head and scalp of the subject was considered as 5 mm During the measurement, the subject was instructed to open or close his eyes every 30 seconds In order to extract the alpha rhythm component, a 7-14 Hz bandpass filter was applied, then the brain signals of both eyes closed and open situation, was processed for spectral analysis to observe the variations in the frequency area

P300 measurement

On P300 measurement, subject was asked to do the oddball paradigm In this task, two kinds of visual stimuli were frequently presented in a random series once every 800 ms on a screen in front of the subject The standard stimulus had a large radius (r=10) and a high probability (p=0.8), and the target stimulus had a small radius (r=8) and a low probability (p=0.2) When a target stimulus was presented, the subject was instructed to press a response button to indicate it as quickly as possible, but no indication for a standard stimulus In order to reduce the influence caused by alpha rhythm, the sensor head was set on frontal, parietal and both temporal regions (position Fz, Pz, T3 and T4)

in four experiments Moreover, the subject was asked to keep his eyes open (blinking was accepted) Different from alpha rhythm, on P300 data processing, after applying a bandpass filter in 1-9 Hz frequency band, a range filter was applied to remove the data with over range, which was normally caused by subject movement Then the data were divided into the standard and the target conditions, and 100 standard and target conditions were chosen for arithmetic averaging in the same method respectively Flowchart of the real-time data processing module was shown in Fig.3

RESULTS

Alpha rhythm

Fig.4shows a part of the filtered sensor signals (3 seconds) in both eyes closed and open situations, Fig.5 (a) and(b)show the spectral density and the sum power spectrum level of all 60 seconds measurement (30 seconds with eyes open and 30 seconds with eyes closed) The results show that when the subject opened his eyes, not only the signal amplitude was attenuated, but also the sum

056635-4 Wang et al. AIP Advances 7, 056635 (2017)

FIG 3 Flowchart of the real-time data processing module.

power spectrum in every Hz level of the 8-13 Hz frequency band was observably attenuated than eyes closed situation, especially in 10 Hz

P300

The averaged P300 brain activity waveforms measured in four regions (Fz, Pz T3 and T4) are displayed in Fig.6 The results show that in parietal and the right temporal region, positive deflections

FIG 4 Filtered sensor signals in both eyes closed and open situations.

FIG 5 The spectral density and the sum power spectrum level of all 60 seconds measurement.

with a latency of 300 ms can be observably elicited by target stimuli, but barely elicited by standard stimuli, but in frontal and left temporal region, the deflections are negative, and the latency times are 300-400 ms

DISCUSSION

Alpha rhythm and P300 measurement

We referred the experiment methods and compared our results with other alpha rhythm research via EEG and MEG,5,8 it was confirmed that when the subject closed his/her eyes, the intensity of the alpha rhythm was considerably reduced and the main spectral contribution was observed in the

10 Hz frequency component

On P300 measurement, we referred to other results of MEG research based on visual stimuli elicited P300,9 and our past study,10 the deflections can be elicited by Target stimuli but not by Standard stimuli around 300-400 ms and the results showed similar characteristics to our results in this study Furthermore, the absolute value of a P300 peak was measured in about 19.97-37.23 pT in this study, based on the calculation model we reported previously, the P300 peak was estimated as 21.71 pT, and it is about 120 times bigger than the estimated value of SQUIDs based on the same model.11

About the direction of brain magnetic field, as shown in reference8, Fig.3, this figure shows some neuromagnetic field maps of eight subjects Six of them presented that the polarity located over the left and right hemisphere can be opposite And we considered that was the reason why the P300 results measured in frontal and left temporal regions had the opposite polarity with the results measured in parietal and right temporal regions

MI sensor measurement system

The main goal of this study was to develop a brain activity measurement system based on the

MI sensor, which can process the data in real-time While the results presented the capabilities of the MI sensor, as a small-scale, low-cost and highly sensitive magnetic sensor, for applications of biomagnetic field measurement In our next step, we expect to improve the system into a multi-channel system to reduce the noise produced by the body using multivariate statistics, for example, Principal

056635-6 Wang et al. AIP Advances 7, 056635 (2017)

FIG 6 The averaged P300 brain activity waveforms measured in the frontal, the parietal and both the temporal regions.

Component analysis (PCA) and Independent Component Analysis (ICA) On the other hand, we are also focusing on making the circuit all into digital circuit using FPGA, and we expect to obtain a lower noise level

In this study, we measured the alpha rhythm and the P300 brain activity in the frontal, the parietal and both the temporal regions using a real-time measuring and processing system based on highly sensitive MI sensor After comparing our results with other relevant research, the reliability of our data was confirmed and it presents the capabilities of the MI sensor for applications of brain activity measurement

1U Aydin, J Vorwerk, M D¨umpelmann, P K¨upper, H Kugel, M Heers, J Wellmer, C Kellinghaus, J Haueisen, S Rampp,¨

H Stefan, and C H Wolters, “Combined EEG/MEG can outperform single modality EEG or MEG source reconstruction

in presurgical epilepsy diagnosis,” PloS ONE

2 F A Mussa-Ivaldi and L E Miller, “Brain–machine interfaces: Computational demands and clinical needs meet basic neuroscience”, Trends Neurosci.26(6), 329–334 (2003).

3 L.V Panina, K Mohri, K Bushida, and M Noda, “Giant magneto-impedance and magneto-inductive effects in amorphous alloys,” J Appl Phys.76, 6198 (1994).

4 K Mohri, T Uchiyama, L P Shen, C M Cai, L V Panina, Y Honkura, and M Yamamoto, “Amorphous wire and CMOS IC-based sensitive micromagnetic sensors utilizing magnetoimpedance (MI) and stress-impedance (SI) effects,” IEEE Trans Magn.38, 3063 (2002).

5 J L Cantero, M Atienza, and R M Salas, “Human alpha oscillations in wakefulness, drowsiness period, and REM sleep: Different electroencephalographic phenomena within the alpha band,” Neurophysiol Clin.32, 54–71 (2002).

6 J Polich, “Updating P300: An integrative theory of P3a and P3b,” Clinical Neurophysiology118(10), 2128–2148 (2007).

7 T Uchiyama, K Mohri, Y Honkura, and L V Panina, “Recent advances of pico-tesla resolution magneto-impedance sensor based on amorphous wire CMOS IC MI sensor,” IEEE Transactions on Magnetics48, 11 (2012).

8 C Ciulla, T Takeda, and H Endo, “MEG characterization of spontaneous alpha rhythm in the human brain,” Brain Topography11, 3 (1999).

9 A Mecklinger, B Maess, B Opitz, E Pfeifer, D Cheyne, and H Weinberg, “A MEG analysis of the P300 in visual discrimination tasks,” Electroencephalogr Clin Neurophysiol.108, 45–56 (1998).

10 K Wang, S Tajima, D Song, N Hamada, C Cai, and T Uchiyama, “Auditory evoked field measurement using magneto-impedance sensors,” Journal of Applied Physics117, 17B306 (2015).

11 K Wang, S Tajima, Y Asano, Y Okuda, N Hamada, C Cai, and T Uchiyama, “Detection of P300 brain waves using

a magneto-impedance sensor,” in 2014 the 8th International Conference on Sensing Technology (Liverpool, UK, 2014),

pp 2–4.