Low-complexity hardware design methodology for reliable and automated removal of ocular and muscular artifact from EEG

https://doi.org/10.1016/j.cmpb.2018.02.009Get rights and content

Highlights

  • A reliable, robust and automated low complexity hardware design methodology to remove blink and muscular artifacts real time without the need of any extra electrode.

  • Denoised EEG can be obtained for online diagnosis in a pervasive personalized healthcare environment.

  • Furthermore, this can be achieved with comparable accuracy when compared with state of the art approaches.

  • In this paper, FPGA implementation based on the proposed methodology is carried out and hardware complexity along with power consumption (only 75 micro-Watt) are also reported.

  • Therefore it can be implemented in a low complexity fashion on a chip-set to ensure the battery backup, that drives electronics, sustains for longer time than the state of the art approaches.

Abstract

Background and objective

EEG is a non-invasive tool for neuro-developmental disorder diagnosis and treatment. However, EEG signal is mixed with other biological signals including Ocular and Muscular artifacts making it difficult to extract the diagnostic features. Therefore, the contaminated EEG channels are often discarded by the medical practitioners which may result in less accurate diagnosis. Many existing methods require reference electrodes, which will create discomfort to the patient/children and cause hindrance to the diagnosis of the neuro-developmental disorder and Brain Computer Interface in the pervasive environment. Therefore, it would be ideal if these artifacts can be removed real time on the hardware platform in an automated fashion and then the denoised EEG can be used for online diagnosis in a pervasive personalized healthcare environment without the need of any reference electrode.

Methods

In this paper we propose a reliable, robust and automated methodology to solve the aforementioned problem. The proposed methodology is based on the Haar function based Wavelet decompositions with simple threshold based wavelet domain denoising and artifacts removal schemes. Subsequently hardware implementation results are also presented. 100 EEG data from Physionet, Klinik für Epileptologie, Universität Bonn, Germany, Caltech EEG databases and 7 EEG data from 3 subjects from University of Southampton, UK have been studied and nine exhaustive case studies comprising of real and simulated data have been formulated and tested. The proposed methodology is prototyped and validated using FPGA platform.

Results

Like existing literature, the performance of the proposed methodology is also measured in terms of correlation, regression and R-square statistics and the respective values lie above 80%, 79% and 65% with the gain in hardware complexity of 64.28% and improvement in hardware delay of 53.58% compared to state-of-the art approaches. Hardware design based on the proposed methodology consumes 75 micro-Watt power.

Conclusions

The automated methodology proposed in this paper, unlike the state of the art methods, can remove blink and muscular artifacts real time without the need of any extra electrode. Its reliability and robustness is also established after exhaustive simulation study and analysis on both simulated and real data. We believe the proposed methodology would be useful in next generation personalized pervasive healthcare for Brain Computer Interface and neuro-developmental disorder diagnosis and treatment.

Introduction

Neuro-developmental disorders (NDD) including Attention Deficit Hyperactivity Disorder (ADHD), Schizophrenia, Down syndrome, Autism Spectrum Disorder (ASD), intellectual retardation, learning disablement are impairments in the evolution of the brain or the central nervous system which manifest early in the development, often during infancy or before child enters into the socio-academic education. While the symptoms and behavior of NDD including language and speech learning, motor synchronization, behavior, retention, imagination under-development, communication [1] differ from individual to individual, some children with such disabilities in the childhood develop permanent damages. For example, children with ASD show impairment in social interaction, deficit in communication and motor coordination, repetitive or stereotyped behavior, lack of cognitive skills, language loss, and atypical visual perception [2], [3], [4], [5], [6].

To diagnose NDD for ASD, Autism Diagnostic Observation Schedule (ADOS) and Autism Diagnostic Interview-Revised (ADI-R) are used comprising of a series structured tasks and interviews respectively involving the interactions among patient, examiner and parents. The examiner identifies the patient's response to the tasks and suggests a proper treatment procedure [7]. However, such procedures involve a constant observation on the children, significant amount of parenting, treatment time and huge long-term expenses. The recent Neuroimaging techniques that discovered an overgrowth of the cortical white matter and abnormal pattern in frontal and temporal lobe during prenatal and postnatal period of brain evolution generally require a sedation and radioactive dye [8]. Both of the above-mentioned procedure requires high quality medical facilities for intensive care in home environment.

On the other hand EEG- electrical recording of the brain systematic activity along the scalp, measured by the voltage fluctuations resulting from the ionic current which flows within the neuron [8], [9], [10], is cost-effective and non-invasive tool for the exploration of different brain regions for cognitive and other event-related activities of a subject [8], [10]. The diagnostic feature extracted from the EEG signals can reveal the brain functionality in the area of particular task and can be used as a biomarker to classify between the NDD and healthy control. In fact, EEG has also been arguably the most widely used mechanism for acquiring brain computer interface (BCI) signals from the brain for the control of computers or other devices via the modulation of neuro-logical activity in the participant's' brain without the need of any activation of the efferent nervous system [11]. However in daily life, since EEG signal is mixed with other biological signals of non-interest [10], [12], including blink and muscle artifacts, it is difficult to extract the diagnostic features. Therefore, Medical Practitioners during offline visual observation, tend to discard the EEG channels containing these artifacts [2] which results in less accurate diagnosis [13]. Similarly, in case of online automated diagnosis using EEG in tele-health framework under internet of things, these artifacts may cause wrong diagnosis triggering false alarm and causing panic. Recently, there is an attempt to propose an online and automated EEG artifacts removal scheme in [11], [14], [15] targeting BCI where, although the processing is done online, the acquisition of the data is still done off-line which deviates from the need of our envisaged goal of having a pervasive personalized healthcare monitoring system.

Therefore, a robust methodology which would remove the effect of these artifacts as well as retrieve EEG amidst the presence of these artifacts would be extremely helpful for BCI and also for enhanced diagnosis of NDD in real-time online personalized home care environment. But since, the frequency spectrum of blink and muscular artifact overlaps with the normal EEG signal, it poses a commendable challenge in achieving the target of retrieving artifact free EEG in real-time automated fashion on hardware platform. To tackle this challenge the researchers use Independent Component Analysis (ICA) Blind source Separation and wavelet based time frequency algorithm to remove the ocular artifacts [10], [16], [17], [18], [19], [20], [21], [22], [23] and muscular artifacts [13], [15], [20], [22], [24] from the EEG [41], [42], [43]. Although these methods are non-invasive, most of these methods (except [11], [12], [14], [15]) require external ocular electrodes near the eyes, which will cause discomfort thereby making it unsuitable for the personalized remote health care. Furthermore, such arrangements make the patients/children conscious about the presence of these extra electrodes, which causes hindrance to the appropriate diagnosis of the NDD. In [12] the need of these ocular electrodes have been eradicated however, this recent method does the processing of the data acquired offline [12].

Motivated by the fore-mentioned challenges, in this paper, we propose:

  • a reliable, robust and automated low complexity hardware design methodology to remove blink and muscular artifacts real time without the need of any extra electrode;

  • a mechanism for online EEG based NDD diagnosis in a pervasive personalized healthcare environment;

  • a methodology with comparable accuracy when compared with state of the art approaches and that can be implemented in a low complexity fashion on a chip-set to ensure the battery backup, that drives electronics, sustains for longer time than the state of the art approaches and

  • an FPGA implementation along with the hardware complexity analysis with power consumption of only 75 micro-Watt.

The paper is organized as follows, Section 1 describes materials and methods, Section 2 proposes the methodology, Section 3 presents the results, necessary comparisons and relevant discussions and finally Section 4 concludes the paper.

Section snippets

Prerequisites

Fig. 1 depicts the block diagram of the system based on the proposed methodology. A 21 and 64 channel EEG signal mixed with muscle and ocular artifacts in random ratio, is passed through FastICA [25] block. Considering A = mixing matrix, S = source matrix, X = input mixed matrix, FastICA works on X = A*S and estimates Y = W * X where Y ≈ S, W = Unmixing matrix and Y = estimated independent component matrix. The proposed method works on the output of the ICA (e.g. FastICA has been used in this

Results

The EEG signals database obtained from Physionet [32], Klinik für Epileptologie, Universität Bonn, Germany [33] consisting of five sets from healthy volunteer and epileptic patient with different activities, Caltech [34] and University of Southampton [43] were recorded at sampling frequency of 256 Hz, 173.6 Hz, 160 Hz and 500 Hz respectively.

Like all existing techniques, we also chose two categories of data for validation – 1. Simulated artifacts with Clean EEG (case study – I to IX in Table 2)

Conclusion

The automated methodology proposed in this paper, unlike the state of the art methods, can remove blink and muscular artifacts real time on the hardware without the need of any extra electrode. Its reliability and robustness is also established after exhaustive simulation study and analysis on both simulated and real data. The performance of the proposed methodology is measured in terms of correlation, regression and R-square statistics and it has been found that their average values lie above

Acknowledgment

The project is partly funded under the Early Carrier Research Grant from the Science and Engineering Research Board, Govt. of India with Sanction No: ECR/2015/000148 dated 14th July 2016. Amit Acharyya would also like to acknowledge his Visvesvaraya Young Faculty Fellowship by DEITY, Govt. of India.

References (43)

  • S. Rondeau “Electroencephalogram use in Autistic Disorder Assessment” Naturop. Doctor News Rev....
  • Methodological and Clinical Features

    Psychophysiology

    (2012)
  • Rajesh Patel

    Ocular artifact suppression from EEG using ensemble empirical mode decomposition with principal component analysis

    Comput. Electric. Eng.

    (2015)
  • Ian Daly et al.

    Martin Billinger, and Gernot Muller-Putz. "FORCe: fully online and automated artifact removal for brain-computer interfacing

    IEEE Trans. Neural Syst. Rehabil. Eng.

    (2015)
  • L. Frølich et al.

    Classification of independent components of EEG into multiple artifact classes

    Psychophysiology

    (2015)
  • Tzyy-Ping Jung et al.

    Sejnowskil “Removing Electroencephalographic Artifacts: Comparison between ICA and PCA

    Neural Netw. Signal Process.

    (1998)
  • MdRakibul Mowla

    Artifacts-matched blind source separation and wavelet transform for multichannel EEG denoising

    Biomed. Signal Process. Control

    (2015)
  • V. Krishnaveni et al.

    Automated Identification and removal Of Ocular Artifacts From EEG using Wavelet Transform

    Measur. Sci. Rev.

    (2006)
  • M. Mennes et al.

    Validation of ICA as a tool to remove eye movement artifacts from EEG/ERP

    Psychophysiology

    (2010)
  • C.A. Joyce et al.

    Automated removal of eye movement and blink artifacts from EEG data using blind source separation

    Psychophysiology

    (2004)
  • Burger Christiaan et al.

    Removal of EOG artifacts by combining wavelet neural network and independent component analysis

    Biomed. Signal Process. Control

    (2015)
  • Cited by (0)

    View full text