Model of Software and Hardware Infrastructure for Electrophysiology

Petr Jeˇzek

, Jan

Stˇebet´ak

, Petr Br˚uha

and Roman Mouˇcek

New Technologies for the Information Society, University of West Bohemia,

Univerzitni 8, 306 14, Pilsen, Czech Republic

Department of Computer Science and Engineering, University of West Bohemia,

Univerzitni 8, 306 14, Pilsen, Czech Republic

Keywords:

Neuroinformatics, Electrophysiology, Infrastructure, Data Management, Semantic Web, Analytic Tool,

Semantic Framework.

Abstract:

Large amounts of EEG/ERP (electroencephalography, event-related potential) data, various data formats and

non-standardized domain description lead to incompatible results and interpretations of EEG/ ERP experimen-

tal data/metadata and to difﬁcult communication between interested laboratories. Authors’ research group has

solved these problems and has contributed to the building of a neuroinformatics infrastructure by developing

and integrating data management and analytic tools for EEG/ERP research. The model of the software and

hardware infrastructure for electrophysiology, and the context and architecture of the developed EEG/ERP

Portal, serving to manage, share and process EEG/ERP experiments, are presented. Other additional tools are

brieﬂy described.

1 INTRODUCTION

Our research group as a member of the Czech Na-

tional Node of International Neuroinformatics Coor-

dinating Facility (INCF) (INCF, 2012) participates in

deﬁnition and development of standardized formats

for electrophysiology research. Our efforts resulted

in the central custom solution - the EEG/ERP (elec-

troencephalography, event-related potential) Portal.

In this paper we brieﬂy present the infrastructure

solutions in neuroscience as they are provided by the

member countries of INCF. Then the basic ideas and

the model of software and hardware infrastructure for

electrophysiology is introduced. In the next sections

the overall concept and architecture of the EEG/ERP

Portal is described. Since we need to register the

Portal as a recognizable data source, we also imple-

mented the Semantic Framework module (Section 5)

that enables users transformation of the common ob-

ject oriented source code into semantic web languages

(Berners-Lee et al., 2001). Semantic gaps between

the object-oriented code and its semantic web rep-

resentation are bridged by extension of the object-

oriented code using Java annotations. The EEG/ERP

Portal is also supposed to be integrated with various

external tools; these are described in Section VI.

2 STATE OF THE ART

The neuroinformatics infrastructure is being built in

several INCF national nodes in parallel. The INCF

portal (INCF, 2012) includes, for example, a soft-

ware center for easy storage and sharing of neuroin-

formatics software tools, a content management sys-

tem for national nodes presentation and provides ac-

cess to supercomputing resources for the neuroinfor-

matics community.

The Neuroscience Information Framework

(NIF) (Gupta et al., 2008) is a dynamic inventory

of registered Web-based neuroscience resources

containing data, materials, and tools. It advances

research in neuroscience by enabling access to public

research data and tools through an open source

environment. Currently more than 4,000 sources are

registered within NIF.

The Carmen portal (Carmen, 2012) provides stor-

age of experimental data, metadata and analysis code.

Users can also analyze their experimental data. Their

analytical tools are implemented as web services.

Within the Japan National Node, neuroinformatics

databases are organized and shared with the public in

platforms. The platforms serve as storages of exper-

imental data and documents; some of them provide

analytic tools.

Semantic web solutions for EEG/ERP data man-

352

Ježek P., Št

ebeták J., Br˚uha P. and Mou

cek R..

Model of Software and Hardware Infrastructure for Electrophysiology.

DOI: 10.5220/0004245103520356

In Proceedings of the International Conference on Health Informatics (HEALTHINF-2013), pages 352-356

ISBN: 978-989-8565-37-2

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

agement focus mainly on building ontologies. The

ontology built within the NEMO project provides for-

mal semantic deﬁnitions of concepts in ERP domain,

including ERP patterns, spatial, temporal, functional

(cognitive/behavioral)attributes of these patterns, and

data acquisition and analytic methods and parameters

(NEMO, 2011).

The fundamental method for obtaining ERP com-

ponents from the EEG signal is averaging. The meth-

ods used for ERP components detection include the

matching pursuit algorithm, wavelet transform and

Hilbert-Huang transform (

Rondik et al., 2011).

3 MODEL OF SOFTWARE

AND HARDWARE

INFRASTRUCTURE

The basic idea of the model of the software and hard-

ware infrastructure for electrophysiology comes from

the set of main activities performed by researchers

during electrophysiological experiments.

First of all, researchers present a hypothesis and

design a protocol for a speciﬁc experiment. Then they

perform the experiment and collect data and related

metadata. During the experiment they have to syn-

chronize the EEG signal obtained from the scalp of

the tested subject with presented stimuli.

Second, they analyse the data using various pro-

cessing methods, interpret the data and publish re-

sults. In the long term, they do not care about the

data and lose them.

The biggest problemsfor science are the following

ones: since data are not published, conclusions and

interpretations cannot be later reproduced or veriﬁed,

the methods used for data analysis are lost or their

detailed parameters are not later traceable.

The basic aim of building the software and hard-

ware infrastructure for electrophysiology is to in-

crease both effectiveness and efﬁciency of scientiﬁc

research in this ﬁeld. The integrated system is con-

sidered as a service (compare to (Watson et al., 2007))

providing usually the web-based interface. The main

features of this service include a long-term and sus-

tainable storage of data and related metadata collected

from experiments, various methods for data process-

ing, workﬂows for data processing and sharing of

data, documents, methods and workﬂows in groups.

The software infrastructure with the central

EEG/ERP portal is shown in Fig. 1. The hardware in-

frastructure includes stimulators for ERP experiments

and complex electronic switches for synchronization

of impulses.

Sharing is very helpful for scientiﬁc community.

However, the present organization of science does not

support it entirely. Scientists usually want to be the

ﬁrst to publish their ideas or results to gain recogni-

tion and money for their future research. Therefore it

is not possible to fully open publication of ideas, data,

or processing methods at every stage of the scientiﬁc

process. As the result, sharing of knowledge has to

be limited at least at early stage of the scientiﬁc pro-

cess to deﬁned groups of people. These groups can be

gradually enlarged as the scientiﬁc process continues.

Finally, after successful publication, the full access to

data and to used methods is possible. It is clear that

some data, for example, personal data of tested sub-

jects have to be secured according to law. Laws or

other privacy issues can even prevent any sharing of

some data. After reading the paper other researchers

can be interested in the data or their processing. For

example, they want to use the same methods or work-

ﬂows, which are applied to their data or they want to

use original data to verify their own methods. Then

new valuable results available for publication can be

produced with adequate effort.

4 EEG/ERP PORTAL

The purpose of the EEG/ERP Portal is to serve as

a managing tool for EEG/ERP experiments. It en-

ables interested laboratories sharing and interchange

of the stored experiments (data, metadata, experimen-

tal scenarios, etc.). The features of the EEG/ERP

Portal also include sharing of knowledge, working

in groups, content management system, and full-text

search.

The data are protected by the system of user ac-

counts with deﬁned user roles. Individual users are

grouped into self-managed groups. On the basis of

activities that the user can perform four user roles are

recognized (Reader, Experimenter, Group Adminis-

trator, and Supervisor).

We prepared a simple wizard that guides the

logged user through the process of adding an exper-

iment. Each experiment contains raw data supple-

mented by related metadata. We deﬁned a set of meta-

data which the user is instructed to ﬁll in through the

prepared forms. These metadata are organized in sev-

eral semantic groups. The experimenter can also de-

cide if the experiment is private or public. Public ex-

periments are downloadable for all registered users

(without personal data of tested subjects), while pri-

vate experiments are downloadable only within the

experimenter’s group.

ModelofSoftwareandHardwareInfrastructureforElectrophysiology

353

Figure 1: Component Model of EEG/ERP Infrastructure.

5 SEMANTIC WEB EXTENSION

Before we started the registration of the EEG/ERP

Portal as a recognizable neuroscience data source

within the NIF (Gupta et al., 2008), we had de-

scribed EEG/ERP experiments by a suitable ontol-

ogy. The Semantic Web uses a triple oriented repre-

sentation described by Resource Description Frame-

work (RDF) and its extension Ontology Web Lan-

guage (OWL). Despite advantages that the Seman-

tic Web (semantic description of stored data) brings,

the EEG/ERP Portal is designed using object oriented

modeling (OOM).

We tested several frameworks to enrich an ob-

ject oriented code by missing semantics. ActiveRDF

(Oren et al., 2007) is a library for accessing RDF data

from Ruby programs. eClass (Liu et al., 2007) is an

extended Java class with changed syntax that contains

additional information about ﬁelds and methods. The

Semantic Object Framework (Po-Huan et al., 2009)

adds missing semantics using embedded comments.

The tested frameworks are difﬁcult to use either

because they do not include suitable means for adding

missing semantics, or they require a modiﬁcation in

common programming syntax.

6 EXTERNAL LIBRARIES

AND TOOLS

Since the EEG/ERP Portal cooperates with a set of

associated submodules we implemented several com-

munication interfaces for external tools. The tools

working with the data stored in the EEG/ERP Por-

tal can be divided into two groups. The ﬁrst group

includes tools that are accessible through an inter-

net browser. These tools are implemented as stand-

alone libraries integrated within the EEG/ERP Portal

directly. The second group of tools includes desktop

tools that are run locally on the user’s computer. The

desktop tools access data in the EEG/ERP Portal and

manage them locally.

6.1 Libraries of EEG/ERP Portal

The libraries, implemented outside the system, are

gradually added into the EEG/ERP Portal. The li-

braries are simultaneously placed into the open source

repositories where they can be used by the interested

community.

HEALTHINF2013-InternationalConferenceonHealthInformatics

354

6.1.1 Semantic Framework

With regards to the difﬁculties mentioned in Section 5

we decided to implement a custom framework based

on common programming technologies. It resulted

in the development of the Semantic Framework that

uses Java Annotations (Java, 2012). These annota-

tions enable us to add advanced OWL constructs into

the object oriented code. The Semantic Framework

uses the modiﬁed JenaBean (Jena, 2011) in combina-

tion with the OWL API (OWL, 2012). The JenaBean

was extended by the set of annotations mapped to ap-

propriate OWL constructs (Mouˇcek and Jeˇzek, 2010).

6.1.2 Waveforms Reader

The data obtained during the EEG/ERP experiments

are stored in the format given by the used analogue-

digital converter. We provide a module for play-

ing the stored signal in the web-based interface of

the EEG/ERP Portal. The Waveforms Reader sepa-

rates individual waveforms that are visualized using

HTML5 and CSS3 technologies. The rendering is

smooth even if the signal is sought or zoomed.

6.1.3 Analytic Methods

EEG/ERP data are supposed to be further processed.

We implemented the methods described in Section 2

as single libraries. When the user executes a method,

this method is running in a separate thread in the back-

ground. When the execution of the method ends, the

results are visualized in a chart. It ensures that a par-

allel work may be done although the execution of the

methods may be time-consuming.

6.2 External Tools

External tools work independently of the EEG/ERP

Portal. The EEG/ERP Portal provides an interface

for accessing stored experiments using Web Services

technology. The Apache CXF service framework is

used. The Web Service interface provides several

methods for accessing users experiments with their

data, metadata and scenarios. The Web Service is se-

cured by user credentials. When any request for other

functionality occurs the provided interface can be ex-

tended. An interested client only implements a Web

Service client.

6.2.1 JERPA

JERPA (Jeˇzek and Moucˇcek, 2011) is a solution for

visualization of EEG/ERP records. We extended the

present solution about the possibility to download ex-

periments from the EEG/ERP Portal. The user can

download data of his/her experiments to the local stor-

age and visualize them. The functionality for upload-

ing a new experiment to the EEG/ERP Portal is also

provided. The local storage is synchronized with the

EEG/ERP Portal in regular intervals.

6.2.2 Analytic Tools

Since processing of analytic methods is a time con-

suming task we designed and implemented the system

running on the separate server. This system imple-

ments analytic methods described in Section 2. The

system contains a simple web-based interface where

the user can list available methods and accessible

server resources. The implemented methods are ac-

cessed using the Web Service as well. When any user

wants to use this system he/she has to register.

6.2.3 Ofﬂine EEG/ERP Portal

Since there are many situations in which the Internet

connection is not available during performing experi-

ments, we prepared an ofﬂine (simpliﬁed) version of

the EEG/ERP Portal as a desktop application. The

performed experiments are stored in the local storage.

If Internet connection is available, the local storage is

synchronized with the EEG/ERP Portal.

6.2.4 Statistical Methods

The library of statistical methods was implemented as

a standalone product that runs on the separate server

as well. Currently we have implemented one way

and two way analysis of variance (ANOVA) and one

way and two way multivariate analysis of variance

(MANOVA).

6.2.5 Annotation Tool

Since we provide the EEG/ERP experiments in the

Semantic Web form using the Semantic Framework,

the domain expert has to prepare a set of annotated

POJO objects. The annotation tool is a standalone

system that allows users to load POJOs from the per-

sistence layer of the EEG/ERP Portal and annotate

them. These annotated POJOs are exported from the

Annotation Tool and integrated within the EEG/ERP

Portal.

7 CONCLUSIONS

The presented software and hardware infrastructure

ModelofSoftwareandHardwareInfrastructureforElectrophysiology

355

combines research in informatics with research in

electrophysiology. Its aim is to help researchers in

their work, which includes especially long-term col-

lection, maintenance, usage and share of EEG/ERP

data and related tools. The set of metadata is deﬁned

reﬂecting experience from experiments carried out in

our laboratory, medical expertise providedby the Uni-

versity Hospital in Pilsen, INCF recommendations,

and analysis of related books and scientiﬁc papers.

The central point of the infrastructure, the EEG/ERP

Portal, is being developed as an open source using

common, basically object oriented technologies.

Since the registration of the EEG/ERP Portal as a

recognizable data source requires providing its onto-

logical description, transformation of the object ori-

ented code into the semantic web technologies is re-

quired. The Semantic Framework provides this trans-

formation and adds missing semantics using Java An-

notations. The integration of the developed frame-

work into the EEG/ERP Portal ensures the online

serialization of the stored experiments into semantic

web languages. Since processing of EEG/ERP exper-

iments includes a usage of a large set of methods, the

ways to run these methods were brieﬂy presented.

Our future work will focus on expansion and im-

provement of the infrastructure. It includes building

of the limited social network within the EEG/ERP

Portal, cooperation of the Portal with social networks

used by the community, the full registration of the

EEG/ERP Portal within NIF, extensive work on the

methods for EEG/ERP signal processing and building

of workﬂow patterns. An equally important task is to

continue in the standardization of data and metadata

structures.

ACKNOWLEDGEMENTS

The work was supported by the UWB grant SGS-

2010-038 Methods and Applications of Bio- and

Medical Informatics and by the European Regional

Development Fund (ERDF), Project ”NTIS - New

Technologies for Information Society”, European

Centre of Excellence, CZ.1.05/1.1.00/02.0090.

REFERENCES

Berners-Lee, T., Hendler, J., and Lassila, O. (2001). The

semantic web, scientiﬁc american magazine.

Carmen (2012). Carmenportal. http://www.carmen.org.uk/.

Gupta, A., Bug, W., Marenco, L., Qian, X., Condit, C., Ran-

garajan, A., M¨uller, H. M., Miller, P. L., Sanders, B.,

Grethe, J. S., Astakhov, V., Shepherd, G., Sternberg,

P. W., and Martone, M. E. (2008). Federated access

to heterogeneous information resources in the neuro-

science information framework (nif). Neuroinformat-

ics 6, pages 205–217.

INCF (2012). International neuroinformatics co-

ordinating facility. http://www.incf.org/about/

what-is-neuroinformatics.

Java (2012). Java annotations. http:// download.oracle.com/

javase/1.5.0/docs/guide/language/annotations.html.

Jena (2011). Jena framework. http://jena.sourceforge.net.

Jeˇzek, P. and Moucˇcek, R. (2011). Integration of signal

processing methods into eeg/erp system. In Traver,

V., Fred, A. L. N., Filipe, J., and Gamboa, H., editors,

HEALTHINF, pages 563–566. SciTePress.

Liu, F., Wang, J., and Dillon, S. T. (2007). Web informa-

tion representation, extraction and reasoning based on

existing programming technology. In Computational

Inteligence 37, pages 147–168.

Mouˇcek, R. and Jeˇzek, P. (2010). System for storage and

management of eeg/erp experiments - generation of

ontology. In Filipe, J. and Cordeiro, J., editors, ICEIS

(1), pages 415–420. SciTePress.

NEMO (2011). Neural electro magnetic ontologies.

http://nemo.nic.uoregon.edu/wiki/NEMO.

Oren, E., Delbru, R., Gerke, S., Haller, A., and Decker, S.

(2007). Activerdf: object-oriented semantic web pro-

gramming. pages 817–824.

OWL (2012). Owl api. http://owlapi.sourceforge.net/.

Po-Huan, C., Chi-Chuan, L., and Kuo-Ming, C. (2009). In-

tegrationg semanic web and object-oriented program-

ming for cooperative design. Journal of University

Computer Science,vol. 15, no. 9.

Rondik, T., Ciniburk, J., Mouˇcek, R., and Mautner, P.

(2011). Erp components detection using wavelet

transform and matching pursuit algorithm. In Applied

Electronics (AE), 2011 International Conference on,

pages 1 –4.

Watson, P., Jackson, T., Pitsilis, G., Gibson, F., Austin, J.,

Fletcher, M., Liang, B., and Lord, P. (2007). The

CARMEN Neuroscience Server. In Proceedings of

the UK e-Science All hands Meeting, pages 135–141.

HEALTHINF2013-InternationalConferenceonHealthInformatics

356