BIOSIG

Standardization and Quality Control in Biomedical Signal Processing using the

BioSig Project

A. Schlögl, C. Vidaurre

Fraunhofer FIRST-IDA, Kekulestrasse 7, 12489, Berlin, Germany

Ernst Hofer, Thomas Wiener

Institute of Biophysics, Medical University of Graz, Harrachgasse 21, 8010, Graz, Austria

Clemens Brunner, Reinhold Scherer

Institute for Knowledge Discovery, University of Technology Graz, 8010, Graz, Austria

Franco Chiarugi

Institute of Computer Science (ICS), Foundation for Research and Technology - Hellas (FORTH) P.O. Box 1385, Vassilika

Vouton, GR 711 10 Heraklion, Crete, Greece

Keywords: Standardization, Quality control, electroencephalogram, Brain Computer Interfaces, Free/libre open source

software (FLOSS).

Abstract: Biomedical signal processing is an important but underestimated area of medical informatics. In order to

overcome this limitation, the open source software library BioSig has been established. The tools can be

used to compare the recordings of different equipment providers, it provides validated methods for artifact

processing and supports over 40 different data formats (more than any other software in this area). BioSig

provides reference implementations for biomedical signal processing questions and holds the top rank

among all biomedical signal processing projects registered at SourceForge. Thus is provides standardization

and quality control for the field of biomedical signal processing.

1 INTRODUCTION

Biomedical signal processing is an important area of

medical informatics and is used in many subject area

(neurology, psychiatry, cardiology, pulmology,

cognitive neuroscience, psychology, biophysics,

biomedical engineer etc) with many different

applications. Unlike other areas of medical

informatics (Imaging, labor diagnostics, patient

information system etc.), biomedical signals are not

well represented in health information systems.

A likely explanation is the fact that many

different disciplines and many small groups do

biomedical signal processing. The interaction

between the various groups is not well organized;

often the same well-known methods are

implemented again and again, the wheel is re-

invented again and again. In order to overcome this

problem, the open source software project BioSig

was founded with the aim to provide a software

library for biomedical signal processing. Motivated

by the successful development model of the Linux

operating system, it was decided that the library

should be also open source, everyone is invited to

use and to contribute to BioSig, and the GNU

General Purpose License (GPL) ensures that BioSig

will stay open source.

Section 2, presents several subprojects that have

been developed within BioSig. Section 3 provides

some numbers about the success of BioSig, Sections

4 and 5 discuss open issues summarize the project.

403

Schlögl A., Vidaurre C., Hofer E., Wiener T., Brunner C., Scherer R. and Chiarugi F. (2008).

BIOSIG - Standardization and Quality Control in Biomedical Signal Processing using the BioSig Project.

In Proceedings of the First International Conference on Bio-inspired Systems and Signal Processing, pages 403-409

DOI: 10.5220/0001065904030409

 SciTePress

2 METHOD

Software development and programming is an

important aspect of biomedical signal processing. In

order to address different needs, several

programming languages are supported. The Matlab

scripting language is widely used in biomedical

signal processing and engineering. For this reason,

the first part of BioSig has been implemented in

Matlab language. Matlab is an proprietary software

product from “The Mathworks Inc”, short TMW, but

there are also some open source alternatives

available; Octave (http://www.octave.org) is

probably the most widely known alternative. Special

effort was undertaken to make the code also

compatible to Octave. This part is now call “BioSig

for Octave and Matlab”, or short “BioSig4OctMat”.

Moreover, C/C++ is a very flexible

programming language and provides a very efficient

(i.e. fast) software; although the software

development takes more time. There is now a

common C/C++ interface to access various data

formats including the SCP-ECG standard (EN1064),

the HL7aECG, GDF (Schlögl et al. 1999b, Schlögl,

2006b) and several other data formats.

Furthermore, projects for the languages Python

(BioSig4Python) and Java (BioSig4Java) have been

started.

2.1 Converter between SCP-ECG and

HL7 aECG

SCP-ECG (Standard Communications Protocol for

Computer-Assisted Electrocardiography) is a

European standard (EN1064:2005) for interpretive

resting ECG. This ECG Standard is the result of an

EU supported project that European, American and

Japanese Manufacturers and Users have jointly

worked and agreed on (1989-1990). In 1993 it

became a European ENV, later was positively

balloted within AAMI (AAMI EC71), and finally

became a European EN at the beginning of 2005.

In 2002 the FDA launched the need of having

the full disclosure digital waveforms submitted for

the support of clinical trials with a flexible XML

schema and a rich set of annotations. The main

American manufacturers thus defined the so-

called FDA XML Data Format Design Specification

that was an FDA XML-based specification covering

the design for the waveform data format as well as

the relevant submission information. This

specification of rising popularity, known also as

Annotated ECG in XML, also became a part of HL7

V.3 currently balloted (HL7 aECG).

OpenECG is a world-wide network supporting

interoperability in electrocardiography through the

consistent implementation of standards. In 2007 it

has about 850 members from more than 60

countries. The development of open source

converters among ECG formats is supported and

encouraged by OpenECG.

At the beginning of 2006 a international working

group formed by people with different expertise was

created by the OpenECG network, with the support

of IEEE 1073 and CEN and the coordination of TU

Graz and Biosig, for the development of an open

source two way converter in C++ between the SCP-

ECG and the HL7 aECG standards. In the

conversion, GDF, the BioSig format was used as an

intermediate form. ECG data sets available in the

OpenECG portal (http://www.openecg.net) were

used to test the converter in different environments

included Linux and Cygwin.

Once the first version was available, there were

some issues that remained open and most of them

were related to an incomplete mapping between the

two standards. This information was an important

retrofit for the relevant Standard Developing

Organizations and some actions in order to solve

these open issues have already been done. In fact, a

harmonization of the ECG lead standard

terminology between the two standards has already

been done leading to the creation of the SCP-ECG

amendment (EN1064:2005+A1:2007) and a similar

revision for the HL7 aECG standard.

The converter has been released as open source

and is currently available in the Biosig Sourceforge

site (http://biosig.sf.net/). Figure 1 shows SCP data

that has been converted into the GDF v1 data format

(Schlögl et al. 2007b).

2.2 BioSig for BCI Research

Brain computer interfacing is one topic closely

related to EEG processing that needs of techniques

capable to work under on-line conditions. Besides,

efficient methods for artifact rejection and/or

correction, online feature extraction techniques,

classifiers, single trial analysis and performance

measurements are important issues in this field.

A Brain Computer Interface (BCI) consists in

general of 4 modules: EEG pre-processing, feature

extraction, classification and feedback. Biosig

provides useful tools for on-line artefact processing;

BIOSIGNALS 2008 - International Conference on Bio-inspired Systems and Signal Processing

404

Figure 1: Screenshot of SigViewer showing 10s of 8-channel ECG data. The smaller “window” shows the editable event

table. The data set was converted from the SCP (EN1064) to the GDF v1 (Schögl et al. 1999b) data format using

BioSig4C++.

several on-line feature extractors are also available,

such as adaptive autoregressive parameters or band

power estimates. It also provides many classifiers

(including but not limited to LDA, QDA/MDA,

SVM, NBC, etc) and several single trial analysis

methods to test the performance of systems/subjects.

It can be used to provide initial conditions to all

these modules before starting a BCI on-line session

(see also for the section “rtsBCI below), in which

the system is in general tuned for the subject. Also,

it is especially useful for the analysis of

experimental BCI data. BioSig was the reference

tool for the development of on-line adaptive

classifiers (which were tested in BCI experiments)

(Vidaurre et al. 2006).

2.3 Artifact Processing and Quality

Control

Biomedical signal recordings are often contaminated

by various artifacts. BioSig provides several tools to

address this issue. This include tools for (i) quality

control and determining the saturation values of the

recording systems are provided (Schlögl et al

1999a), (ii) a fully automated reduction methods of

EOG artifacts in EEG recording (Schlögl et al

2007a), and (iii) inverse filtering for detecting

muscle artifacts (Schlögl et al 2000).

2.4 Coupling Analyzing

In order to investigate the interaction and coupling

between brain areas, various coupling measures like

coherency, phase, partial coherence, partial directed

coherence (PDC), directed transfer function (DTF)

etc. can be used. As shown in Schlögl and Supp

(2006), all these coupling measures can be derived

from a multivariate autoregressive (MVAR) model.

The MVAR estimator (Schlögl, 2006a) can be also

applied to data with missing values, thus the various

coupling measures can be obtained from data with

missing values, too. BioSig supports also a non-

parametric statistical analysis using a jackknife

procedure (Efron, 1981) for estimating the

confidence intervals of all the coupling measures.

BIOSIG - Standardization and Quality Control in Biomedical Signal Processing Using the BioSig Project

405

2.5 QRS Detection and HRV Analysis

An important area of biomedical signal processing is

analysing the electrocardiogram (ECG). BioSig

contains several algorithms for QRS detection

(Nygard et al, 1983, Afonso et al 1999), detecting

extrasystoles and other irregular detections (Mateo

and Lugano, 2003) and analyzing the heart rate

variability (Taskforce, 1996). Furthermore, the

efficient algorithm of Berger et al. (1986) for an

equidistant sampling of the heart rate is

implemented.

2.6 rtsBCI

The Graz-BCI open source software package rtsBCI

provides a framework for the development and rapid

prototyping of real-time BCI systems. The software

is based on Matlab/Simulink (The Mathworks, Inc,

Natick, MA, USA) running on Microsoft Windows

(Microsoft Corporation, Redmond, WA, USA) and

licensed under the GNU GPL. For hard real-time

computing and the generation of stand-alone C code

the Real-Time Windows Target (RTWT) and the

Real-Time Workshop (RTW), respectively, are

required. Both toolboxes are extensions of Simulink.

Furthermore, BioSig for Octave and Matlab is

needed for data format handling, and TCP/UDP/IP

toolbox for network communication support.

Additionally to these software requirements, a data

acquisition device is indispensable.

After installation, all rtsBCI modules are listed

in the Simulink Library Browser and can be used to

design (model) the BCI system. Several Matlab

functions and Simulink blocks for (i) data

acquisition and conversion, (ii) storage, (iii) digital

signal processing (e. g. band power feature

estimation, Split-Radix discrete Fourier transform,

adaptive autoregressive parameters (AAR) estimated

with Kalman filtering, linear discriminant analysis,

etc.), (iv) visualization (e. g. signal scope,

presentation of cue information or feedback of a

moving bar), (v) paradigm control (cue-based and

self-paced operation mode) and (vi) network support

(e. g. remote monitoring) are available.

Tunable parameters as well as other information

relevant for the experiment (e.g. subject information,

amplifier settings, electrode setup, paradigm timing)

are stored in an individual configuration file (.INI

file). Before a model is executed, the configuration

is transferred to the model and stored altogether with

the biosignals for further analysis. The division of

model and parameters makes it very easy to deal

with changes: For example, a new classifier requires

only the replacement of the classification block. A

new subject requires only the modification of the

related data in the configuration file.

Modular architecture and rapid prototyping

allow a fast extension and incorporation of new

software as well as hardware components. This

flexibility is a big advantage as is the fact that

Matlab is very popular. The period of vocational

adjustment is reduced, as well as the costs, because

only a reduced number of toolboxes are required.

2.7 SigViewer

SigViewer is a powerful stand-alone viewing and

scoring program for biosignals, originally designed

to process electroencephalogram (EEG) signals.

SigViewer has among its features the ability to load

multi-channel signals such as EEG, ECG, EMG, and

EOG recordings, and display these in various scales.

At the moment, only GDF v1 (Schlögl et al. 1999b)

is supported, but as a workaround, users can convert

other data formats to GDF using the function

“save2gdf” (available in BioSig4OctMat and

BioSig4C++).

Figure 1 shows a screenshot of

SigViewer displaying ECG data.

The other major capability besides the viewing

functions is the scoring of biosignals, which permits

the user to make various annotations to the signals

(e.g. mark segments as artifactious, mark specific

events, like QRS-complexes, etc) and save this

information into a file.

It is also possible to view basic information

about a specific file (e.g. number of channels,

sampling frequency, number of events, time of

recording, and so on). In addition to graphically

scoring the data, the event table is available as a list-

based widget for viewing and deleting events and

annotations (

Figure 1).

SigViewer is written in C++ using the open-

source platform-independent graphical user interface

(GUI) toolkit Qt 4 (Trolltech

). SigViewer runs

under many different operating systems such as

Linux, Windows and Mac OS X – in other words, it

is designed to be platform-independent (or more

accurately cross platform). Moreover, it does not

depend on any proprietary software, making it a

truly free program. The source code does not have to

be changed when compiling binaries for specific

platforms, it is enough to take one and the same

source tree and compile it on the target platform.

BIOSIGNALS 2008 - International Conference on Bio-inspired Systems and Signal Processing

406

2.8 Analysis of Cardiac Near-Field

(CNF) Signals

An ongoing research project is the investigation of

the cardiac near-field (CNF) signal. The

spatiotemporal electrical activation at the surface of

heart tissue is assumed to propagate with a smooth

elliptical wave front. Micro-obstacles like embedded

connective tissue may affect the smoothness of the

wave front which results in complex activation

sequences at microscopic size scale. The

investigation of these mechanisms by analysis of the

local activation in microscopic dimensions is

expected to gain deeper insight into structure-related

arrhythmias. For this purpose electrophysiological

in-vitro experiments have been carried out using

autorhythmic or electrically stimulated heart tissue

preparations from Rabbits or Guinea Pigs.

The gradient of electric potential at the cardiac

surface, the Cardiac Near-Field (CNF), can be

computed from four extra-cellular potentials Φ

..Φ

recorded with ultra-densely placed electrodes

(electrode spacing 50µm). Such a sensor with an

appropriate data acquisition system (sampling rates

of 100 kHz per channel) has been developed

recently (Hofer et al. 2006). It has been shown that

the multivariate signal Φ= [Φ

, Φ

]

can be

used to determine local parameters of the

propagating electrical activation, namely velocity

and direction (Plank et al. 2000). Currently, research

is aimed on developing robust procedures for the

calculation of these parameters and the evaluation of

their accuracy (Wiener et al. 2007). In practice, the

two major problems are: First, the acquired signals

are affected by inherent electrode noise and by

stimulus artifacts. Second, in case of structural

discontinuities in the underlying tissue, Φ may be a

composition of multiple local and distal electrical

activation sequences. Therefore, the formation of the

CNF in case of normal and complex activation

sequences in the tissue has been extensively

investigated in computer simulations (Plank et al.

2003).

The BioSig toolbox for Matlab is being used for

off-line analysis of waveforms of Φ obtained from

experiments and computer simulations because it

provides comprehensive procedures for time-

frequency analysis of the multivariate stochastic

process Φ. Moreover the implemented tools can deal

with missing values due to the removal of stimulus

artifacts. Developed signal processing procedures

are being validated by applying them to noise-free

waveforms Φ obtained from computer simulations.

2.9 Miscellaneous

Many smaller algorithms are also included.

Examples are the so-called “Paynter filter” (Bruce et

al. 1977, Platt et al. 1998) to estimate the envelope

of EMG power. There are also many EEG

parameters like Hjorth parameters (Hjorth, 1975),

Barlow parameters (Goncharova and Barlow, 1990),

a global linear descriptor (Wackerman, 1999) and

the brainrate parameter (Pop-Jordanova and Pop-

Jordanov, 2005) supported. Furthermore, methods

for multiple statistical tests for avoiding the problem

of alpha inflation are supported (Hemmelmann et al.

2005), and many different plotting functions for

EEG analysis, like the visualization of Coherence

according to Nolte et al. (2004).

3 RESULTS

BioSig addresses all aspects of biomedical signal

processing, starting with the support for over 40

different data formats, quality control and artifact

processing, methods for signal processing and

feature extraction classification of single trial EEG,

and statistical tests including the multiple

comparison problem. Currently, the main application

areas are research on EEG-based Brain-Computer

Interfaces, coupling analysis of EEG/ECoG/MEG,

processing of EEG artifacts, conversion of different

data formats.

BioSig provides reference implementations of

many biomedical signal processing algorithms and

for many application areas including, EEG, ECoG,

MEG, ECG and HRV analysis, Brain Computer

Interface research, analysing brain connectivity. The

software algorithms can be copied, used, modified

and distributed under the terms of the GNU GPL

(http://www.gnu.org/copyleft/gpl.html). The open

source software library for biomedical signal

processing BioSig is available from

http://biosig.sf.net.

An open source converter between ECG

standardized ECG data formats SCP-ECG (EN1064)

and HL7aECG and several other data formats is

available. Future plans include the development of a

common data format for all biomedical signals.

BIOSIG - Standardization and Quality Control in Biomedical Signal Processing Using the BioSig Project

407

Table 1: Ranking of the Biosig project at SoureForge

among various application areas. The 2

column shows

the ranking of BioSig and the total the number of projects

for each application area.

Topic Ranking 2007-10-22

/ number of projects

SourceForge 456 / 160 049

Biosignals (keyword

search: “EEG, ECG”)

1 / 27

Medical Science

applications

8 / 543

Human Machine

Interfaces

2 / 612

Dataformats 20 / 2139

The BioSig software library is widely adopted.

Currently, the download rate is far beyond 600 per

month and increasing. As of Oct 2007, BioSig is the

highest ranked project for biomedical signal

processing (search term “EEG ECG”) at

SourceForge http://sourceforge.net, a platform that

hosts over 160 000 open source projects. Within the

last two years (Sep 2005 – Sep 2007), the monthly

ranking fluctuated between 2906 (Jan 2007) and 380

(Aug 2007), the overall rank is within the top 2% of

all hosted projects. Besides SourceForge, parts of

BioSig have been incorporated in other projects (e.g.

EEGLab http://www.sccn.ucsd.edu/eeglab/), which

is not considered in the above statistics.

4 DISCUSSION

Although BioSig is routinely used in several

application areas like BCI research, data conversion,

coupling analysis, etc. there are several topics which

are not or only suboptimally supported. Examples

for the current limitations are the following.

(i) Two components (rtsBCI and SViewer)

require proprietary software (Simulink and Matlab).

It would be desirable to have a fully open source

solution without this requirement. SViewer is going

to be replaced by SigViewer, but it is still useful

because SViewer supports more data formats.

Nevertheless, many users do have Matlab anyway,

therefore it is reasonable to distribute these tools.

(ii) The situation on the viewing and scoring

software is not perfect. The SViewer requires the

proprietary Matlab software and is relatively slow,

SigViewer supports only very few data formats.

(iii) The conversion between different data

formats is not always perfect, and can lose some

information (demographic data, annotations, see also

Schlögl et al. 2007b). For this reason, it is important

to unify the various data formats for biomedical

signal processing.

(iv) Support for many specialized application

areas (like advanced ECG analysis, ...) depend on

the contribution and evaluation of expert users. In

order to maintain the growth of BioSig with the aim

to become “the” software library for biomedical

signal processing, participation of users and experts

of the various areas of biomedical signal processing

is crucial. We think of advanced ECG analysis (P-

and T-wave detection, classification of arrhythmias),

the source localization problem in EEG analysis, or

analyzing the activity of spiking neurons.

5 CONCLUSIONS

The BioSig project contains many software tools for

biomedical signal processing. Because BioSig

provides an open source software library, there is no

need to “re-invent the wheel”, but the existing

software algorithms can be used and improved.

These algorithms provide reference

implementations, and can be validated and improved

by everyone. This is an efficient mechanism for

standardization and quality control of software for

biomedical signal processing.

The aim of BioSig providing a software library

for biomedical signal processing has been already

reached, BioSig is also the #1 project for biomedical

signal processing on Sourceforge. The open question

is not whether but how BioSig can help integrating

biosignal analysis into the health information

system.

ACKNOWLEDGEMENTS

This work was supported by the EU grant

“BrainCom” (FP6-2004-Mobility-5 Grant No

024259) and the Austrian Science Fund FWF P-

19993-N15.

REFERENCES

Afonso, V., Tompkins, W., Nguyen, T., Luo S. 1999. ECG

beat detection using filter banks. IEEE Trans. Biomed.

Eng. 46(2):192-202.

BIOSIGNALS 2008 - International Conference on Bio-inspired Systems and Signal Processing

408

Berger R.D., Akselrod S., Gordon D., Cohen R.J. 1986.

An efficient algorithm for spectral analysis of heart

rate variability. IEEE Trans Biomed Eng. 33(9):900-4.

Bruce, E. N., M. D. Goldman, and J. Mead. A digital

computer technique for analyzing respiratory muscle

EMGs. J. Appl. Physiol. 43: 551-556, 1977

Efron, B. 1981. "Nonparametric estimates of standard

error: The jackknife, the bootstrap and other methods",

Biometrika, 68, 589-599.

EN1064 2005. Health informatics. Standard

communication protocol. Computer-assisted

electrocardiography

Goncharova, I.I. and Barlow, J.S. 1990. Changes in EEG

mean frequency and spectral purity during

spontaneous alpha blocking. Electroencephalogr Clin

Neurophysiol. 76(3):197-204.

Hemmelmann C, Horn M, Suesse T, Vollandt R, Weiss S.

2005. New concepts of multiple tests and their use for

evaluating high-dimensional EEG data. J Neurosci

Methods. 142(2):209-17.

Hjorth, B. 1975. Time Domain Descriptors and their

Relation to particulare Model for Generation of EEG

activity. in G. Dolce, H. Kunkel: CEAN Computerized

EEG Analysis, Gustav Fischer, p.3-8.

Hofer, E., Keplinger, F., Thurner, T., Wiener, T., Sanchez-

Quintana, D., Climent, V., Plank, G., 2006. A new

floating sensor array to detect electric near fields of

beating heart preparations. In Biosens. Bioelectron.,

vol. 21, pp 2232-2239.

Mateo, J., Laguna, P. 2003. Analysis of Heart Rate

Variability in Presence of Ectopic Beats Using the

Heart Timing Signal. IEEE Transactions on

biomedical engineering, 50(3): 334-343.

Nolte G, Bai O, Wheaton L, Mari Z, Vorbach S, Hallett

M. 2004. Identifying true brain interaction from EEG

data using the imaginary part of coherency. Clin

Neurophysiol.115(10):2292-307.

Nygards, M.-E. and Sörnmo, L. 1983. Delineation of the

QRS complex using the envelope of the e.c.g.-Med. &

Biol. Eng. & Comput., 21, 538-547.

Plank, G., Hofer, E., 2000. Model study of vector-loop

morphology during electrical mapping of microscopic

conduction in cardiac tissue. In Ann. Biomed. Eng.,

vol. 28, pp 1244-1252.

Plank, G., Vigmond, E., Leon, L.J., Hofer, E., 2003.

Cardiac near-field morphology during conduction

around a microscopic obstacle – a computer

simulation study. In Ann. Biomed. Eng., vol. 31, pp

1206-1212.

Platt, R.S., Hajduk, E.A., Hulliger, M., Easton,P.A. 1998.

A modified Bessel filter for amplitude demodulation

of respiratory electromyograms. J. Appl. Physiol.

84(1): 378-388.

Pop-Jordanova, N. and Jordan Pop-Jordanov J. 2005.

Spectrum-weighted EEG frequency ("Brainrate") as a

quantitative indicator of arousal Contributions, Sec.

Biol. Med. Sci., MASA, XXVI, 2: 35 - 42

Schlögl, A., Kemp, B., Penzel, T., Kunz, D., Himanen, S.-

L., Värri, A., Dorffner, G., Pfurtscheller, G., 1999a.

Quality Control of polysomnographic Sleep Data by

Histogram and Entropy Analysis. Clin. Neurophysiol.

110(12): 2165 - 2170.

Schlögl, A., Filz, O., Ramoser, H., Pfurtscheller, G.

1999b. GDF version 1 - A general dataformat for

biosignals, available at:

http://www.dpmi.tugraz.at/schloegl/matlab/eeg/gdf4/T

R_GDF.pdf

Schlögl, A. 2000. The electroencephalogram and the

adaptive autoregressive model: theory and

applications. Shaker Verlag, Aachen, Germany.

Schlögl, A. 2006a. Comparison of Multivariate

Autoregressive Estimators. Signal processing. 86(9):

2426-9.

Schlögl, A. 2006b. GDF version 2 - A general dataformat

for biosignals. http://arxiv.org/abs/cs.DB/0608052

Schlögl, A. and Supp, G. 2006. Analyzing event-related

EEG data with multivariate autoregressive parameters.

(Eds.) C. Neuper and W. Klimesch, Event-related

Dynamics of Brain Oscillations. Analysis of dynamics

of brain oscillations: methodological advances.

Progress in Brain Research 159: 135 – 147.

Schlögl, A., Keinrath, C., Zimmermann, D., Scherer, R.,

Leeb, R., Pfurtscheller, G. 2007a. A fully automated

correction method of EOG artifacts in EEG

recordings. Clin. Neurophys. 118(1):98-104.

Schlögl, A., Chiarugi, F., Cervesato, E Apostolopoulos,

E., Chronaki, C.E.. 2007b. Two-Way Converter

between the HL7 aECG and SCP-ECG Data Formats

Using BioSig.

Conference on Computers In Cardiology,

Taskforce Heart Rate Variability: Standards of

Measurement, physilogical interpretation and clinical

use. Taskforce of the European Society for Cardiology

and the North American Society of Pacing and

Electrophysiology. European Heart Journal (1996)

17, 354-381.

Vidaurre, C., Schlögl, A.,Cabeza, R., Scherer, R.

Pfurtscheller, G. 2006. A fully on-line adaptive BCI,

IEEE Trans. Biomed. Eng., 53(8): 1214–1219.

Wackermann, J. 1999. Towards a quantitative

characterization of functional states of the brain: from

the non-linear methodology to the global linear

descriptor. International Journal of Psychophysiology,

34: 65-80.

Wiener, T., Thurner, T., Prassl, A.J., Plank, G., Hofer, E.,

2007. Accuracy of local conduction velocity

determination from non-fractionated cardiac activation

signals. In EMBC’07, 29th Annual International

Conference of the IEEE Engineering in Medicine and

Biology Society.

BIOSIG - Standardization and Quality Control in Biomedical Signal Processing Using the BioSig Project

409