Erpdb: an Integrated Database of Erp Data for Neuroinformatics Research

Event-related potential (ERP) is the measurement of the brain's electrical activity in response to different types of events, such as attention, words, thinking, or sounds. By measuring the brain's response to such events, we can learn how different types of information are processed. As the mass of recorded ERP data explodes, an automatic and accurate tool to store, manage, and retrieve data readily is of increasing concern in neuroinformatics. In this paper, we describe a relational ERP database that has been constructed using the SQL server 2000 database management system and an IIS web server that has been setup for data retrieval through a custom web interface (http://202.113.232.103:8088/erpdb/index.asp). A novel database structure has been used to store ERP data of different activity channels, which provides a rapid and accurate way for data retrieval within any given range on the time zone with various searching options. The database is divided into: (1) subjects' information and record information and (2) ERP data, which has been structured and standardized in a database table supplemented with unrestricted text files. It can integrate or exchange data with other clinical databases or computer-based information systems through a program based on ADO techniques. Users are able to readily retrieve ERP data through the user-friendly web page interface. All online resources of the database are freely available to the scientific community. As the database develops further, we anticipate it will become a valuable tool that will make a great contribution to everyday clinical practice, teaching, and research work inneuroscience and psychology in the future. ERPs, one of the most important kinds of electrophysiological data, are the measurement of the brain's electrical activity (EEG signals) in response to different types of events, such as attention, words, thinking, or sounds-usually external stimuli or subject's movements (Picton & Hillyard, 1988). ERPs may either precede an event or follow it. They consist of a sequence of positive and negative voltage-fluctuation-labeled components, which are useful as measures of covert information processing (differences between conditions can be obtained through comparison to ERP in the absence of behavioral responding) (David & Ray, 2000).


Event-Related Potentials (ERPs)
ERPs, one of the most important kinds of electrophysiological data, are the measurement of the brain's electrical activity (EEG signals) in response to different types of events, such as attention, words, thinking, or soundsusually external stimuli or subject's movements (Picton & Hillyard, 1988).ERPs may either precede an event or follow it.They consist of a sequence of positive and negative voltage-fluctuation-labeled components, which are useful as measures of covert information processing (differences between conditions can be obtained through comparison to ERP in the absence of behavioral responding) (David & Ray, 2000).

S743
One of the main reasons for using ERPs in psychology is that they present an intriguing possibility to obtain information about how the intact human brain processes signals and prepares for actions.Cognitive ERP has been used as a marker of cognitive function in patients with psychiatric and neurological disorders (Katada, Sato, et al., 2004).In contrast to overt behavioral measures and slow brain imaging techniques such as functional magnetic resonance, ERP reflects the information processing in the cortex in real time.Although little is known about neural correlation and functional meaning of ERP components, even this limited knowledge has led to inferences that could never be made if only the final results of many processes were recorded.
ERP signals are small (1-30 microvolt) and are embedded in electroencephalographic (EEG) activity unrelated to the eliciting event.However, they can be recorded non-invasively from the scalp by means of signal averaging techniques.The size (signal-to-noise ratio) and period (the component's frequency) of the ERP component can decide the number of trials required to produce a noise-free ERP.
There are three measurable aspects of the ERP waveform: amplitude, latency, and scalp distribution (Johnson, 1998).Component amplitude provides an index of neural activation extent (how the component responds functionally to experimental variables); component latency (the point in time at which the peak occurs) reveals the timing of this activation; and a component's scalp distribution (the pattern of voltage gradient over the scalp at any point in time) provides information on the overall pattern of activated brain areas.

ERP Research-The requirement of database
The neuroscience research on ERP results in substantial amounts of digital signals and textual information.As clinical techniques develop rapidly, the accumulating ERP data have created one of the largest datasets in biological science.Functional capability and flexibility in data retrieval depend on how they are structured and standardized.Usually, the raw data, together with additional information or interpretations, are stored in non-structured and non-standardized free text.In addition, the ERP tests are only designed for retrieving one by one (Aurlien, Gjerde, Gilhus, et al., 1999).As a result, an automatic and accurate tool (database) to store, manage and retrieve data readily is of increasing concern in neuroinformatics.
Databases offer accurate and timely data and clinical details for neuroscience research (Finnerup, Fuglsang-Frederiksen, Rossel, et al., 1999).Several neuroscience databases and computer-based information systems have been established previously.One example is Aurlien et al, who report an EEG description system facilitating categorization of EEG data in a computerized database (Aurlien, et al., 1999).Another example is the epilepsy and EEG information system, developed as part of a European multicentre project PRESTIGE.The aim of PRESTIGE is to bridge the gap between research and everyday clinical practice by deploying healthcare telematics technology for guideline generation, dissemination, and routine application (research-based and consensus-based) (Finnerup, et al., 1999)  [6] .In addition, in the field of neural disease, Babiloni has set up a database online with respect to the Alzheimer disease (Babiloni, Babiloni, Carducci, et al., 2001).The Brain Resource International Database, supported by the Brain Resource Company in Australia, is the world's first integrated brain function database.It contains 33 dimensions of brain function through using standardized positions to identify the best biomarkers in a number of Brain diseases (http://www.brainresource.com/).Many research efforts drew data from the database (Hermens, Michael, & Simon, 2005;Brickman, Paul, & Cohen, 2005;Clark, Veltmeyer, & Hamilton, 2004;Hermens, Eleonore, & Simon, 2005), but databases for ERP Data Science Journal, Volume 6, Supplement, 13 October 2007 S744 are still scarce.Even aspects such as how ERP raw data as well as the textual information can be structured and standardized are lacking.The purpose of this project is to develop an online standardized relational database of ERP data along with textual information about descriptions and interpretations.This database will also be used for collection of ERP data.

DESIGN AND IMPLEMENTATION
The ERPDB Database has been developed and implemented using the SQL Server 2000 relational database system.Having specially extended full support to XML (Extensible Markup Language), SQL Server 2000 has the flexibility, accessibility, and security functions required by data storage components of web sites.Because of its unparalleled advantage, SQL Server 2000 has already established a leading position as an integrated solution for database and data analysis (Yang, 2004).
An IIS web server has been set up for data retrieval through a custom web interface using ASP and VBScript language.Our server is running on the Windows 2000 Advanced Server operating system rather than other operating systems such as Unix or Linux.First, the SQL Server 2000 must be based on a Windows Server Series operating system.In addition, any other platforms to store and analyze data are based on Windows, such as Microsoft Excel, Compaq Visual Fortran, Matlab, Microsoft Visual Studio, and so on.The tools and integrated development environment operated on Windows are more available, compared to similar products operated on Unix or Linux.Thus it is more convenient and efficient for biology or medicine researchers not experienced in information technology.
The ERP datasets used to populate this database are in usual textual style, e.g.part of the data is provided by Dr. Sue Surguy, King's College, University of London.A program has been developed to read those textual data and convert them into the database with the standardized relational structure.The program is developed with the VB6 language and connects to the database by ADO techniques.As an object-oriented visualization-programming tool, Microsoft Visual Basic is easy to learn, flexible, and convenient.Microsoft ActiveX Data Objects (ADO), as the COM interface based on OLE DB, is one of the most important application interfaces (API).Through the encapsulation to OLE DB API, ADO provides consistent and rapid data access, supporting various developmental needs, including the establishment of applications, tools, phylum, database clients, and business objects of the Internet.The main advantage of ADO is that it is simple and fast, requiring relatively less memory and less disk space (Xia & Fanxing, 2005).The program can be used for any other ERP raw data to convert them into the database.Since ADO can be used not only in connecting SQL Server database, but also in connecting databases based on many other database management systems, this program can retrieve data from many other databases and convert them into one database.This is important for data transferring or data sharing between different information centers.

DATA CONTENT
The ERPDB database is divided into two parts: (1) subjects' primary and record information and (2) ERP data, which has been structured and standardized in a database table, supplemented with unrestricted text files.Three tables are designed for the data.Data Science Journal, Volume 6, Supplement, 13 October 2007

Table
SubjectsInfo stores the information of every subject in ERP recording experiments including person's