Technical report A new way of building a database of EEG ®ndings H. Aurlien * , I.O. Gjerde, N.E. Gilhus, O.G. Hovstad, B. Karlsen, H. Skeidsvoll Haukeland University Hospital, Section of Clinical Neurophysiology, Department of Neurology, University of Bergen, Norway. Accepted 1 February 1999 Abstract Whereas computer-based electroencephalography (EEG) is widely applied, the EEG interpretations are usually not stored in a way that favours exploitation of modern computer technology. This paper reports an EEG description system facilitating categorization of EEG data in a computerized database. The system interactively communicates with the digital EEG system and also with the general patient admin- istrative system. The main new quality of this system is the methods for data input and automatic data retrieval from several systems, rather than the establishment of a database of EEG data itself. The EEGs are visually analysed and categorized. Manually marked EEG events are automatically transferred to the database and such events as well as de®ned electrode positions within these epochs are directly linked to their corresponding descriptions. The database is updated without demand for ®lling in the events in the database in a second operation. Thereby, the EEG interpreter builds the database while analysing the EEG. This system provides an improved accessibility of EEG data for clinical, normative, educational and scienti®c use. q 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Electroencephalography; Databases; Computer-systems; Medical-records systems; Reference standards; Automatic data processing 1. Introduction In recent years, computer-based EEG has been widely applied. Digital techniques have several advantages compared with the previous, paper-based systems (Burgess, 1993; Nuwer, 1997; Blum, 1998; Quinonez, 1998; Swartz, 1998). For the electroencephalographer (EEGer), it gives the opportunity for a better evaluation through user-selected montages, vertical and horizontal scaling, and ®lter adjust- ments. For the laboratory, it means a far better storage and faster retrieval of EEGs. Digital storing of EEGs results in vast amounts of digital signals as well as textual informa- tion. Functionality and ¯exibility in data retrieval depend on how they are structured and standardized. Usually, the data concerning EEG interpretations are stored in non-structured and non-standardized free text, and the EEG tests are designed for retrieving one by one only. Already the EEG pioneers developed comprehensive databases/archives with structured and categorized EEG data (Gibbs et al., 1943). Gibbs and Gibbs (1950) describe a record ®ling system for routine EEGs using carbon paper to obtain quadruplicates of the reports and then organising them in a name ®le as well as in diagnosis ®les according to both electroencephalographic and clinical diagnosis. Schwab (1951) describes an even more advanced system using punch cards where the tests could be retrieved accord- ing to clinical diagnosis, patient demographic data, various EEG parameters or combinations of these. However, even using modern computers the routines demanded to establish such systems can be inconvenient and time consuming. This is possibly why they are not available in most modern EEG systems. Thus, retrospective EEG studies still depend on the manual ®ltering and analysis of EEG reports (Maher et al., 1995; Dean et al., 1997; Bautista et al., 1998; Dury and Beydoun, 1998). The parameters needed for advanced computerized retrieval are not included in current Interna- tional Federation of Clinical Neurophysiology (IFCN) stan- dards (Nuwer et al., 1998). The challenge is to take advantage of the opportunities given by the new compu- ter-based technology. The EEGer should be enabled and encouraged to effectively structure and standardize the EEG interpretation and then store the categorized data in a computerized EEG database. To avoid the need for dupli- cate manual entering of information, external available databases relevant for these purposes should be linked to the EEG database. Such linking of databases would provide an easier and more convenient access to key data and give a continuous overview of the laboratory's patient population. All previous instances of speci®c diagnoses or EEG events or groups of these could be retrieved, improving educational programs, scienti®c studies and quality-control measures. Clinical Neurophysiology 110 (1999) 986±995 CLINPH 98724 1388-2457/99/$ - see front matter q 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S1388-2457(99)00037-1 * Corresponding author. Tel.: 1 47-55-975-000, fax: 1 47-55-975-164. E-mail address: harald.aurlien@haukeland.no (H. Aurlien)