EEG.pl

After 70 years of clinical applications of electroencephalography, visual analysis still remains the only accepted method. Application of signal processing methods in clinical electroencephalography is explicitly discouraged [1]. Clinical neuroscience faces a tremendous inter-subject variability. Given also the variety of methods--and their flavors--applied for the analysis of EEG/MEG, a coherent progress may be achieved only via comparison of the same algorithms on large amount of datasets from different laboratories. Our story begins in 1994, when the matching pursuit algorithm was proposed by mathematicians [2]. Authors provided a free implementation, which made possible instant applications to EEG. After a decade, it led to a thesis about the methodological unification of visual and computer EEG analysis [3]. From the very beginning, we shared our implementations of the algorithm and other relevant software. Around 2001, when the idea of creating Polish neuroinformatics network was put forward by prof. Andrzej Wróbel, we created EEG.pl: "an open repository for software, publications and datasets related to the analysis of brain potentials". Unfortunately, sole availability of complete algorithms with source code does not imply the dissemination of methodology within the main target audience of clinical neurophysiologists. Therefore, signalml.org hosts a large project of an open system for EEG display and analysis, which offers access to the advanced methods of signal processing in a commercial-grade, user-friendly environment. It features an open interface for incorporating advanced signal processing methods, while the inherent incompatibility of various digital EEG formats is addressed via the SignalML metadescription [4]. On the low level, it provides the first user friendly interface to the matching pursuit decomposition of signals. Higher level functions, implemented via the interface to the Matlab(R) code, include automatic detection of artifacts and sleep stages in polysomnographic recordings.