SEV – a software toolbox for large scale analysis and visualization of polysomnography data

SEV is a graphical toolbox designed in MATLAB for displaying polysomnography (PSG) signals recorded during sleep studies, prototyping signal-processing algorithms and automating sleep feature extraction methods across large collections or cohorts of such studies. Format imported are European Data Formats and event/hypnogram files. Time-series analysis can be performed using a suite of classifiers, filters and signal decomposition tools (e.g. wavelets) developed internally or implemented from validated methods published by others. Power spectral analysis can be performed using either periodogram averaging or multiple spectrum independent component analysis. The tool is highly configurable and provides a simple framework for classifier optimization and extensibility. MATLAB's parallel processing toolbox is utilized during batch processing. Output formats include MySQL database entry, tab-delimited text and MATLAB archive (.MAT). The tool is well suited for genetic or epidemiological sleep research questions requiring rigorous, robust and reproducible evaluation of a PSG-based sleep study cohort. Current built-in applications include modules to detect and quantify rapid eye movements and spindle activity (using existing algorithms), inter-channel electroencephalography coherence and a detector developed in house to quantify periodic leg movements during sleep. SEV is open source and freely available under a common creative license.