SignalML 2.0
Filed under:
General neuroinformatics
Zbigniew Jędrzejewski-Szmek (Faculty of Physics, University of Warsaw), Aleksander Michalski (Nencki Intitute of Experimental Biology), Kamil Góźdź (Ericpol), Jakub Hałun (Ericpol), Grzegorz Stadnik (Ericpol), Joanna Jędrzejewska-Szmek (Faculty of Physics, University of Warsaw), Paweł Kordowski (Faculty of Physics, University of Warsaw), Piotr Durka (Faculty of Physics, University of Warsaw)
SignalML is an XML-based language for description of biomedical time series formats, originally proposed in [1]. Independent of platform and programming language, it is a meta-format which describes other formats. SignalML acts as an import/export interface for other software: descriptions of formats are written once, without the need to write actual code to handle separate formats or employ SignalML in software written in different languages.
SignalML 2.0 treats each file as a source of parameters (data of the 'header-type': sampling rate, number of channels, date of recording, etc.) and raw signal data. Format description provides information about the location of parameters within files, description of other parameters to be calculated using those read from files, and of the layout and format of samples. Notable new features of SignalML include:
* a specification of the way that a description of a format in SignalML should be interpreted. This includes the most basic algebraic expressions, allowing to create a SignalML support library in any programming language,
* generalization of the descriptions:
o the ability to specify multiple files, with individual formats,
o generalization of layouts due to a configurable mapping function calculating sample positions,
o support for more binary representations (big-endian, little-endian, different widths, hexadecimal and octal prefixes in text representation),
o ability to use convenient algebraic expressions referring to other parameters wherever a value must be provided,
SignalML tries to reuse existing standards and conventions:
* XML is used as the lexical layer of SignalML;
* XPath is used to extract information from other XML files;
* the expression language is largely based on Python;
* regular expressions are Perl-compatible;
* variable types (int, float, string, bytes) and the textual representation of numbers are taken directly from Python 3.x;
*bit layout of values in binary files is described using dtype from NumPy;
*units are represented using standard SI notation, including Greek prefixes.
Svarog (Signal Viewer, Analyzer and Recorder on GPL, http://svarog.pl) is using an implementation of SignalML 2.0.
[1] SignalML: metaformat for description of biomedical time series. Durka, P. J. and Ircha, D. Computer Meth Prog Biomedicine 2004, vol. 76, pp. 253-259
SignalML 2.0 treats each file as a source of parameters (data of the 'header-type': sampling rate, number of channels, date of recording, etc.) and raw signal data. Format description provides information about the location of parameters within files, description of other parameters to be calculated using those read from files, and of the layout and format of samples. Notable new features of SignalML include:
* a specification of the way that a description of a format in SignalML should be interpreted. This includes the most basic algebraic expressions, allowing to create a SignalML support library in any programming language,
* generalization of the descriptions:
o the ability to specify multiple files, with individual formats,
o generalization of layouts due to a configurable mapping function calculating sample positions,
o support for more binary representations (big-endian, little-endian, different widths, hexadecimal and octal prefixes in text representation),
o ability to use convenient algebraic expressions referring to other parameters wherever a value must be provided,
SignalML tries to reuse existing standards and conventions:
* XML is used as the lexical layer of SignalML;
* XPath is used to extract information from other XML files;
* the expression language is largely based on Python;
* regular expressions are Perl-compatible;
* variable types (int, float, string, bytes) and the textual representation of numbers are taken directly from Python 3.x;
*bit layout of values in binary files is described using dtype from NumPy;
*units are represented using standard SI notation, including Greek prefixes.
Svarog (Signal Viewer, Analyzer and Recorder on GPL, http://svarog.pl) is using an implementation of SignalML 2.0.
[1] SignalML: metaformat for description of biomedical time series. Durka, P. J. and Ircha, D. Computer Meth Prog Biomedicine 2004, vol. 76, pp. 253-259
Preferred presentation format:
Poster
Topic:
General neuroinformatics
Latest news for Neuroinformatics 2011
Follow INCF on Twitter
