Biosig Package

Biosig is a toolbox for processing biomedical signals like ECG, EEG etc. It can read about 50 different data formats including EDF, GDF, BDF, CFS, Heka, ABF, HL7aECG, SCP (EN1064), BrainVision, and many more. The heavy lifting is done in "libbbiosig" the biosig library implemented in C/C++. The Python module of Biosig provides an interface between libbiosig and Python. Currently, two functions are provided, one for reading the data samples, and one for extracting the meta information, like event table, scaling factors, sampling rates, etc.

More information is available at

• Biosig project homepage
• Biosig wiki page

Installation:

GNU/Linux (Debian, Ubuntu, and family)

  sudo apt install python3-biosig

or if you want to install in your own venv

  sudo apt install libbiosig-dev  # make sure that libbiosig and its prerequisites are in place
  pip install biosig

MacOSX/Homebrew

  brew install biosig             # make sure libbiosig is in place.
  pip install biosig

MS-Windows (Python >=3.9,<=3.14)

  pip install biosig

Other *NIX (Linux, Unix) OS

• install libbiosig from source (configure && make && make install)
• pip install biosig

Usage

  import biosig
  import json
  import tomllib

read header/metainformation

  HDR=json.loads(biosig.header(FILENAME))

If some text fields are not ASCII/UTF8 encoded, this might not work. If that is the case, use

	CODEX = 'ISO-8859-15'
	HDR = json.loads(biosig.jsonheader(FILENAME,CODEX))

or

	CODEX = 'ISO-8859-15'
	tomlHDR = biosig.tomlheader(FILENAME,CODEX)
	H2 = tomllib.loads(tomlHDR)
	H2['CHANNEL']

read data

  DATA=biosig.data(FILENAME)

A more elaborate example is this demo

showing

• how to load data
• displaying the traces with matplotlib
• splitting the traces into sweeps

History

In the past there were several attempts of providing Python support. A first attempt using pure python should to be very slow and a lot of implementation effort, only some early version of GDF has been supported. Later, Swig was used for providing a python binding. This was reasonable flexible, and the libbiosig implementation could be mostly used. Disadvantages were the effort to maintain swig.i interface file, which is currently most likely broken. Known issues are string support, lack of Python3 support. Also this attempt is considered deprecated.

The third attempt is using "module extensions". Currently, two functions, one for reading the header information (in JSON format) and one for reading the data samples is provided.