Read EDF Bio-Electricity Datafile

I was studying about sleeping steps, using this cap-sleeping dataset: https://physionet.org/content/capslpdb/1.0.0/ as reference.

However I encountered some problems decoding the EEG sequencial datafile which ends with *.edf.

This datasets contains several kinds of files, including *.edf, *.m, *.xlsx, *.st, *.txt. The following explainations are about these types of files:

---

*.edf:

Multi-Channels raw-signal datafile, and the full name of edf was European Data Format, whose name should be widely known in the realm of signal-processing. For python users, we can use pyEDFlib to analyze this file type.

https://pyedflib.readthedocs.io/en/latest/

pyEDFlib is a python library to read/write EDF+/BDF+ files based on EDFlib.

EDF means [European Data Format](http://www.edfplus.info/) and was firstly published [1992](http://www.sciencedirect.com/science/article/pii/0013469492900097). In 2003, an improved version of the file protokoll named EDF+ has been published and can be found [here](http://www.sciencedirect.com/science/article/pii/0013469492900097).

The EDF/EDF+ format saves all data with 16 Bit. A version which saves all data with 24 Bit, was introduces by the compony [BioSemi](http://www.biosemi.com/faq/file_format.htm).

The definition of the EDF/EDF+/BDF/BDF+ format can be found under [edfplus.info](http://www.edfplus.info/).

This python toolbox is a fork of the [toolbox from Christopher Lee-Messer](https://bitbucket.org/cleemesser/python-edf/) and uses the [EDFlib](http://www.teuniz.net/edflib/) from Teunis van Beelen. The EDFlib is able to read and write EDF/EDF+/BDF/BDF+ files.

*.st: 

The scores for each recording in PhysioBank-compatible format.

*.txt:

The scores for each recording in REMlogic report format.

The .txt score files have the following fields:

• Sleep stage (W=wake, S1-S4=sleep stages, R=REM, MT=body movements)
• Body position (Left, Right, Prone, or Supine; not recorded in some subjects)
• Time of day [hh:mm:ss]
• Event (either a sleep stage (SLEEP-S0..S4, REM, MT), or a phase A of CAP)
• Duration (in seconds)
• Location (the signal(s) in which the event can be observed)

*.m:

  Matlab scripts. for reading the .txt score files easily. By launching ScoringReader.m, an array containing the scoring of the macrostructure, and three arrays containing the starting time of the CAP A phases, their duration and their subtype, are generated. The hypnogram is reported as 0 for wake, 1 to 4 for the sleep stages, according to R&K rules, and 5 for REM. The Matlab function CAP.m computes, starting from these variables, the CAP time and CAP rate according to Terzano’s rules.

gender-age.xlsx:

report of age and gender of the subjects. The first row is what is wrong with the subjects. Here is a list. These names also showed in txt files.

---

Since we get the definitions, we can use tools to step out for reaching the benchmark.

for python user, use pip to install the edf analyze library:

pip install pyedflib

Then run the test code to draw a fragment of one sequencial data file:

>>> import pyedflib
>>> import numpy as np
>>> file_name = r"\\192.168.1.103\ssd\cap-sleep-database-1.0.0\n1.edf"
>>> f = pyedflib.EdfReader(file_name)
>>> n = f.signals_in_file
>>> f
<pyedflib.edfreader.EdfReader object at 0x000001AEDC0488D8>
>>> n
21
>>> signal_labels = f.getSignalLabels()
>>> signal_labels
['ROC-LOC', 'LOC-ROC', 'F2-F4', 'F4-C4', 'C4-P4', 'P4-O2', 'F1-F3', 'F3-C3', 'C3-P3', 'P3-O1', 'C4-A1', 'EMG1-EMG2', 'ECG1-ECG2', 'TERMISTORE', 'TORACE', 'ADDOME', 'Dx1-DX2', 'SX1-SX2', 'Posizione', 'HR', 'SpO2']
>>> len(signal_labels)
21
>>> sample = f.getNSamples()
>>> len(sample)
21
>>> sample
array([17725440, 17725440, 17725440, 17725440, 17725440, 17725440,
       17725440, 17725440, 17725440, 17725440, 17725440,  8862720,
       17725440,  2215680,  2215680,  2215680,  8862720,  8862720,
        2215680,    34620,    34620])
>>> f.readSignal(1)
array([ -28.96970238,  -19.93384158,   -8.21164377, ..., -244.94509158,
       -242.44191392, -241.61769689])
>>> signal_1 = f.readSignal(1)
>>> len(signal_1)
17725440
>>> import matplotlib.pyplot as plt
>>> plt.scatter([_ for _ in range(10000)], signal_1[:10000])
<matplotlib.collections.PathCollection object at 0x000001AEEF75BC50>
>>> plt.show()

 

Now with the labels, you can run test on this dataset.