An EMG signal (EMG.edf)

Description

The first file (EMG.edf), contains an EMG segment at minimum effort. EMG signals typically consume a lot of resources because they have to be sampled at high rates. The present recording is continuous, so it could have been coded as an EDF file. It includes ten data records. Each data record can be seen as equivalent to a sweep of EMG in raster mode. The duration of each data record is 100 ms. Since we use EDF+, we need an EDF Annotations signal that is maintained empty.

The header

This is the content of the global header record

Version	0
Patient	X M 29-JAN-1956 X
Recording	Startdate 01-JAN-2002 EMG jop N1 L_Biceps_Brachii
Startdate	01.01.02
Starttime	09.30.00
Bytes in header	768
Reserved	EDF+C
Data records	10
Duration	0.100
Signals	2

This is the content of the signal header record

	First signal	Second signal
Label	EMG L_Biceps_Bra	EDF Annotations
Transducer	Concentric electrode
Phys. dimension	uV
Phys. minimum	-1000	0
Phys. maximum	1000	1
Dig. minimum	-2000	-32768
Dig. maximum	2000	32767
Prefiltering	HP:3Hz LP:10kHz
Samples	2000	500
Reserved

We can see that we have 2000 samples for each data record, so the sampling rate is 20 kHz.

The traces

The content of the first data record can be seen in figure 2.1

Figure 2.1: Plot of the record 0

The annotations

Let's see the annotations of each data record (each line is the annotation included in one data record). They only include an empty stamp indicating the beginning of each data record

 
+0.00[20][20]
+0.10[20][20]
+0.20[20][20]
+0.30[20][20]
+0.40[20][20]
+0.50[20][20]
+0.60[20][20]
+0.70[20][20]
+0.80[20][20]
+0.90[20][20]

The sign [20] indicates the decimal character 20 (hexadecimal 14).

Some comments

Notice that in this case the coding in EDF+ is very similar to the coding of the same signal in EDF. The main difference is the introduction of an EDF Annotation signal marking the time. The conversion from EDF files to EDF+ files is made by adding this signal and by making some minor modification in the header.

We could have chosen a sampling rate of the annotation signal of 10 and the same time stamp could have been included. We preferred the sampling of 500 allowing a free space to code some a posteriori processing (e.g., detecting the different Motor Unit Potentials appearing in the signal), to mark the conditions of the acquisition (e.g., spontaneous activity, weak effort) or detail the procedure (e.g., modification of needle insertion).