EEG Database for classification purpose

[Hanna (Hanna Carla)]

Hi, i’m using this database CHB-MIT Scalp EEG Database v1.0.0 (physionet.org)] on my master’s research project. I’m trying to use the annotations files but it’s in the edf.seizure extension. Does anyone ever worked with these files and knows how to open?

[drammock (Dan McCloy)]

I looked at the dataset. all the .seizures files are only 54 bytes long, they don’t contain any actual annotations. I think they exist only to indicate which recordings contained seizures, not to mark where in the file the seizures occurred. You can read them with MNE-Python, but only if you change them to have an .edf extension. for example, this shell command:

mv chb01_03.edf.seizures chb01_03.seizures.edf

renames the file, and then in python:

import mne
mne.read_annotations('chb01_03.seizures.edf')

will return <Annotations | 0 segments>

If you have further questions about the dataset you should probably contact the authors / dataset uploaders directly.

[ZeenatFirdosh]

Thank you!!
I wil work on it and will try implement what you said…

[Teuniz (Teunis van Beelen)]

Please have a look here:

how to load .edf.seizure file of CHB-MIT dataset

These files are in “MIT annot” format. The annotation files belonging to EDF files which contain a seizure, have at least two annotations which point to the beginning and the end of the seizure. (for example file chb05_06.edf.seizures) After opening the EDF file in EDFbrowser, go to Tools → Import annotations/events and select the tab MIT/WFDB to import the annotations from that annotations file. Click on the import button and in the file dialog, set the filter to “all files”. The descriptio…