tueg-tools
Tools for working with the Temple University Hospital EEG (TUHEEG) Corpus
Download EEG data
The Temple group provide various scripts/instructions for downloading their dataset, but in our experience some users struggle to make these work, and there are known limitations. Hence tueg-tools includes a (hopefully) easy-to-use, platform-independent method for downloading the full dataset (TUEG), a specific subset (defined by root URL), or some size-limited portion of the specified subset for anyone with basic python and git skills.
Just clone this repo, cd to its folder and then in python...
import tueg_tools
ds = tueg_tools.Dataset('C:/Your_path_to_where_data_should_be_downloaded')
ds.download('https://www.isip.piconepress.com/projects/tuh_eeg/downloads/tuh_eeg_abnormal/',
username='Your Username', password='Your Password', maxSize=10**11)Note that in the above code, the optional argument maxSize=10**11 limits the download to the
first 100 GB (10^11 bytes).
You will need to insert your own username and password. You can obtain these by requesting access from the maintainers of the TUHEEG Corpus, as described on their downloads page.
The URL specified in this example points to the TUAB subset. You can specify the URL for any of the TUHEEG subsets* and this function will walk through their sitemap and reproduce the folder structure on your machine (stopping early if maxSize is specified).
*Hopefully. Only TUAB and TUEG have been tested.
Iterating over the dataset
The tueg_tools.Dataset class includes various generator methods to simplify iterating over multiple sessions/recordings in the dataset. Look through the code to get a full idea of what's been implemented so far. As a basic example:
ds = tueg_tools.Dataset('C:/Your_Path_to_where_data_was_downloaded')
for eeg in ds.eeg_gen():
# Do something with each of the EEG records in this dataset.