search

sccn / ICLabel

Automatic EEG IC classification plugin for EEGLAB
https://iclabel.ucsd.edu/tutorial
53 stars 19 forks source link

Is there a way to download the original dataset? #9

Closed ghost closed 4 years ago

ghost commented 4 years ago

Hi,

First of all, congratulations on this project! :)

I was wondering if the ICs + labels were also available to download. I have a few ideas for new features and would like to test them, but I would need the original ICs and their corresponding labels. I imagine it might be a huge dataset! Is there any way to access the original dataset?

Thanks!

arnodelorme commented 4 years ago

@lucapton can the dataset be downloaded?

lucapton commented 4 years ago

There is a dataset available but not with the raw IC time series. It includes labels and a number of extracted features from each component. If you look at my other repositories, it’s called ICLabel-dataset and includes a python class for loading and managing the dataset. You’ll have to then download a number of files containing the features and labels. I need to fix the documentation because the automatic downloads have been giving people issues.

On Jun 11, 2020, at 15:35, rafaelgm notifications@github.com wrote:

 Hi,

First of all, congratulations on this project! :)

I was wondering if the ICs + labels were also available to download. I have a few ideas for new features and would like to test them, but I would need the original ICs and their corresponding labels. I imagine it might be a huge dataset! Is there any way to access the original dataset?

Thanks!

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

ghost commented 4 years ago

Hi, thanks for your responses. I already have the dataset with the labels and features (around 25GB), but in order to create new features I would need the raw IC time series. Are they available for download, or will they be available in the future?

lucapton commented 4 years ago

I understand but unfortunately they are not available and likely will not be. The ICs were extracted from datasets on a living server so their current state is very different than they were originally. It was was in feasible at the time to make a full duplicate of the server for preservation. Furthermore, doing so may have violated any number of IRB guidelines as the datasets come from many different experiments, each with their own IRB agreement.

I understand this is likely disappointing.

On Jun 16, 2020, at 09:09, rafaelgm notifications@github.com wrote:

 Hi, thanks for your responses. I already have the dataset with the labels and features (around 25GB), but in order to create new features I would need the raw IC time series. Are they available for download, or will they be available in the future?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

ghost commented 4 years ago

Sure, it's perfectly understandable. Thanks, anyway. :)