re : events of dataset for motor imagery provided in physionet.org

samhitha_95
Posts: 16
Joined: 18 Aug 2016, 02:09

re : events of dataset for motor imagery provided in physionet.org

Post by samhitha_95 » 18 Aug 2016, 02:15

We could not find the annotations (T0,T1,T2) for the event types present in .edf file but we found those annotations in the .event files provided along with the edf file . However , the event types in both these files do not match . I am providing the link to these datasets below. Could someone please help us resolve this issue asap. Thank you .

https://www.physionet.org/pn4/eegmmidb/

pbrunner
Posts: 344
Joined: 17 Sep 2010, 12:43

Re: re : events of dataset for motor imagery provided in physionet.org

Post by pbrunner » 18 Aug 2016, 11:24

Samhitha,

I assume you are using the EEGLab or FieldTrip version of read_edf. This implementation will give you the error message that the "EDF Annotations" have been skipped, which results in the problem that you have described.
Warning: Skipping "EDF Annotations" as continuous data channel because of inconsistent sampling frequency (80 Hz)
As a workaround you can use a different function to read the EDF+ files, e.g., the edfread function from MATLAB Exchange. I have verified that this allows you to load the 65th channel of the data that contains the EDF Annotations.

https://www.mathworks.com/matlabcentral ... 00-edfread

Alternatively, you can look at the EDF+ website for other Viewers. I would suggest that you verify that the different viewers/import methods provide you with the same data:
http://www.edfplus.info/downloads/

Let me know if this resolves the issues.

Regards, Peter

samhitha_95
Posts: 16
Joined: 18 Aug 2016, 02:09

Re: re : events of dataset for motor imagery provided in physionet.org

Post by samhitha_95 » 19 Aug 2016, 00:25

Thanks a lot for the reply . We were using wfdb toolbox to view the event types in .event files and then when we realized that the event types from .event were not matching with those in .edf ( which we saw in epoch extraction step of EEGlab ) . So , we initially thought that the discrepancy could be due to viewing both the files in different readers ( .event in wfdb and .edf in EEGlab ). So we switched to wfdb toolbox completely to view both the files and found the discrepancy persisted . So we are presuming that it is not because of the reader , there is a discrepancy , but the files themselves don't match . Could you please confirm if this could be the case . Thank you.
PS: I have tried reading the .edf file with the link you mentioned above . It gives the same annotations as in EEGlab which is different from the annotations in .event file

pbrunner
Posts: 344
Joined: 17 Sep 2010, 12:43

Re: re : events of dataset for motor imagery provided in physionet.org

Post by pbrunner » 19 Aug 2016, 10:59

Samhitha,

we initially provided the data without the .event files, so I assume that physionet.org created the event files. Looking at these event files, they indeed are wrong. For that reason I would recommend that you disregard them. We will contact physionet.org to resolve this issue. In the meantime I can offer you two workarounds:

(1) You use the annotations in the EDF+ file.
(2) I can provide you with the original BCI2000 data files.

Regards, Peter

samhitha_95
Posts: 16
Joined: 18 Aug 2016, 02:09

Re: re : events of dataset for motor imagery provided in physionet.org

Post by samhitha_95 » 20 Aug 2016, 02:13

Thank you very much . Could you please mail the original data files to
f2012646@hyderabad.bits-pilani.ac.in

pbrunner
Posts: 344
Joined: 17 Sep 2010, 12:43

Re: re : events of dataset for motor imagery provided in physionet.org

Post by pbrunner » 22 Aug 2016, 13:28

Samhitha,

please see your inbox for the link to the data.

Regards, Peter

rita_alkhoury
Posts: 3
Joined: 12 Nov 2016, 16:47

Re: re : events of dataset for motor imagery provided in physionet.org

Post by rita_alkhoury » 13 Nov 2016, 13:07

prof. Peter Brunner
please can you send me the original data to rita.kh1993@gmail.com
I have the same probleme :(

pbrunner
Posts: 344
Joined: 17 Sep 2010, 12:43

Re: re : events of dataset for motor imagery provided in physionet.org

Post by pbrunner » 13 Nov 2016, 19:42

Rita,

please see your inbox for the link to the data.

Regards, Peter

MagdielCAS
Posts: 1
Joined: 13 Dec 2016, 22:26

Re: re : events of dataset for motor imagery provided in physionet.org

Post by MagdielCAS » 13 Dec 2016, 22:29

prof. Peter Brunner

please can you send me the original data as soon as possible to magdielc.a.sousa@alu.ufc.br
I have the same problem :(

kamalakshat
Posts: 1
Joined: 24 Feb 2017, 06:24

Events of dataset for motor imagery provided in physionet.org

Post by kamalakshat » 24 Feb 2017, 06:32

Hello everyone,

I have downloaded the Matlab file from Physionet ATM. The 65th channel shows different numbers as 12331, 5140, 11008, etc. and a trail of zeroes at some points. How to relate these with the events. How to find the one for left and the one for right?

Thanks,

Kamal

zitazsi
Posts: 1
Joined: 01 Apr 2017, 03:45

Re: re : events of dataset for motor imagery provided in physionet.org

Post by zitazsi » 01 Apr 2017, 03:50

prof. Peter Brunner

Could you please send me the original data to zitazsi94@gmail.com ?

I have the same problem.

robertleng
Posts: 1
Joined: 04 Apr 2017, 05:29

Re: re: events of dataset for motor imagery provided in physionet.org

Post by robertleng » 04 Apr 2017, 05:40

Hi Prof. Peter Brunner and everyone,

I downloaded the motor imagery data set from physionet.org, but use the mne python toolkit to read the data.

So my question is, what do the annotations in channel 65 of the .edf file correspond to?

The annotations I see are 0, 1, 2 and 3. I am able to tell that 0 marks transition between stimuli, 1 indicates rest. I am not sure what 2 and 3 indicates.
I've looked at the physio.net page, but they use T0,T1,T2 annotations, and no information can be found for 0,1,2,3 annotations.

Thank you!

Robert

rkaygupta
Posts: 3
Joined: 06 Nov 2017, 09:57

Re: Events of dataset for motor imagery provided in physionet.org

Post by rkaygupta » 06 Nov 2017, 10:36

kamalakshat wrote: 24 Feb 2017, 06:32 Hello everyone,

I have downloaded the Matlab file from Physionet ATM. The 65th channel shows different numbers as 12331, 5140, 11008, etc. and a trail of zeroes at some points. How to relate these with the events. How to find the one for left and the one for right?

Thanks,

Kamal
I have the same doubt, can anyone clarify/explain?And why are they in the baseline data, because there is not cue involved there. Also, if the recording is of 2 min(for baseline) and sampling rate is 160 Hz then no. of data points=9600 , we have 9760 , is it implied that we ignore the points after 9600?
Thanks a lot

romanbara
Posts: 1
Joined: 23 Apr 2018, 08:01

Re: re : events of dataset for motor imagery provided in physionet.org

Post by romanbara » 23 Apr 2018, 08:02

Hello Prof. Peter Brunner
please can you send me the original data to romanbaravalle@gmail.com

I notice I have the same problem.

Thanks,
Román

charlotte
Posts: 1
Joined: 02 Jul 2018, 09:18

Re: re : events of dataset for motor imagery provided in physionet.org

Post by charlotte » 02 Jul 2018, 09:23

Dear Prof. Peter Brunner
I downloaded the motor imagery data set from physionet.org。I have the problem to read the .edf.event files. If it in your convenient to send me the original data,I would appreciate.
MY email 2051352986@qq.com

Post Reply

Who is online

Users browsing this forum: No registered users and 22 guests