This project has moved. For the latest updates, please go here.

Remove blinks from raw data

Mar 12, 2012 at 11:37 PM

Hi!

I wanted to ask if it`s possible to remove/replace automaticaly blinks and partial blinks from raw data.

There is this option that shows blinks during video so I thought it`s possible. What is that option "eliminate data of this trail from statistical analyses" for?

If there is the possibility to remove blinks, which procedure is used?

Is there something like a handbook for OGAMA? I think some of the moduls are not described in the documentation yet.

 

greeting,

Ewald

 

 

Coordinator
Mar 13, 2012 at 9:23 AM

Hi Ewald,
what are ýou excatly intending to do?
Ogama is handling blinks and missing data the following way -> if there is no data for a specific time stamp (-> DBNull entry in database) this timestamp is just handled as it was not there. If the database entry is (0,0) it is handled as a blink. For fixation calculation this is handled within the fixation calculation algorithm (-> see documentation at \Ogama\Modules\Common\Tools\FixationDetection.cs) so normally there is no need to "remove" blinks... what does this mean?

The "eliminate data" options is useful, if you have trials that are obviously miscalibrated, so they are not useful for statistical analysis. With this button you are able to mark them as "bad trial", so they are ignored for statistical calculations.

The only documentation is the one at www.ogama.net and the source code documentation supplied with the installation. Sorry, we are not enough people to write a professional documentation... you are invited to participate!

Regards,
Adrian

Mar 23, 2012 at 3:04 PM

Sorry, forgot to sign the e-mail answering and saw your reply only now.

I want to measure cognitive load from eye dilation. One step to do this is to detect blinks and partial blinks and delete or interpolate them.

Do I understand correct. Ogama deletes measures when it does not "see" the pupil.

But there are zeros in the raw data for both eyes. Is the raw data unfiltered? see picture http://www.flickr.com/photos/78271679@N07/6862437032

 

Greetings,

Ewald

 

 

 

 

Mar 23, 2012 at 3:08 PM

one thing i forgot.

There are many dublicates (seems every measurepoint is doubled). What does that mean?

I think this issue was already discussed but i can't find it any longer.

 

Greetings Ewald

 

Coordinator
Mar 24, 2012 at 7:29 PM

Hi Ewald,

> Do I understand correct. Ogama deletes measures when it does not "see" the pupil.
The data in the database is unfiltered, it stays in the database just as it comes from the hardware devices during recording or import. So it does not deletes anything. The fixation detection algorithm itself does care for missing samples as it tries to ignore them as described in the code annotations. This all is true for the gaze position samples. The pupil measurements are not handled in any way in Ogama at the moment, they are just recorded and written to the database.

I have never seen the duplicates issue before. By default time duplicates are written only, if a gaze sample and an event occur at the same time. So it would be interesting with which device under which cirumstances you get these duplicates.

Regards, Adrian

Mar 24, 2012 at 9:05 PM

Hi Adrian,

Ok fine, thats good.

I tested with a Tobii x120 at 120herz. It was the same slideshow I already sended you. If you want I send you also the pics.

 

Greetings,

Ewald

Coordinator
Mar 24, 2012 at 9:22 PM

Does the duplicates occur on every recording with the Tobii or just occasionally?

Mar 24, 2012 at 9:41 PM

They were not ever there. I have one recording that has no dublicates but since then it is always happening. It may have something to do with that strange timeformat I have now. Measures two digits after the comma of millisecounds. Before it was only millisecounds

24923
24923