How to identify, detect, classify the Concentration

HanHan Republic of Korea
edited May 2023 in Ganglion

Hi, OpenBCI!

We are going to use Focus Widget to identify the concentration and then send the user's concentration to the Unity3D project through Network Widget to proceed with the game.

As we know that brain waves related to concentration are well collected from the frontal lobe, we placed the electrodes in the positions AF7, Fp1, Fpz, Fp2, and AF8.

I'm looking at FFT, Focus Widget, Band Power, and I'm especially looking at Band Power's Beta, Gamma waves.

Resting, watching videos, or solving problems does not show any pattern in Band Power.

Is there a proper way to identify the concentration?

Thank you always.

Comments

  • Beta or gamma band power as measured in frontal regions of the scalp can be considered a lossy function that discards all of cognition except whether one is awake.

    To get the information you want to extract properly, you might need to measure gamma at the brain surface or even within the brain, with something such as implanted electrodes. The signal/noise ratio at the usual distance from brain to scalp surface electrode is otherwise far too small.

  • wjcroftwjcroft Mount Shasta, CA

    This type of classification is still an active research subject. As can be seen from the range of papers:

    https://www.google.com/search?q=detecting+concentration+from+eeg

    In the academic publishing field, a "review paper" is one that surveys the published research in a particular area, in a fairly exhaustive manner. Attempting to find patterns and consensus. In that search above, I came across at least one such review, below. It is focused on 'attention' vs 'concentration', but I believe there is significant overlap in those terms.

    https://www.frontiersin.org/articles/10.3389/fphys.2021.727840/full
    "Attention Detection in Virtual Environments Using EEG Signals: A Scoping Review"
    Rhaíra Helena Caetano e Souza1,2* and Eduardo Lázaro Martins Naves1
    The competitive demand for attention is present in our daily lives, and the identification of neural processes in the EEG signals associated with the demand for specific attention can be useful to the individual’s interactions in virtual environments. Since EEG-based devices can be portable, non-invasive, and present high temporal resolution technology for recording neural signal, the interpretations of virtual systems user’s attention, fatigue and cognitive load based on parameters extracted from the EEG signal are relevant for several purposes, such as games, rehabilitation, and therapies. However, despite the large amount of studies on this subject, different methodological forms are highlighted and suggested in this work, relating virtual environments, demand of attention, workload and fatigue applications. In our summarization, we discuss controversies, current research gaps and future directions together with the background and final sections.


    I also assume you have read the material on the Focus Widget, and how the metric was derived. The page also advises there are different variants of the ML that can be selected.

    https://docs.openbci.com/Software/OpenBCISoftware/GUIWidgets/#focus-widget

    Since the ML models were trained independent of the current subject, one would expect that fact to strongly influence accuracy of the widget.

    https://brainflow.readthedocs.io/en/stable/UserAPI.html?highlight=mindfulness#brainflow-ml-model

    Note also that Brainflow API itself uses the term 'mindfulness' vs 'concentration' or 'attention'. I don't know how these models were trained, but in meditation disciplines such as vipassana and TM, there are distinct differences in object-less meditation, remaining in the present moment, and concentration based practices. If you have further questions on how the Brainflow metrics were trained, I suggest asking on the Brainflow Slack, which you can sign up for here:

    https://brainflow.org/

    William

  • HanHan Republic of Korea
    1. Can I create the functions of Focus Widget (Concentration, Relaxing) provided by the OpenBCI GUI as a Java program?
    2. Can I get a code example of collecting actual EEG values with Ganglion Board and then applying FFT using BrainFlow SDK?
    3. "Concentration Index = (SMR + Mid Beta) / Theta" Is this formula appropriate for identifying a concentration?
    4. Is there a way to figure out Concentration or Relaxing without using the OpenBCI GUI?

    Very Thanks Open BCI Team.

  • wjcroftwjcroft Mount Shasta, CA

    Hi @Han, I merged your new thread into your previous thread on the same subject.

    @Han said:
    1. Can I create the functions of Focus Widget (Concentration, Relaxing) provided by the OpenBCI GUI as a Java program?

    Yes, Java examples are on the Brainflow docs site.

    https://brainflow.readthedocs.io/en/stable/Examples.html#java-eeg-metrics

    1. Can I get a code example of collecting actual EEG values with Ganglion Board and then applying FFT using BrainFlow SDK?

    https://brainflow.readthedocs.io/en/stable/Examples.html#java-band-power
    https://brainflow.readthedocs.io/en/stable/UserAPI.html#java-api-reference [search for 'fft' in this page]

    1. "Concentration Index = (SMR + Mid Beta) / Theta" Is this formula appropriate for identifying a concentration?

    My hunch is that this is going to be too crude, as it is looking at the entirety of beta activity, not selectively.

    1. Is there a way to figure out Concentration or Relaxing without using the OpenBCI GUI?

    https://brainflow.readthedocs.io/en/stable/Examples.html#java-eeg-metrics

    Regards, William

  • wjcroftwjcroft Mount Shasta, CA

    @wjcroft said:
    ... [from previous March 29 comment]
    I also assume you have read the material on the Focus Widget, and how the metric was derived. The page also advises there are different variants of the ML that can be selected.

    https://docs.openbci.com/Software/OpenBCISoftware/GUIWidgets/#focus-widget

    Since the ML models were trained independent of the current [specific, unique] subject, one would expect that fact to strongly influence accuracy of the widget.

    https://brainflow.readthedocs.io/en/stable/UserAPI.html?highlight=mindfulness#brainflow-ml-model

    ... If you have further questions on how the Brainflow metrics were trained, I suggest asking on the Brainflow Slack, which you can sign up for here:

    https://brainflow.org/

    As an addendum to the March 29 comments quoted above, I would suggest you investigate how you could TRAIN the Brainflow ML 'mindfulness' ('concentration') models to specifically work with you or your preferred subjects. On sample scenarios where you ask the subjects to engage in the desired task.

    So to facilitate that, post questions on the Brainflow Slack, where the developer Andrey is responding.

    Regards,

Sign In or Register to comment.