{
  "id": 471502,
  "title": "What are the EEG electrode references?",
  "url": "/competitions/hms-harmful-brain-activity-classification/discussion/471502",
  "author_name": "",
  "post_date": "2024-01-28T14:55:28.325535900Z",
  "votes": 3,
  "comment_count": 1,
  "views": 0,
  "content": "<p>The electric potential measured by each of the electrodes must be in comparison to some other point on the head (see [<a href=\"https://www.youtube.com/watch?v=AcW97nMLGEs&amp;ab_channel=JeremyMoeller])\" target=\"_blank\">https://www.youtube.com/watch?v=AcW97nMLGEs&amp;ab_channel=JeremyMoeller])</a>. The reference one chooses has advantages and disadvantages based on what an expert's needs are. However, in the EEG files the headers for each electrode are simply 'F7' or 'O1', rather than 'F7-T3' or 'O1-O2'. The competition description does not elaborate on which reference scheme is being used as far as I saw. This information would be beneficial for both context and for computing hand-crafted features. </p>\n<p>Of course, regardless of the reference and assuming it is the same for all electrodes, if we make new features by taking the difference between the EEG traces we have the unknown reference will cancel out, i.e. 'F7 - X' - 'T3 - X' = 'F7 - T3'. Have the competition hosts stated anywhere what the unknown reference is?</p>",
  "messages": [
    {
      "id": "2624068",
      "postDate": "01/28/2024 14:55:28",
      "content": "<p>The electric potential measured by each of the electrodes must be in comparison to some other point on the head (see [<a href=\"https://www.youtube.com/watch?v=AcW97nMLGEs&amp;ab_channel=JeremyMoeller])\" target=\"_blank\">https://www.youtube.com/watch?v=AcW97nMLGEs&amp;ab_channel=JeremyMoeller])</a>. The reference one chooses has advantages and disadvantages based on what an expert's needs are. However, in the EEG files the headers for each electrode are simply 'F7' or 'O1', rather than 'F7-T3' or 'O1-O2'. The competition description does not elaborate on which reference scheme is being used as far as I saw. This information would be beneficial for both context and for computing hand-crafted features. </p>\n<p>Of course, regardless of the reference and assuming it is the same for all electrodes, if we make new features by taking the difference between the EEG traces we have the unknown reference will cancel out, i.e. 'F7 - X' - 'T3 - X' = 'F7 - T3'. Have the competition hosts stated anywhere what the unknown reference is?</p>",
      "rawMarkdown": "The electric potential measured by each of the electrodes must be in comparison to some other point on the head (see [https://www.youtube.com/watch?v=AcW97nMLGEs&ab_channel=JeremyMoeller]). The reference one chooses has advantages and disadvantages based on what an expert's needs are. However, in the EEG files the headers for each electrode are simply 'F7' or 'O1', rather than 'F7-T3' or 'O1-O2'. The competition description does not elaborate on which reference scheme is being used as far as I saw. This information would be beneficial for both context and for computing hand-crafted features. \n\nOf course, regardless of the reference and assuming it is the same for all electrodes, if we make new features by taking the difference between the EEG traces we have the unknown reference will cancel out, i.e. 'F7 - X' - 'T3 - X' = 'F7 - T3'. Have the competition hosts stated anywhere what the unknown reference is?",
      "votes": null
    },
    {
      "id": "2624841",
      "postDate": "01/29/2024 03:13:13",
      "content": "<p>The online reference/GND (the one used during the EEG recording) usually does not show up in the data as a channel since that would mean amplifying and recording a 0 PD. What you can do (and what the expert in the video is saying) is use a new offline reference so that you can compare activity in different regions of the brain. </p>\n<p>What most people in the competition have been doing is use the Anterior-Posterior Bipolar Montage (one of the montages described in the video) to create 4 chains of electrodes (for a total of 16 new signals) and then create spectrograms to compare brain activity in 4 different regions. If you take a look at any of the example_figures PDFs you will see that this is pretty much what the competition's hosts did, and it makes sense since most if not all the activities we are trying to classify are lateralized.</p>\n<p>I recommend you reading these:<br>\n<a href=\"https://pressrelease.brainproducts.com/referencing/\" target=\"_blank\">https://pressrelease.brainproducts.com/referencing/</a><br>\n<a href=\"https://www.acns.org/UserFiles/file/ACNSStandardizedCriticalCareEEGTerminology_rev2021.pdf\" target=\"_blank\">https://www.acns.org/UserFiles/file/ACNSStandardizedCriticalCareEEGTerminology_rev2021.pdf</a></p>",
      "rawMarkdown": "The online reference/GND (the one used during the EEG recording) usually does not show up in the data as a channel since that would mean amplifying and recording a 0 PD. What you can do (and what the expert in the video is saying) is use a new offline reference so that you can compare activity in different regions of the brain. \n\nWhat most people in the competition have been doing is use the Anterior-Posterior Bipolar Montage (one of the montages described in the video) to create 4 chains of electrodes (for a total of 16 new signals) and then create spectrograms to compare brain activity in 4 different regions. If you take a look at any of the example_figures PDFs you will see that this is pretty much what the competition's hosts did, and it makes sense since most if not all the activities we are trying to classify are lateralized.\n\nI recommend you reading these:\nhttps://pressrelease.brainproducts.com/referencing/\nhttps://www.acns.org/UserFiles/file/ACNSStandardizedCriticalCareEEGTerminology_rev2021.pdf",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2624841,
      "author_name": "gabrielribeirocsilva",
      "author_url": "",
      "post_date": "01/29/2024 03:13:13",
      "content": "<p>The online reference/GND (the one used during the EEG recording) usually does not show up in the data as a channel since that would mean amplifying and recording a 0 PD. What you can do (and what the expert in the video is saying) is use a new offline reference so that you can compare activity in different regions of the brain. </p>\n<p>What most people in the competition have been doing is use the Anterior-Posterior Bipolar Montage (one of the montages described in the video) to create 4 chains of electrodes (for a total of 16 new signals) and then create spectrograms to compare brain activity in 4 different regions. If you take a look at any of the example_figures PDFs you will see that this is pretty much what the competition's hosts did, and it makes sense since most if not all the activities we are trying to classify are lateralized.</p>\n<p>I recommend you reading these:<br>\n<a href=\"https://pressrelease.brainproducts.com/referencing/\" target=\"_blank\">https://pressrelease.brainproducts.com/referencing/</a><br>\n<a href=\"https://www.acns.org/UserFiles/file/ACNSStandardizedCriticalCareEEGTerminology_rev2021.pdf\" target=\"_blank\">https://www.acns.org/UserFiles/file/ACNSStandardizedCriticalCareEEGTerminology_rev2021.pdf</a></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2624068": "The electric potential measured by each of the electrodes must be in comparison to some other point on the head (see [https://www.youtube.com/watch?v=AcW97nMLGEs&ab_channel=JeremyMoeller]). The reference one chooses has advantages and disadvantages based on what an expert's needs are. However, in the EEG files the headers for each electrode are simply 'F7' or 'O1', rather than 'F7-T3' or 'O1-O2'. The competition description does not elaborate on which reference scheme is being used as far as I saw. This information would be beneficial for both context and for computing hand-crafted features. \n\nOf course, regardless of the reference and assuming it is the same for all electrodes, if we make new features by taking the difference between the EEG traces we have the unknown reference will cancel out, i.e. 'F7 - X' - 'T3 - X' = 'F7 - T3'. Have the competition hosts stated anywhere what the unknown reference is?",
    "2624841": "The online reference/GND (the one used during the EEG recording) usually does not show up in the data as a channel since that would mean amplifying and recording a 0 PD. What you can do (and what the expert in the video is saying) is use a new offline reference so that you can compare activity in different regions of the brain. \n\nWhat most people in the competition have been doing is use the Anterior-Posterior Bipolar Montage (one of the montages described in the video) to create 4 chains of electrodes (for a total of 16 new signals) and then create spectrograms to compare brain activity in 4 different regions. If you take a look at any of the example_figures PDFs you will see that this is pretty much what the competition's hosts did, and it makes sense since most if not all the activities we are trying to classify are lateralized.\n\nI recommend you reading these:\nhttps://pressrelease.brainproducts.com/referencing/\nhttps://www.acns.org/UserFiles/file/ACNSStandardizedCriticalCareEEGTerminology_rev2021.pdf"
  },
  "source": "meta"
}