{
  "id": 235367,
  "title": "Why are we only asked to use the green filter to predict when blue filter indicates the nucleus label?",
  "url": "/competitions/hpa-single-cell-image-classification/discussion/235367",
  "author_name": "chongyixiang",
  "post_date": "2021-04-29T07:15:27.171000",
  "votes": 2,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hello. Why are we only asked to use the green filter to predict when blue filter contains information about the nucleus label itself? Wouldn't the blue filter, as well as other filters. should be used to predict the labels? </p>",
  "messages": [
    {
      "id": 1287578,
      "postDate": "2021-04-29T07:38:37.560Z",
      "content": "<p>They are to be used. The green filter contains the protein of interest , and our prediction is the organelle in which the protein is found to be localized. You need to first understand where the protein exactly is , which is only in the green channel , and then also find out what type of organelle it is , where the protein is located, which requires the references of other channels ( yellow can be skipped , it is pretty much the same as red). Most approaches being used involve segmentation of cells , and predicting on individual cells for every image (RGB). </p>",
      "rawMarkdown": "They are to be used. The green filter contains the protein of interest , and our prediction is the organelle in which the protein is found to be localized. You need to first understand where the protein exactly is , which is only in the green channel , and then also find out what type of organelle it is , where the protein is located, which requires the references of other channels ( yellow can be skipped , it is pretty much the same as red). Most approaches being used involve segmentation of cells , and predicting on individual cells for every image (RGB). ",
      "votes": 3,
      "replies": [
        {
          "id": 1288574,
          "postDate": "2021-04-30T06:11:34.017Z",
          "content": "<p>Great explanation.</p>\n<p>One small thing though is that ER (yellow) and MT (red) are similar but not fully the same, and do show different things. When annotating manually, the different aspects of the two are used to better understand what we are seeing.</p>",
          "rawMarkdown": "Great explanation.\n\nOne small thing though is that ER (yellow) and MT (red) are similar but not fully the same, and do show different things. When annotating manually, the different aspects of the two are used to better understand what we are seeing.",
          "votes": 1
        }
      ]
    },
    {
      "id": 1287553,
      "postDate": "2021-04-29T07:15:27.170Z",
      "content": "<p>Hello. Why are we only asked to use the green filter to predict when blue filter contains information about the nucleus label itself? Wouldn't the blue filter, as well as other filters. should be used to predict the labels? </p>",
      "rawMarkdown": "Hello. Why are we only asked to use the green filter to predict when blue filter contains information about the nucleus label itself? Wouldn't the blue filter, as well as other filters. should be used to predict the labels? ",
      "votes": 2
    }
  ],
  "comments": [
    {
      "id": 1287578,
      "author_name": "Satwik",
      "author_url": "",
      "post_date": "2021-04-29T07:38:37.560000",
      "content": "<p>They are to be used. The green filter contains the protein of interest , and our prediction is the organelle in which the protein is found to be localized. You need to first understand where the protein exactly is , which is only in the green channel , and then also find out what type of organelle it is , where the protein is located, which requires the references of other channels ( yellow can be skipped , it is pretty much the same as red). Most approaches being used involve segmentation of cells , and predicting on individual cells for every image (RGB). </p>",
      "votes": 3,
      "replies": [
        {
          "id": 1288574,
          "author_name": "Casper Winsnes",
          "author_url": "",
          "post_date": "2021-04-30T06:11:34.017000",
          "content": "<p>Great explanation.</p>\n<p>One small thing though is that ER (yellow) and MT (red) are similar but not fully the same, and do show different things. When annotating manually, the different aspects of the two are used to better understand what we are seeing.</p>",
          "votes": 1,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1287578": "They are to be used. The green filter contains the protein of interest , and our prediction is the organelle in which the protein is found to be localized. You need to first understand where the protein exactly is , which is only in the green channel , and then also find out what type of organelle it is , where the protein is located, which requires the references of other channels ( yellow can be skipped , it is pretty much the same as red). Most approaches being used involve segmentation of cells , and predicting on individual cells for every image (RGB). ",
    "1287553": "Hello. Why are we only asked to use the green filter to predict when blue filter contains information about the nucleus label itself? Wouldn't the blue filter, as well as other filters. should be used to predict the labels? "
  }
}