{
  "id": 68590,
  "title": "yellow channel in the dataset",
  "url": "/competitions/human-protein-atlas-image-classification/discussion/68590",
  "author_name": "",
  "post_date": "2018-10-14T22:21:58.423721700Z",
  "votes": 1,
  "comment_count": 5,
  "views": 0,
  "content": "<p>In the dataset there is red, green, blue and yellow file for every sample.\nWhat do we need the yellow channel for?\nI tried to visualize some of the images <a href=\"https://www.kaggle.com/bachrr/human-protein-atlas-eda\">here</a> but couldn't find how to use the yellow channel!</p>",
  "messages": [
    {
      "id": "403917",
      "postDate": "10/14/2018 22:21:58",
      "content": "<p>In the dataset there is red, green, blue and yellow file for every sample.\nWhat do we need the yellow channel for?\nI tried to visualize some of the images <a href=\"https://www.kaggle.com/bachrr/human-protein-atlas-eda\">here</a> but couldn't find how to use the yellow channel!</p>",
      "rawMarkdown": "In the dataset there is red, green, blue and yellow file for every sample.\nWhat do we need the yellow channel for?\nI tried to visualize some of the images [here][1] but couldn't find how to use the yellow channel!\n\n\n  [1]: https://www.kaggle.com/bachrr/human-protein-atlas-eda",
      "votes": null
    },
    {
      "id": "403990",
      "postDate": "10/15/2018 03:38:45",
      "content": "<p>I created \"RGBA\" and put yellow in the A.</p>",
      "rawMarkdown": "I created \"RGBA\" and put yellow in the A.",
      "votes": null
    },
    {
      "id": "404125",
      "postDate": "10/15/2018 09:20:12",
      "content": "<p>Confocal data acquisition uses monochromes cameras, so they are not supposed to be displayed as RGB images and you can display each channel with the color map you fancy.</p>\n\n<p>Traditionally, nuclei are shown blue because of the dyes used to visualize them (DAPI for example, which fluoresces in the blue region of the light spectrum). For the other channels, I guessed they used (some variant of) Green Fluorescent Protein for the green channel, a microtubule binding protein fused to Red Fluorescent Protein for microtubules, and Yellow Fluorescent Protein for the endoplasmic reticulum.</p>\n\n<p>In the end, you can use the yellow for whatever you think is useful, it is useful for cell segmentation since it stains the whole cell.</p>",
      "rawMarkdown": "Confocal data acquisition uses monochromes cameras, so they are not supposed to be displayed as RGB images and you can display each channel with the color map you fancy.\n\nTraditionally, nuclei are shown blue because of the dyes used to visualize them (DAPI for example, which fluoresces in the blue region of the light spectrum). For the other channels, I guessed they used (some variant of) Green Fluorescent Protein for the green channel, a microtubule binding protein fused to Red Fluorescent Protein for microtubules, and Yellow Fluorescent Protein for the endoplasmic reticulum.\n\nIn the end, you can use the yellow for whatever you think is useful, it is useful for cell segmentation since it stains the whole cell.",
      "votes": null
    },
    {
      "id": "404182",
      "postDate": "10/15/2018 11:59:59",
      "content": "<p>For these images the yellow is staining the endoplasmic reticulum, not the cytosol. While we don't use it very often during manual annotation it can come in handy for distinguishing between certain protein patterns.</p>",
      "rawMarkdown": "For these images the yellow is staining the endoplasmic reticulum, not the cytosol. While we don't use it very often during manual annotation it can come in handy for distinguishing between certain protein patterns.",
      "votes": null
    },
    {
      "id": "404310",
      "postDate": "10/15/2018 14:53:27",
      "content": "<p>Did the same. </p>\n\n<p>A network shouldn't care about what is RGB, only our eyes do. </p>",
      "rawMarkdown": "Did the same. \n\nA network shouldn't care about what is RGB, only our eyes do.",
      "votes": null
    },
    {
      "id": "404389",
      "postDate": "10/15/2018 16:36:44",
      "content": "<p>My mistake, thanks for the reply and I edited the answer.</p>",
      "rawMarkdown": "My mistake, thanks for the reply and I edited the answer.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 403990,
      "author_name": "ldm314",
      "author_url": "",
      "post_date": "10/15/2018 03:38:45",
      "content": "<p>I created \"RGBA\" and put yellow in the A.</p>",
      "votes": null,
      "replies": [
        {
          "id": 404310,
          "author_name": "danmoller",
          "author_url": "",
          "post_date": "10/15/2018 14:53:27",
          "content": "<p>Did the same. </p>\n\n<p>A network shouldn't care about what is RGB, only our eyes do. </p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 404125,
      "author_name": "jschnab",
      "author_url": "",
      "post_date": "10/15/2018 09:20:12",
      "content": "<p>Confocal data acquisition uses monochromes cameras, so they are not supposed to be displayed as RGB images and you can display each channel with the color map you fancy.</p>\n\n<p>Traditionally, nuclei are shown blue because of the dyes used to visualize them (DAPI for example, which fluoresces in the blue region of the light spectrum). For the other channels, I guessed they used (some variant of) Green Fluorescent Protein for the green channel, a microtubule binding protein fused to Red Fluorescent Protein for microtubules, and Yellow Fluorescent Protein for the endoplasmic reticulum.</p>\n\n<p>In the end, you can use the yellow for whatever you think is useful, it is useful for cell segmentation since it stains the whole cell.</p>",
      "votes": null,
      "replies": [
        {
          "id": 404182,
          "author_name": "cwinsnes",
          "author_url": "",
          "post_date": "10/15/2018 11:59:59",
          "content": "<p>For these images the yellow is staining the endoplasmic reticulum, not the cytosol. While we don't use it very often during manual annotation it can come in handy for distinguishing between certain protein patterns.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 404389,
          "author_name": "jschnab",
          "author_url": "",
          "post_date": "10/15/2018 16:36:44",
          "content": "<p>My mistake, thanks for the reply and I edited the answer.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "403917": "In the dataset there is red, green, blue and yellow file for every sample.\nWhat do we need the yellow channel for?\nI tried to visualize some of the images [here][1] but couldn't find how to use the yellow channel!\n\n\n  [1]: https://www.kaggle.com/bachrr/human-protein-atlas-eda",
    "403990": "I created \"RGBA\" and put yellow in the A.",
    "404125": "Confocal data acquisition uses monochromes cameras, so they are not supposed to be displayed as RGB images and you can display each channel with the color map you fancy.\n\nTraditionally, nuclei are shown blue because of the dyes used to visualize them (DAPI for example, which fluoresces in the blue region of the light spectrum). For the other channels, I guessed they used (some variant of) Green Fluorescent Protein for the green channel, a microtubule binding protein fused to Red Fluorescent Protein for microtubules, and Yellow Fluorescent Protein for the endoplasmic reticulum.\n\nIn the end, you can use the yellow for whatever you think is useful, it is useful for cell segmentation since it stains the whole cell.",
    "404182": "For these images the yellow is staining the endoplasmic reticulum, not the cytosol. While we don't use it very often during manual annotation it can come in handy for distinguishing between certain protein patterns.",
    "404310": "Did the same. \n\nA network shouldn't care about what is RGB, only our eyes do.",
    "404389": "My mistake, thanks for the reply and I edited the answer."
  },
  "source": "meta"
}