{
  "id": 240046,
  "title": "Question about none, negative, no finding",
  "url": "/competitions/siim-covid19-detection/discussion/240046",
  "author_name": "",
  "post_date": "2021-05-18T11:47:20.028073200Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>In study level csv file each study has exactly 1 label either  <code>negative|typical|indeterminate|atypical</code>. Also, we can see that studies with <strong>Negative for Pneumonia == 1</strong> have images which are all labeled as <code>none 1 0 0 1 1</code>.</p>\n<p>In this case do we consider all <strong>Negative for Pneumonia == 1</strong> cases as <code>none</code> - <code>no finding</code>? Is there a difference between no finding and negative cases? For example, an opacity area which can be labeled as <code>negative</code>?</p>\n<p>In this case are we going to submit a study with <code>Negative for Pneumonia</code> as <code>none 1 0 0 1 1</code> or <code>negative 1 0 0 1 1</code>?</p>\n<p>However, in <a href=\"https://www.kaggle.com/c/siim-covid19-detection/overview/evaluation\" target=\"_blank\">evaluation page</a> the explanation about submission feels like <code>none</code> and <code>negative</code> are separate things. Can you give a more detailed explanation about them, perhaps an explanation on how the annotations took place?</p>",
  "messages": [
    {
      "id": "1313065",
      "postDate": "05/18/2021 11:47:20",
      "content": "<p>In study level csv file each study has exactly 1 label either  <code>negative|typical|indeterminate|atypical</code>. Also, we can see that studies with <strong>Negative for Pneumonia == 1</strong> have images which are all labeled as <code>none 1 0 0 1 1</code>.</p>\n<p>In this case do we consider all <strong>Negative for Pneumonia == 1</strong> cases as <code>none</code> - <code>no finding</code>? Is there a difference between no finding and negative cases? For example, an opacity area which can be labeled as <code>negative</code>?</p>\n<p>In this case are we going to submit a study with <code>Negative for Pneumonia</code> as <code>none 1 0 0 1 1</code> or <code>negative 1 0 0 1 1</code>?</p>\n<p>However, in <a href=\"https://www.kaggle.com/c/siim-covid19-detection/overview/evaluation\" target=\"_blank\">evaluation page</a> the explanation about submission feels like <code>none</code> and <code>negative</code> are separate things. Can you give a more detailed explanation about them, perhaps an explanation on how the annotations took place?</p>",
      "rawMarkdown": "In study level csv file each study has exactly 1 label either  `negative|typical|indeterminate|atypical`. Also, we can see that studies with **Negative for Pneumonia == 1** have images which are all labeled as `none 1 0 0 1 1`.\n\nIn this case do we consider all **Negative for Pneumonia == 1** cases as `none` - `no finding`? Is there a difference between no finding and negative cases? For example, an opacity area which can be labeled as `negative`?\n\nIn this case are we going to submit a study with `Negative for Pneumonia` as `none 1 0 0 1 1` or `negative 1 0 0 1 1`?\n\nHowever, in [evaluation page](https://www.kaggle.com/c/siim-covid19-detection/overview/evaluation) the explanation about submission feels like `none` and `negative` are separate things. Can you give a more detailed explanation about them, perhaps an explanation on how the annotations took place?",
      "votes": null
    },
    {
      "id": "1315383",
      "postDate": "05/19/2021 18:45:09",
      "content": "<p><a href=\"https://www.kaggle.com/keremt\" target=\"_blank\">@keremt</a>, <code>negative</code> is a label for study-level prediction and <code>none</code> is a label for image-level prediction. I tried to explain this in more detail <a href=\"https://www.kaggle.com/c/siim-covid19-detection/discussion/240329\" target=\"_blank\">here</a>. Let me know if that makes sense.</p>",
      "rawMarkdown": "keremt, `negative` is a label for study-level prediction and `none` is a label for image-level prediction. I tried to explain this in more detail [here](https://www.kaggle.com/c/siim-covid19-detection/discussion/240329). Let me know if that makes sense.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1315383,
      "author_name": "hassiahk",
      "author_url": "",
      "post_date": "05/19/2021 18:45:09",
      "content": "<p><a href=\"https://www.kaggle.com/keremt\" target=\"_blank\">@keremt</a>, <code>negative</code> is a label for study-level prediction and <code>none</code> is a label for image-level prediction. I tried to explain this in more detail <a href=\"https://www.kaggle.com/c/siim-covid19-detection/discussion/240329\" target=\"_blank\">here</a>. Let me know if that makes sense.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1313065": "In study level csv file each study has exactly 1 label either  `negative|typical|indeterminate|atypical`. Also, we can see that studies with **Negative for Pneumonia == 1** have images which are all labeled as `none 1 0 0 1 1`.\n\nIn this case do we consider all **Negative for Pneumonia == 1** cases as `none` - `no finding`? Is there a difference between no finding and negative cases? For example, an opacity area which can be labeled as `negative`?\n\nIn this case are we going to submit a study with `Negative for Pneumonia` as `none 1 0 0 1 1` or `negative 1 0 0 1 1`?\n\nHowever, in [evaluation page](https://www.kaggle.com/c/siim-covid19-detection/overview/evaluation) the explanation about submission feels like `none` and `negative` are separate things. Can you give a more detailed explanation about them, perhaps an explanation on how the annotations took place?",
    "1315383": "keremt, `negative` is a label for study-level prediction and `none` is a label for image-level prediction. I tried to explain this in more detail [here](https://www.kaggle.com/c/siim-covid19-detection/discussion/240329). Let me know if that makes sense."
  },
  "source": "meta"
}