{
  "id": 109479,
  "title": "@philculliton  - Rule clarification needed - \"pixel data only\"",
  "url": "/competitions/rsna-intracranial-hemorrhage-detection/discussion/109479",
  "author_name": "steelrose",
  "post_date": "2019-09-19T13:39:33.419000",
  "votes": 16,
  "comment_count": 11,
  "views": 0,
  "content": "<p>@philculliton \nHi Phil,\ncan you please clarify the rule: \"Submission predictions must be based <strong>entirely on the pixel data</strong> in the provided datasets.\"</p>\n\n<p>It's a bit \"confusing\" - i.e. does it mean we <strong>cannot read</strong> the DICOM metadata - especially <strong>\"Window Center\", \"Window Width\", \"Rescale Intercept\" and \"Rescale Slope\"</strong> - for data pre-processing?</p>\n\n<p>Or should we assume that all images have the same window information and hardcode it?</p>",
  "messages": [
    {
      "id": 629930,
      "postDate": "2019-09-19T13:39:33.420Z",
      "content": "<p>@philculliton \nHi Phil,\ncan you please clarify the rule: \"Submission predictions must be based <strong>entirely on the pixel data</strong> in the provided datasets.\"</p>\n\n<p>It's a bit \"confusing\" - i.e. does it mean we <strong>cannot read</strong> the DICOM metadata - especially <strong>\"Window Center\", \"Window Width\", \"Rescale Intercept\" and \"Rescale Slope\"</strong> - for data pre-processing?</p>\n\n<p>Or should we assume that all images have the same window information and hardcode it?</p>",
      "rawMarkdown": "@philculliton \nHi Phil,\ncan you please clarify the rule: \"Submission predictions must be based **entirely on the pixel data** in the provided datasets.\"\n\nIt's a bit \"confusing\" - i.e. does it mean we **cannot read** the DICOM metadata - especially **\"Window Center\", \"Window Width\", \"Rescale Intercept\" and \"Rescale Slope\"** - for data pre-processing?\n\nOr should we assume that all images have the same window information and hardcode it?",
      "votes": 17
    },
    {
      "id": 630067,
      "postDate": "2019-09-19T17:16:03.583Z",
      "content": "<p>According to the host: using the metadata for preprocessing, the way you're describing it, is fine.</p>",
      "rawMarkdown": "According to the host: using the metadata for preprocessing, the way you're describing it, is fine.",
      "votes": 8,
      "replies": [
        {
          "id": 630068,
          "postDate": "2019-09-19T17:17:23.820Z",
          "content": "<p>thanks</p>",
          "rawMarkdown": "thanks"
        },
        {
          "id": 630263,
          "postDate": "2019-09-20T02:14:03.797Z",
          "content": "<p>Thank you, this will be necessary.</p>",
          "rawMarkdown": "Thank you, this will be necessary."
        },
        {
          "id": 631815,
          "postDate": "2019-09-22T18:21:35.770Z",
          "content": "<p><a href=\"/philculliton\">@philculliton</a> could you please clarify what is not allowed, or explain in some other way why was this remark added and what it means? E.g. which fields are allowed and which are not allowed?</p>",
          "rawMarkdown": "@philculliton could you please clarify what is not allowed, or explain in some other way why was this remark added and what it means? E.g. which fields are allowed and which are not allowed?",
          "votes": 4
        },
        {
          "id": 631888,
          "postDate": "2019-09-22T21:55:13.673Z",
          "content": "<p>Is it against the rules to create a 3D construction or another section using series id etc.?</p>",
          "rawMarkdown": "Is it against the rules to create a 3D construction or another section using series id etc.?",
          "votes": 6
        },
        {
          "id": 632470,
          "postDate": "2019-09-23T16:21:24.100Z",
          "content": "<p>Good questions!</p>\n\n<p><a href=\"/kambarakun\">@kambarakun</a> - it is not against the rules to create a 3D construction using the series ID.</p>\n\n<p><a href=\"/lopuhin\">@lopuhin</a> - sorry for the ambiguity.  We're approaching this on a case-by-case basis. If you have a use for the metadata that falls outside of what has already been allowed, please ask, and the host will clarify. All of the provided fields were included so they could be used, but the exact use should be cleared to make sure you're falling within the guidelines of the rules.</p>",
          "rawMarkdown": "Good questions!\n\n@kambarakun - it is not against the rules to create a 3D construction using the series ID.\n\n@lopuhin - sorry for the ambiguity.  We're approaching this on a case-by-case basis. If you have a use for the metadata that falls outside of what has already been allowed, please ask, and the host will clarify. All of the provided fields were included so they could be used, but the exact use should be cleared to make sure you're falling within the guidelines of the rules.",
          "votes": 3
        },
        {
          "id": 632490,
          "postDate": "2019-09-23T17:00:22.747Z",
          "content": "<p>Ok that's fair, thanks for clarification!</p>",
          "rawMarkdown": "Ok that's fair, thanks for clarification!"
        },
        {
          "id": 632503,
          "postDate": "2019-09-23T17:20:25.193Z",
          "content": "<p>You're welcome - thanks for the question!</p>",
          "rawMarkdown": "You're welcome - thanks for the question!"
        },
        {
          "id": 632585,
          "postDate": "2019-09-23T19:05:25.040Z",
          "content": "<p><a href=\"/philculliton\">@philculliton</a> Can we have a dedicated topic/thread for metadata usage Q&amp;A, just like external data sharing? These are the information every participants want to know. We can also avoid duplicated questions.</p>",
          "rawMarkdown": "@philculliton Can we have a dedicated topic/thread for metadata usage Q&amp;A, just like external data sharing? These are the information every participants want to know. We can also avoid duplicated questions.",
          "votes": 1
        },
        {
          "id": 632646,
          "postDate": "2019-09-23T20:51:42.200Z",
          "content": "<p><a href=\"/wowfattie\">@wowfattie</a> - thanks!  Great suggestion.  Done!</p>",
          "rawMarkdown": "@wowfattie - thanks!  Great suggestion.  Done!",
          "votes": 2
        }
      ]
    },
    {
      "id": 630050,
      "postDate": "2019-09-19T16:34:02.220Z",
      "content": "<p>yeah it's nearly impossible to do any thing useful with CT image without window and normalization info :)</p>",
      "rawMarkdown": "yeah it's nearly impossible to do any thing useful with CT image without window and normalization info :)",
      "votes": 1
    }
  ],
  "comments": [
    {
      "id": 630067,
      "author_name": "Phil Culliton",
      "author_url": "",
      "post_date": "2019-09-19T17:16:03.583000",
      "content": "<p>According to the host: using the metadata for preprocessing, the way you're describing it, is fine.</p>",
      "votes": 8,
      "replies": [
        {
          "id": 630068,
          "author_name": "steelrose",
          "author_url": "",
          "post_date": "2019-09-19T17:17:23.820000",
          "content": "<p>thanks</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 630263,
          "author_name": "ldawes",
          "author_url": "",
          "post_date": "2019-09-20T02:14:03.797000",
          "content": "<p>Thank you, this will be necessary.</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 631815,
          "author_name": "Konstantin Lopukhin",
          "author_url": "",
          "post_date": "2019-09-22T18:21:35.770000",
          "content": "<p><a href=\"/philculliton\">@philculliton</a> could you please clarify what is not allowed, or explain in some other way why was this remark added and what it means? E.g. which fields are allowed and which are not allowed?</p>",
          "votes": 4,
          "replies": []
        },
        {
          "id": 631888,
          "author_name": "kambarakun",
          "author_url": "",
          "post_date": "2019-09-22T21:55:13.673000",
          "content": "<p>Is it against the rules to create a 3D construction or another section using series id etc.?</p>",
          "votes": 6,
          "replies": []
        },
        {
          "id": 632470,
          "author_name": "Phil Culliton",
          "author_url": "",
          "post_date": "2019-09-23T16:21:24.100000",
          "content": "<p>Good questions!</p>\n\n<p><a href=\"/kambarakun\">@kambarakun</a> - it is not against the rules to create a 3D construction using the series ID.</p>\n\n<p><a href=\"/lopuhin\">@lopuhin</a> - sorry for the ambiguity.  We're approaching this on a case-by-case basis. If you have a use for the metadata that falls outside of what has already been allowed, please ask, and the host will clarify. All of the provided fields were included so they could be used, but the exact use should be cleared to make sure you're falling within the guidelines of the rules.</p>",
          "votes": 3,
          "replies": []
        },
        {
          "id": 632490,
          "author_name": "Konstantin Lopukhin",
          "author_url": "",
          "post_date": "2019-09-23T17:00:22.747000",
          "content": "<p>Ok that's fair, thanks for clarification!</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 632503,
          "author_name": "Phil Culliton",
          "author_url": "",
          "post_date": "2019-09-23T17:20:25.193000",
          "content": "<p>You're welcome - thanks for the question!</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 632585,
          "author_name": "Guanshuo Xu",
          "author_url": "",
          "post_date": "2019-09-23T19:05:25.040000",
          "content": "<p><a href=\"/philculliton\">@philculliton</a> Can we have a dedicated topic/thread for metadata usage Q&amp;A, just like external data sharing? These are the information every participants want to know. We can also avoid duplicated questions.</p>",
          "votes": 1,
          "replies": []
        },
        {
          "id": 632646,
          "author_name": "Phil Culliton",
          "author_url": "",
          "post_date": "2019-09-23T20:51:42.200000",
          "content": "<p><a href=\"/wowfattie\">@wowfattie</a> - thanks!  Great suggestion.  Done!</p>",
          "votes": 2,
          "replies": []
        }
      ]
    },
    {
      "id": 630050,
      "author_name": "DatNT",
      "author_url": "",
      "post_date": "2019-09-19T16:34:02.220000",
      "content": "<p>yeah it's nearly impossible to do any thing useful with CT image without window and normalization info :)</p>",
      "votes": 1,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "629930": "@philculliton \nHi Phil,\ncan you please clarify the rule: \"Submission predictions must be based **entirely on the pixel data** in the provided datasets.\"\n\nIt's a bit \"confusing\" - i.e. does it mean we **cannot read** the DICOM metadata - especially **\"Window Center\", \"Window Width\", \"Rescale Intercept\" and \"Rescale Slope\"** - for data pre-processing?\n\nOr should we assume that all images have the same window information and hardcode it?",
    "630067": "According to the host: using the metadata for preprocessing, the way you're describing it, is fine.",
    "630050": "yeah it's nearly impossible to do any thing useful with CT image without window and normalization info :)"
  }
}