{
  "id": 155296,
  "title": "what is \"diagnosis=unknown\" in the csv train file?",
  "url": "/competitions/siim-isic-melanoma-classification/discussion/155296",
  "author_name": "",
  "post_date": "2020-06-01T06:19:35.101245200Z",
  "votes": 4,
  "comment_count": 7,
  "views": 0,
  "content": "<p>many of the \"diagnosis attribute\" is labelled as unknown.\ndoes it mean that\n1. it is not diagnosed, hence it can be melanoma, nevus, etc ....\n2. it is diagnosed but is unknown. it is something else but definitely <strong>not</strong> melanoma, nevus, etc ....</p>\n\n<p>there is another \"benign_malignant  attribute\" . I was wondering if an image is diagnosed as \"unknown\", why there is a \"benign_malignant\" label?</p>",
  "messages": [
    {
      "id": "869598",
      "postDate": "06/01/2020 06:19:35",
      "content": "<p>many of the \"diagnosis attribute\" is labelled as unknown.\ndoes it mean that\n1. it is not diagnosed, hence it can be melanoma, nevus, etc ....\n2. it is diagnosed but is unknown. it is something else but definitely <strong>not</strong> melanoma, nevus, etc ....</p>\n\n<p>there is another \"benign_malignant  attribute\" . I was wondering if an image is diagnosed as \"unknown\", why there is a \"benign_malignant\" label?</p>",
      "rawMarkdown": "many of the \"diagnosis attribute\" is labelled as unknown.\ndoes it mean that\n1. it is not diagnosed, hence it can be melanoma, nevus, etc ....\n2. it is diagnosed but is unknown. it is something else but definitely **not** melanoma, nevus, etc ....\n\nthere is another \"benign\\_malignant  attribute\" . I was wondering if an image is diagnosed as \"unknown\", why there is a \"benign\\_malignant\" label?",
      "votes": null
    },
    {
      "id": "870034",
      "postDate": "06/01/2020 12:57:14",
      "content": "<p>I have also had that doubt</p>",
      "rawMarkdown": "I have also had that doubt",
      "votes": null
    },
    {
      "id": "870262",
      "postDate": "06/01/2020 15:34:49",
      "content": "<p>I am assuming the pathologist didn't come to a conclusion if the biopsy was cancerous or not. </p>",
      "rawMarkdown": "I am assuming the pathologist didn't come to a conclusion if the biopsy was cancerous or not.",
      "votes": null
    },
    {
      "id": "870288",
      "postDate": "06/01/2020 15:57:05",
      "content": "<p>I have made the same assumption</p>",
      "rawMarkdown": "I have made the same assumption",
      "votes": null
    },
    {
      "id": "871139",
      "postDate": "06/02/2020 07:00:47",
      "content": "<p>I analyzed this a bit and I got that all the melanoma cells are treated as malignant which is what it should be while all the other diagnosis are treated benign. </p>\n\n<p>So, my analogy on this is, I guess unknown diagnosis is suspicious value, as it may be possible that they did not knew what it was and considered it benign, I think there is a possibility of some of those being melanoma and should have been malignant instead.</p>\n\n<p>And if this is so, I think this is a big problem as the number of unknown diagnosis is quite high (~27100).</p>\n\n<p>I think they only labelled confirmed diagnosis malignant or benign while leaving all not confirmed i.e. unknown diagnosis as benign, I think this unknown diagnosis may contain all the other diagnosis with melanoma too!</p>\n\n<p>Correct me if I am wrong. :)</p>",
      "rawMarkdown": "I analyzed this a bit and I got that all the melanoma cells are treated as malignant which is what it should be while all the other diagnosis are treated benign. \n\nSo, my analogy on this is, I guess unknown diagnosis is suspicious value, as it may be possible that they did not knew what it was and considered it benign, I think there is a possibility of some of those being melanoma and should have been malignant instead.\n\nAnd if this is so, I think this is a big problem as the number of unknown diagnosis is quite high (~27100).\n\nI think they only labelled confirmed diagnosis malignant or benign while leaving all not confirmed i.e. unknown diagnosis as benign, I think this unknown diagnosis may contain all the other diagnosis with melanoma too!\n\nCorrect me if I am wrong. :)",
      "votes": null
    },
    {
      "id": "871150",
      "postDate": "06/02/2020 07:07:41",
      "content": "<p>Or it may be possible that your second point is correct, because if we see, the images are taken over the years, so it can be possible that \"unknown\" diagnosis means something else but not any of the other category like melanoma, nv, etc.</p>",
      "rawMarkdown": "Or it may be possible that your second point is correct, because if we see, the images are taken over the years, so it can be possible that \"unknown\" diagnosis means something else but not any of the other category like melanoma, nv, etc.",
      "votes": null
    },
    {
      "id": "875346",
      "postDate": "06/05/2020 17:22:39",
      "content": "<p>Great question - sorry this is unclear in our data description. </p>\n\n<p>We expect all malignant lesions are correctly labeled as malignant. </p>\n\n<p>Unknown lesions are: lesions that have not been biopsied and therefore the diagnosis is not \"known\" but they are determined by the experts to be \"benign.\" All the unknown lesions have been monitored clinically and photographs have been reviewed to confirm a benign label. </p>",
      "rawMarkdown": "Great question - sorry this is unclear in our data description. \n\nWe expect all malignant lesions are correctly labeled as malignant. \n\nUnknown lesions are: lesions that have not been biopsied and therefore the diagnosis is not \"known\" but they are determined by the experts to be \"benign.\" All the unknown lesions have been monitored clinically and photographs have been reviewed to confirm a benign label.",
      "votes": null
    },
    {
      "id": "875356",
      "postDate": "06/05/2020 17:35:48",
      "content": "<p><a href=\"/veronicarotemberg\">@veronicarotemberg</a> </p>\n\n<p>thanks for the answer. i got it</p>",
      "rawMarkdown": "veronicarotemberg \n\nthanks for the answer. i got it",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 870034,
      "author_name": "maximofn",
      "author_url": "",
      "post_date": "06/01/2020 12:57:14",
      "content": "<p>I have also had that doubt</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 870262,
      "author_name": "vikasvpatil",
      "author_url": "",
      "post_date": "06/01/2020 15:34:49",
      "content": "<p>I am assuming the pathologist didn't come to a conclusion if the biopsy was cancerous or not. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 870288,
      "author_name": "maximofn",
      "author_url": "",
      "post_date": "06/01/2020 15:57:05",
      "content": "<p>I have made the same assumption</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 871139,
      "author_name": "sarques",
      "author_url": "",
      "post_date": "06/02/2020 07:00:47",
      "content": "<p>I analyzed this a bit and I got that all the melanoma cells are treated as malignant which is what it should be while all the other diagnosis are treated benign. </p>\n\n<p>So, my analogy on this is, I guess unknown diagnosis is suspicious value, as it may be possible that they did not knew what it was and considered it benign, I think there is a possibility of some of those being melanoma and should have been malignant instead.</p>\n\n<p>And if this is so, I think this is a big problem as the number of unknown diagnosis is quite high (~27100).</p>\n\n<p>I think they only labelled confirmed diagnosis malignant or benign while leaving all not confirmed i.e. unknown diagnosis as benign, I think this unknown diagnosis may contain all the other diagnosis with melanoma too!</p>\n\n<p>Correct me if I am wrong. :)</p>",
      "votes": null,
      "replies": [
        {
          "id": 871150,
          "author_name": "sarques",
          "author_url": "",
          "post_date": "06/02/2020 07:07:41",
          "content": "<p>Or it may be possible that your second point is correct, because if we see, the images are taken over the years, so it can be possible that \"unknown\" diagnosis means something else but not any of the other category like melanoma, nv, etc.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 875346,
      "author_name": "veronicarotemberg",
      "author_url": "",
      "post_date": "06/05/2020 17:22:39",
      "content": "<p>Great question - sorry this is unclear in our data description. </p>\n\n<p>We expect all malignant lesions are correctly labeled as malignant. </p>\n\n<p>Unknown lesions are: lesions that have not been biopsied and therefore the diagnosis is not \"known\" but they are determined by the experts to be \"benign.\" All the unknown lesions have been monitored clinically and photographs have been reviewed to confirm a benign label. </p>",
      "votes": null,
      "replies": [
        {
          "id": 875356,
          "author_name": "hengck23",
          "author_url": "",
          "post_date": "06/05/2020 17:35:48",
          "content": "<p><a href=\"/veronicarotemberg\">@veronicarotemberg</a> </p>\n\n<p>thanks for the answer. i got it</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "869598": "many of the \"diagnosis attribute\" is labelled as unknown.\ndoes it mean that\n1. it is not diagnosed, hence it can be melanoma, nevus, etc ....\n2. it is diagnosed but is unknown. it is something else but definitely **not** melanoma, nevus, etc ....\n\nthere is another \"benign\\_malignant  attribute\" . I was wondering if an image is diagnosed as \"unknown\", why there is a \"benign\\_malignant\" label?",
    "870034": "I have also had that doubt",
    "870262": "I am assuming the pathologist didn't come to a conclusion if the biopsy was cancerous or not.",
    "870288": "I have made the same assumption",
    "871139": "I analyzed this a bit and I got that all the melanoma cells are treated as malignant which is what it should be while all the other diagnosis are treated benign. \n\nSo, my analogy on this is, I guess unknown diagnosis is suspicious value, as it may be possible that they did not knew what it was and considered it benign, I think there is a possibility of some of those being melanoma and should have been malignant instead.\n\nAnd if this is so, I think this is a big problem as the number of unknown diagnosis is quite high (~27100).\n\nI think they only labelled confirmed diagnosis malignant or benign while leaving all not confirmed i.e. unknown diagnosis as benign, I think this unknown diagnosis may contain all the other diagnosis with melanoma too!\n\nCorrect me if I am wrong. :)",
    "871150": "Or it may be possible that your second point is correct, because if we see, the images are taken over the years, so it can be possible that \"unknown\" diagnosis means something else but not any of the other category like melanoma, nv, etc.",
    "875346": "Great question - sorry this is unclear in our data description. \n\nWe expect all malignant lesions are correctly labeled as malignant. \n\nUnknown lesions are: lesions that have not been biopsied and therefore the diagnosis is not \"known\" but they are determined by the experts to be \"benign.\" All the unknown lesions have been monitored clinically and photographs have been reviewed to confirm a benign label.",
    "875356": "veronicarotemberg \n\nthanks for the answer. i got it"
  },
  "source": "meta"
}