{
  "id": 481717,
  "title": "Is all the images cancerous ?",
  "url": "/competitions/intel-mobileodt-cervical-cancer-screening/discussion/481717",
  "author_name": "",
  "post_date": "2024-03-04T18:56:21.066042600Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I have been reading in some research papers that they first try to classify first with normal vs abnormal images and then to and differentiate with Type it belongs to. Do we have any such information that any of the images are normal cervix images ?</p>",
  "messages": [
    {
      "id": "2681651",
      "postDate": "03/04/2024 18:56:21",
      "content": "<p>I have been reading in some research papers that they first try to classify first with normal vs abnormal images and then to and differentiate with Type it belongs to. Do we have any such information that any of the images are normal cervix images ?</p>",
      "rawMarkdown": "I have been reading in some research papers that they first try to classify first with normal vs abnormal images and then to and differentiate with Type it belongs to. Do we have any such information that any of the images are normal cervix images ?",
      "votes": null
    },
    {
      "id": "2682820",
      "postDate": "03/05/2024 16:24:52",
      "content": "<p>As far as I remember, the type of the cervix is orthogonal to healthy/suspect/full blown cancer. The issue is that for type III (?) (the one for which the crossing between pluri and unicellular is inside the cervix) the physician must perform additional checks.</p>\n<p>Coming back, I think I saw some non-healthy cervices in the dataset. I manually labeled quite a lot, back in the day :(</p>\n<p>Pro-er tip: SAM and other transformer architectures might \"smoke\" this dataset.</p>\n<p>Hope it helps!</p>",
      "rawMarkdown": "As far as I remember, the type of the cervix is orthogonal to healthy/suspect/full blown cancer. The issue is that for type III (?) (the one for which the crossing between pluri and unicellular is inside the cervix) the physician must perform additional checks.\n\nComing back, I think I saw some non-healthy cervices in the dataset. I manually labeled quite a lot, back in the day :(\n\nPro-er tip: SAM and other transformer architectures might \"smoke\" this dataset.\n\nHope it helps!",
      "votes": null
    },
    {
      "id": "2688861",
      "postDate": "03/09/2024 14:17:53",
      "content": "<p>Oh okie, thanks. Its that in some papers I saw researchers buildings models for normal image vs cancerous image for cervical cancer and source of data is shown to be this dataset itself. So I was confused.</p>",
      "rawMarkdown": "Oh okie, thanks. Its that in some papers I saw researchers buildings models for normal image vs cancerous image for cervical cancer and source of data is shown to be this dataset itself. So I was confused.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2682820,
      "author_name": "visoft",
      "author_url": "",
      "post_date": "03/05/2024 16:24:52",
      "content": "<p>As far as I remember, the type of the cervix is orthogonal to healthy/suspect/full blown cancer. The issue is that for type III (?) (the one for which the crossing between pluri and unicellular is inside the cervix) the physician must perform additional checks.</p>\n<p>Coming back, I think I saw some non-healthy cervices in the dataset. I manually labeled quite a lot, back in the day :(</p>\n<p>Pro-er tip: SAM and other transformer architectures might \"smoke\" this dataset.</p>\n<p>Hope it helps!</p>",
      "votes": null,
      "replies": [
        {
          "id": 2688861,
          "author_name": "thomasjv",
          "author_url": "",
          "post_date": "03/09/2024 14:17:53",
          "content": "<p>Oh okie, thanks. Its that in some papers I saw researchers buildings models for normal image vs cancerous image for cervical cancer and source of data is shown to be this dataset itself. So I was confused.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2681651": "I have been reading in some research papers that they first try to classify first with normal vs abnormal images and then to and differentiate with Type it belongs to. Do we have any such information that any of the images are normal cervix images ?",
    "2682820": "As far as I remember, the type of the cervix is orthogonal to healthy/suspect/full blown cancer. The issue is that for type III (?) (the one for which the crossing between pluri and unicellular is inside the cervix) the physician must perform additional checks.\n\nComing back, I think I saw some non-healthy cervices in the dataset. I manually labeled quite a lot, back in the day :(\n\nPro-er tip: SAM and other transformer architectures might \"smoke\" this dataset.\n\nHope it helps!",
    "2688861": "Oh okie, thanks. Its that in some papers I saw researchers buildings models for normal image vs cancerous image for cervical cancer and source of data is shown to be this dataset itself. So I was confused."
  },
  "source": "meta"
}