{
  "id": 362671,
  "title": "Someone has successfully used the bounding box?",
  "url": "/competitions/rsna-2022-cervical-spine-fracture-detection/discussion/362671",
  "author_name": "Pierre Tisseur",
  "post_date": "2022-10-28T13:30:32.196000",
  "votes": 2,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I started late trying to train bounding boxes with raw images. I wonder if anyone has tried to train bounding boxes with segmented images and if this could help the classifier to focus his attention on the fracture?</p>",
  "messages": [
    {
      "id": 2012462,
      "postDate": "2022-11-01T08:15:31.297Z",
      "content": "<p>I tried, but it did not help us.</p>",
      "rawMarkdown": "I tried, but it did not help us.",
      "votes": 1
    },
    {
      "id": 2007770,
      "postDate": "2022-10-28T13:30:32.197Z",
      "content": "<p>I started late trying to train bounding boxes with raw images. I wonder if anyone has tried to train bounding boxes with segmented images and if this could help the classifier to focus his attention on the fracture?</p>",
      "rawMarkdown": "I started late trying to train bounding boxes with raw images. I wonder if anyone has tried to train bounding boxes with segmented images and if this could help the classifier to focus his attention on the fracture?",
      "votes": 2
    },
    {
      "id": 2009103,
      "postDate": "2022-10-29T16:32:54.260Z",
      "content": "<p>We used it with some success in combination with the segmentations (taking the intersection of the two only when the level is positive, can check out our solution for more info). </p>\n<p>It may have helped more if slice specific fx information was needed for scoring. But ultimately for this task, I don't think it was necessary. Most top solutions didn't seem to use it.</p>",
      "rawMarkdown": "We used it with some success in combination with the segmentations (taking the intersection of the two only when the level is positive, can check out our solution for more info). \n\nIt may have helped more if slice specific fx information was needed for scoring. But ultimately for this task, I don't think it was necessary. Most top solutions didn't seem to use it."
    }
  ],
  "comments": [
    {
      "id": 2012462,
      "author_name": "Dieter",
      "author_url": "",
      "post_date": "2022-11-01T08:15:31.297000",
      "content": "<p>I tried, but it did not help us.</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 2009103,
      "author_name": "Jesse",
      "author_url": "",
      "post_date": "2022-10-29T16:32:54.260000",
      "content": "<p>We used it with some success in combination with the segmentations (taking the intersection of the two only when the level is positive, can check out our solution for more info). </p>\n<p>It may have helped more if slice specific fx information was needed for scoring. But ultimately for this task, I don't think it was necessary. Most top solutions didn't seem to use it.</p>",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2012462": "I tried, but it did not help us.",
    "2007770": "I started late trying to train bounding boxes with raw images. I wonder if anyone has tried to train bounding boxes with segmented images and if this could help the classifier to focus his attention on the fracture?",
    "2009103": "We used it with some success in combination with the segmentations (taking the intersection of the two only when the level is positive, can check out our solution for more info). \n\nIt may have helped more if slice specific fx information was needed for scoring. But ultimately for this task, I don't think it was necessary. Most top solutions didn't seem to use it."
  }
}