{
  "id": 340658,
  "title": "How to identify Spine Cx in stack? ",
  "url": "/competitions/rsna-2022-cervical-spine-fracture-detection/discussion/340658",
  "author_name": "",
  "post_date": "2022-07-30T10:18:48.274095300Z",
  "votes": 4,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Thinking if there is a simple heuristic to match detections in image stack to anatomical perception of what spine it is…</p>\n<ul>\n<li>Naive way, split the stack into 7 equal intervals (maybe add offset to the top and button)</li>\n<li>Informed way, compute statistics based on training detection (but how to deal if multiple spines are fractured or annotation is not compete/perfect?) </li>\n</ul>",
  "messages": [
    {
      "id": "1877069",
      "postDate": "07/30/2022 10:18:48",
      "content": "<p>Thinking if there is a simple heuristic to match detections in image stack to anatomical perception of what spine it is…</p>\n<ul>\n<li>Naive way, split the stack into 7 equal intervals (maybe add offset to the top and button)</li>\n<li>Informed way, compute statistics based on training detection (but how to deal if multiple spines are fractured or annotation is not compete/perfect?) </li>\n</ul>",
      "rawMarkdown": "Thinking if there is a simple heuristic to match detections in image stack to anatomical perception of what spine it is...\n\n- Naive way, split the stack into 7 equal intervals (maybe add offset to the top and button)\n- Informed way, compute statistics based on training detection (but how to deal if multiple spines are fractured or annotation is not compete/perfect?)",
      "votes": null
    },
    {
      "id": "1903248",
      "postDate": "08/17/2022 08:26:03",
      "content": "<p>Spliting the stack in 7 won't work, there is extra region on top and bottom, some of these will varry from subject to subject, and just now checked[i started on the problem only today], number of slices also there is a huge variation, 492 and some cases just 202.. Quite non uniform dataset.</p>",
      "rawMarkdown": "Spliting the stack in 7 won't work, there is extra region on top and bottom, some of these will varry from subject to subject, and just now checked[i started on the problem only today], number of slices also there is a huge variation, 492 and some cases just 202.. Quite non uniform dataset.",
      "votes": null
    },
    {
      "id": "1906513",
      "postDate": "08/20/2022 00:13:30",
      "content": "<p>is there an order slice in the training folder with the number?<br>\nMust process slices in numerical order?</p>",
      "rawMarkdown": "is there an order slice in the training folder with the number?\nMust process slices in numerical order?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1903248,
      "author_name": "krish78",
      "author_url": "",
      "post_date": "08/17/2022 08:26:03",
      "content": "<p>Spliting the stack in 7 won't work, there is extra region on top and bottom, some of these will varry from subject to subject, and just now checked[i started on the problem only today], number of slices also there is a huge variation, 492 and some cases just 202.. Quite non uniform dataset.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1906513,
      "author_name": "rogerautomatizado",
      "author_url": "",
      "post_date": "08/20/2022 00:13:30",
      "content": "<p>is there an order slice in the training folder with the number?<br>\nMust process slices in numerical order?</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1877069": "Thinking if there is a simple heuristic to match detections in image stack to anatomical perception of what spine it is...\n\n- Naive way, split the stack into 7 equal intervals (maybe add offset to the top and button)\n- Informed way, compute statistics based on training detection (but how to deal if multiple spines are fractured or annotation is not compete/perfect?)",
    "1903248": "Spliting the stack in 7 won't work, there is extra region on top and bottom, some of these will varry from subject to subject, and just now checked[i started on the problem only today], number of slices also there is a huge variation, 492 and some cases just 202.. Quite non uniform dataset.",
    "1906513": "is there an order slice in the training folder with the number?\nMust process slices in numerical order?"
  },
  "source": "meta"
}