{
  "id": 514897,
  "title": "Questions Regarding Study and Image Series Identification in Sample Submissions",
  "url": "/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/514897",
  "author_name": "",
  "post_date": "2024-06-26T04:29:05.805003400Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>I have a few questions.</p>\n<ul>\n<li><p>Why does the sample submission row_id correspond only to the study_id, rather than {study_id}{series_id}, or even {study_id}{series_id}_{image number}?</p></li>\n<li><p>Does that imply that a single study is made up of multiple image series captured from different angles of a single person? Are we diagnosing this individual to check if any of their five bones (L1L2, L2L3, L3L4, L4L5, L5S1) have any of the five diseases (2 foraminal, 2 subarticular, 1 canal stenosis)?</p></li>\n<li><p>Regarding the train_label_coordinates.csv file, the instance_number is just a subset of the image series_id. Does this mean an expert can just take that instance_number.dcm and diagnose that the patient has some kind of disease at x,y? <br>\nBut for other images in that series, which do not appear in the train_label_coordinates.csv instance_number, are they useless, or less useful for diagnosing compared to the ones whose instance numbers are collected in train_label_coordinates.csv?</p></li>\n</ul>\n<hr>\n<p>This is really a complex dataset structure🥹</p>",
  "messages": [
    {
      "id": "2890324",
      "postDate": "06/26/2024 04:29:05",
      "content": "<p>I have a few questions.</p>\n<ul>\n<li><p>Why does the sample submission row_id correspond only to the study_id, rather than {study_id}{series_id}, or even {study_id}{series_id}_{image number}?</p></li>\n<li><p>Does that imply that a single study is made up of multiple image series captured from different angles of a single person? Are we diagnosing this individual to check if any of their five bones (L1L2, L2L3, L3L4, L4L5, L5S1) have any of the five diseases (2 foraminal, 2 subarticular, 1 canal stenosis)?</p></li>\n<li><p>Regarding the train_label_coordinates.csv file, the instance_number is just a subset of the image series_id. Does this mean an expert can just take that instance_number.dcm and diagnose that the patient has some kind of disease at x,y? <br>\nBut for other images in that series, which do not appear in the train_label_coordinates.csv instance_number, are they useless, or less useful for diagnosing compared to the ones whose instance numbers are collected in train_label_coordinates.csv?</p></li>\n</ul>\n<hr>\n<p>This is really a complex dataset structure🥹</p>",
      "rawMarkdown": "I have a few questions.\n\n- Why does the sample submission row_id correspond only to the study_id, rather than {study_id}{series_id}, or even {study_id}{series_id}_{image number}?\n\n- Does that imply that a single study is made up of multiple image series captured from different angles of a single person? Are we diagnosing this individual to check if any of their five bones (L1L2, L2L3, L3L4, L4L5, L5S1) have any of the five diseases (2 foraminal, 2 subarticular, 1 canal stenosis)?\n\n- Regarding the train_label_coordinates.csv file, the instance_number is just a subset of the image series_id. Does this mean an expert can just take that instance_number.dcm and diagnose that the patient has some kind of disease at x,y? \nBut for other images in that series, which do not appear in the train_label_coordinates.csv instance_number, are they useless, or less useful for diagnosing compared to the ones whose instance numbers are collected in train_label_coordinates.csv?\n\n\n---\n\nThis is really a complex dataset structure🥹",
      "votes": null
    },
    {
      "id": "2890389",
      "postDate": "06/26/2024 05:23:36",
      "content": "<blockquote>\n  <p>Why does the sample submission row_id correspond only to the study_id, rather than {study_id}{series_id}, or even {study_id}{series_id}_{image number}?</p>\n</blockquote>\n<p>Because the goal of this competition is to predict from Axial T2, Sagittal T1 and Sagittal T2/STIR series the severity (normal/mild to moderate to severe) of the conditions at each level. See the following observation:</p>\n<table>\n<thead>\n<tr>\n<th>condition</th>\n<th>Axial T2</th>\n<th>Sagittal T1</th>\n<th>Sagittal T2/STIR</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>Left Subarticular Stenosis</td>\n<td>9608</td>\n<td>0</td>\n<td>0</td>\n</tr>\n<tr>\n<td>Right Subarticular Stenosis</td>\n<td>9612</td>\n<td>0</td>\n<td>0</td>\n</tr>\n<tr>\n<td>Left Neural Foraminal Narrowing</td>\n<td>0</td>\n<td>9860</td>\n<td>0</td>\n</tr>\n<tr>\n<td>Right Neural Foraminal Narrowing</td>\n<td>0</td>\n<td>9859</td>\n<td>0</td>\n</tr>\n<tr>\n<td>Spinal Canal Stenosis</td>\n<td>0</td>\n<td>5</td>\n<td>9748</td>\n</tr>\n</tbody>\n</table>\n<p>Each condition has a particular series to refer to. As for why there is a deviation with 5 Spinal Canal Stenosis with Sagittal T1 see my comment <a href=\"https://www.kaggle.com/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/514891#2890326\" target=\"_blank\">here</a>.</p>\n<blockquote>\n  <p>Does that imply that a single study is made up of multiple image series captured from different angles of a single person?</p>\n</blockquote>\n<p>Yes, and I suggest you to check out the post: <a href=\"https://www.kaggle.com/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/507101\" target=\"_blank\">Radiologist's Insights I</a></p>\n<blockquote>\n  <p>Does this mean an expert can just take that instance_number.dcm and diagnose that the patient has some kind of disease at x,y?<br>\n  But for other images in that series, which do not appear in the train_label_coordinates.csv instance_number, are they useless, or less useful for diagnosing compared to the ones whose instance numbers are collected in train_label_coordinates.csv?</p>\n</blockquote>\n<p>I can't speak on an expert's behalf, but I think that point is where the epicenter of the condition is located. It may need the adjacent slices or comparison with a slice in the orthogonal plane [that is if a point is on the axial slice we may need the sagittal slice crossing the point to get the full picture].</p>\n<p>It is not that the other slices are not relevant but that we need to locate this region ourselves during prediction as there is no <code>test_label_coordinates.csv</code>.</p>",
      "rawMarkdown": "> Why does the sample submission row_id correspond only to the study_id, rather than {study_id}{series_id}, or even {study_id}{series_id}_{image number}?\n\nBecause the goal of this competition is to predict from Axial T2, Sagittal T1 and Sagittal T2/STIR series the severity (normal/mild to moderate to severe) of the conditions at each level. See the following observation:\n| condition                        |   Axial T2 |   Sagittal T1 |   Sagittal T2/STIR |\n|:---------------------------------|-----------:|--------------:|-------------------:|\n| Left Subarticular Stenosis       |       9608 |             0 |                  0 |\n| Right Subarticular Stenosis      |       9612 |             0 |                  0 |\n| Left Neural Foraminal Narrowing  |          0 |          9860 |                  0 |\n| Right Neural Foraminal Narrowing |          0 |          9859 |                  0 |\n| Spinal Canal Stenosis            |          0 |             5 |               9748 |\n\nEach condition has a particular series to refer to. As for why there is a deviation with 5 Spinal Canal Stenosis with Sagittal T1 see my comment [here](https://www.kaggle.com/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/514891#2890326).\n\n> Does that imply that a single study is made up of multiple image series captured from different angles of a single person?\n\nYes, and I suggest you to check out the post: [Radiologist's Insights I](https://www.kaggle.com/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/507101)\n\n> Does this mean an expert can just take that instance_number.dcm and diagnose that the patient has some kind of disease at x,y?\nBut for other images in that series, which do not appear in the train_label_coordinates.csv instance_number, are they useless, or less useful for diagnosing compared to the ones whose instance numbers are collected in train_label_coordinates.csv?\n\nI can't speak on an expert's behalf, but I think that point is where the epicenter of the condition is located. It may need the adjacent slices or comparison with a slice in the orthogonal plane [that is if a point is on the axial slice we may need the sagittal slice crossing the point to get the full picture].\n\nIt is not that the other slices are not relevant but that we need to locate this region ourselves during prediction as there is no `test_label_coordinates.csv`.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2890389,
      "author_name": "coderrkj",
      "author_url": "",
      "post_date": "06/26/2024 05:23:36",
      "content": "<blockquote>\n  <p>Why does the sample submission row_id correspond only to the study_id, rather than {study_id}{series_id}, or even {study_id}{series_id}_{image number}?</p>\n</blockquote>\n<p>Because the goal of this competition is to predict from Axial T2, Sagittal T1 and Sagittal T2/STIR series the severity (normal/mild to moderate to severe) of the conditions at each level. See the following observation:</p>\n<table>\n<thead>\n<tr>\n<th>condition</th>\n<th>Axial T2</th>\n<th>Sagittal T1</th>\n<th>Sagittal T2/STIR</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>Left Subarticular Stenosis</td>\n<td>9608</td>\n<td>0</td>\n<td>0</td>\n</tr>\n<tr>\n<td>Right Subarticular Stenosis</td>\n<td>9612</td>\n<td>0</td>\n<td>0</td>\n</tr>\n<tr>\n<td>Left Neural Foraminal Narrowing</td>\n<td>0</td>\n<td>9860</td>\n<td>0</td>\n</tr>\n<tr>\n<td>Right Neural Foraminal Narrowing</td>\n<td>0</td>\n<td>9859</td>\n<td>0</td>\n</tr>\n<tr>\n<td>Spinal Canal Stenosis</td>\n<td>0</td>\n<td>5</td>\n<td>9748</td>\n</tr>\n</tbody>\n</table>\n<p>Each condition has a particular series to refer to. As for why there is a deviation with 5 Spinal Canal Stenosis with Sagittal T1 see my comment <a href=\"https://www.kaggle.com/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/514891#2890326\" target=\"_blank\">here</a>.</p>\n<blockquote>\n  <p>Does that imply that a single study is made up of multiple image series captured from different angles of a single person?</p>\n</blockquote>\n<p>Yes, and I suggest you to check out the post: <a href=\"https://www.kaggle.com/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/507101\" target=\"_blank\">Radiologist's Insights I</a></p>\n<blockquote>\n  <p>Does this mean an expert can just take that instance_number.dcm and diagnose that the patient has some kind of disease at x,y?<br>\n  But for other images in that series, which do not appear in the train_label_coordinates.csv instance_number, are they useless, or less useful for diagnosing compared to the ones whose instance numbers are collected in train_label_coordinates.csv?</p>\n</blockquote>\n<p>I can't speak on an expert's behalf, but I think that point is where the epicenter of the condition is located. It may need the adjacent slices or comparison with a slice in the orthogonal plane [that is if a point is on the axial slice we may need the sagittal slice crossing the point to get the full picture].</p>\n<p>It is not that the other slices are not relevant but that we need to locate this region ourselves during prediction as there is no <code>test_label_coordinates.csv</code>.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2890324": "I have a few questions.\n\n- Why does the sample submission row_id correspond only to the study_id, rather than {study_id}{series_id}, or even {study_id}{series_id}_{image number}?\n\n- Does that imply that a single study is made up of multiple image series captured from different angles of a single person? Are we diagnosing this individual to check if any of their five bones (L1L2, L2L3, L3L4, L4L5, L5S1) have any of the five diseases (2 foraminal, 2 subarticular, 1 canal stenosis)?\n\n- Regarding the train_label_coordinates.csv file, the instance_number is just a subset of the image series_id. Does this mean an expert can just take that instance_number.dcm and diagnose that the patient has some kind of disease at x,y? \nBut for other images in that series, which do not appear in the train_label_coordinates.csv instance_number, are they useless, or less useful for diagnosing compared to the ones whose instance numbers are collected in train_label_coordinates.csv?\n\n\n---\n\nThis is really a complex dataset structure🥹",
    "2890389": "> Why does the sample submission row_id correspond only to the study_id, rather than {study_id}{series_id}, or even {study_id}{series_id}_{image number}?\n\nBecause the goal of this competition is to predict from Axial T2, Sagittal T1 and Sagittal T2/STIR series the severity (normal/mild to moderate to severe) of the conditions at each level. See the following observation:\n| condition                        |   Axial T2 |   Sagittal T1 |   Sagittal T2/STIR |\n|:---------------------------------|-----------:|--------------:|-------------------:|\n| Left Subarticular Stenosis       |       9608 |             0 |                  0 |\n| Right Subarticular Stenosis      |       9612 |             0 |                  0 |\n| Left Neural Foraminal Narrowing  |          0 |          9860 |                  0 |\n| Right Neural Foraminal Narrowing |          0 |          9859 |                  0 |\n| Spinal Canal Stenosis            |          0 |             5 |               9748 |\n\nEach condition has a particular series to refer to. As for why there is a deviation with 5 Spinal Canal Stenosis with Sagittal T1 see my comment [here](https://www.kaggle.com/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/514891#2890326).\n\n> Does that imply that a single study is made up of multiple image series captured from different angles of a single person?\n\nYes, and I suggest you to check out the post: [Radiologist's Insights I](https://www.kaggle.com/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/507101)\n\n> Does this mean an expert can just take that instance_number.dcm and diagnose that the patient has some kind of disease at x,y?\nBut for other images in that series, which do not appear in the train_label_coordinates.csv instance_number, are they useless, or less useful for diagnosing compared to the ones whose instance numbers are collected in train_label_coordinates.csv?\n\nI can't speak on an expert's behalf, but I think that point is where the epicenter of the condition is located. It may need the adjacent slices or comparison with a slice in the orthogonal plane [that is if a point is on the axial slice we may need the sagittal slice crossing the point to get the full picture].\n\nIt is not that the other slices are not relevant but that we need to locate this region ourselves during prediction as there is no `test_label_coordinates.csv`."
  },
  "source": "meta"
}