{
  "id": 183621,
  "title": "Images for negative and positive value of the same week",
  "url": "/competitions/osic-pulmonary-fibrosis-progression/discussion/183621",
  "author_name": "",
  "post_date": "2020-09-17T13:43:11.885063200Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hello!</p>\n<p>When parsing sample_submission.csv for predictions, I found such entries like:<br>\n       ID00419637202311204720264_-12, …<br>\nand<br>\n       ID00419637202311204720264_12, …</p>\n<p>While image train/ID00419637202311204720264/12.dcm do exists,<br>\ntrain/ID00419637202311204720264/-12.dcm doesn't, so how to make predictions for negative weeks?</p>",
  "messages": [
    {
      "id": "1014500",
      "postDate": "09/17/2020 13:43:11",
      "content": "<p>Hello!</p>\n<p>When parsing sample_submission.csv for predictions, I found such entries like:<br>\n       ID00419637202311204720264_-12, …<br>\nand<br>\n       ID00419637202311204720264_12, …</p>\n<p>While image train/ID00419637202311204720264/12.dcm do exists,<br>\ntrain/ID00419637202311204720264/-12.dcm doesn't, so how to make predictions for negative weeks?</p>",
      "rawMarkdown": "Hello!\n\nWhen parsing sample_submission.csv for predictions, I found such entries like:\n       ID00419637202311204720264_-12, ...\nand\n       ID00419637202311204720264_12, ...\n\nWhile image train/ID00419637202311204720264/12.dcm do exists,\ntrain/ID00419637202311204720264/-12.dcm doesn't, so how to make predictions for negative weeks?",
      "votes": null
    },
    {
      "id": "1014580",
      "postDate": "09/17/2020 15:04:31",
      "content": "<p>Imaging (.dcm files) is only done at week 0 (or if it is done more often, we are not given these images). Imaging and spirometry (FVC, Percent etc. in the tabular data) are not necessarily done on the same dates. </p>\n<p>See this information in the competition description:</p>\n<blockquote>\n  <p>In the dataset, you are provided with a baseline chest CT scan and associated clinical information for a set of patients. A patient has an image acquired at time Week = 0 and has numerous follow up visits over the course of approximately 1-2 years, at which time their FVC is measured. </p>\n</blockquote>\n<p>In the sample submission file the part of the string after \"_\" indeed refers to the week. In contrast, I believe (admittedly, I could not quickly find a 100% clear statement about this) that for the .dcm files, the number xx in \"/xx.dcm\" refers to the layer/slice-index of an image in the 3D CT scan (i.e. it is 3D in the sense that you stack image 1.dcm -&gt; 2.dcm -&gt; 3.dcm -&gt; … to \"zoom\" through the 3D imaging layer by layer). I think there's some pretty nice notebooks that explore the data that and a number of nice GIFs people have created that illustrate this.</p>",
      "rawMarkdown": "Imaging (.dcm files) is only done at week 0 (or if it is done more often, we are not given these images). Imaging and spirometry (FVC, Percent etc. in the tabular data) are not necessarily done on the same dates. \n\nSee this information in the competition description:\n> In the dataset, you are provided with a baseline chest CT scan and associated clinical information for a set of patients. A patient has an image acquired at time Week = 0 and has numerous follow up visits over the course of approximately 1-2 years, at which time their FVC is measured. \n\nIn the sample submission file the part of the string after \"_\" indeed refers to the week. In contrast, I believe (admittedly, I could not quickly find a 100% clear statement about this) that for the .dcm files, the number xx in \"/xx.dcm\" refers to the layer/slice-index of an image in the 3D CT scan (i.e. it is 3D in the sense that you stack image 1.dcm -> 2.dcm -> 3.dcm -> ... to \"zoom\" through the 3D imaging layer by layer). I think there's some pretty nice notebooks that explore the data that and a number of nice GIFs people have created that illustrate this.",
      "votes": null
    },
    {
      "id": "1014611",
      "postDate": "09/17/2020 15:36:35",
      "content": "<p>Thanks a lot, Kaggling Gander! ;)</p>",
      "rawMarkdown": "Thanks a lot, Kaggling Gander! ;)",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1014580,
      "author_name": "bjoernholzhauer",
      "author_url": "",
      "post_date": "09/17/2020 15:04:31",
      "content": "<p>Imaging (.dcm files) is only done at week 0 (or if it is done more often, we are not given these images). Imaging and spirometry (FVC, Percent etc. in the tabular data) are not necessarily done on the same dates. </p>\n<p>See this information in the competition description:</p>\n<blockquote>\n  <p>In the dataset, you are provided with a baseline chest CT scan and associated clinical information for a set of patients. A patient has an image acquired at time Week = 0 and has numerous follow up visits over the course of approximately 1-2 years, at which time their FVC is measured. </p>\n</blockquote>\n<p>In the sample submission file the part of the string after \"_\" indeed refers to the week. In contrast, I believe (admittedly, I could not quickly find a 100% clear statement about this) that for the .dcm files, the number xx in \"/xx.dcm\" refers to the layer/slice-index of an image in the 3D CT scan (i.e. it is 3D in the sense that you stack image 1.dcm -&gt; 2.dcm -&gt; 3.dcm -&gt; … to \"zoom\" through the 3D imaging layer by layer). I think there's some pretty nice notebooks that explore the data that and a number of nice GIFs people have created that illustrate this.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1014611,
      "author_name": "polishch",
      "author_url": "",
      "post_date": "09/17/2020 15:36:35",
      "content": "<p>Thanks a lot, Kaggling Gander! ;)</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1014500": "Hello!\n\nWhen parsing sample_submission.csv for predictions, I found such entries like:\n       ID00419637202311204720264_-12, ...\nand\n       ID00419637202311204720264_12, ...\n\nWhile image train/ID00419637202311204720264/12.dcm do exists,\ntrain/ID00419637202311204720264/-12.dcm doesn't, so how to make predictions for negative weeks?",
    "1014580": "Imaging (.dcm files) is only done at week 0 (or if it is done more often, we are not given these images). Imaging and spirometry (FVC, Percent etc. in the tabular data) are not necessarily done on the same dates. \n\nSee this information in the competition description:\n> In the dataset, you are provided with a baseline chest CT scan and associated clinical information for a set of patients. A patient has an image acquired at time Week = 0 and has numerous follow up visits over the course of approximately 1-2 years, at which time their FVC is measured. \n\nIn the sample submission file the part of the string after \"_\" indeed refers to the week. In contrast, I believe (admittedly, I could not quickly find a 100% clear statement about this) that for the .dcm files, the number xx in \"/xx.dcm\" refers to the layer/slice-index of an image in the 3D CT scan (i.e. it is 3D in the sense that you stack image 1.dcm -> 2.dcm -> 3.dcm -> ... to \"zoom\" through the 3D imaging layer by layer). I think there's some pretty nice notebooks that explore the data that and a number of nice GIFs people have created that illustrate this.",
    "1014611": "Thanks a lot, Kaggling Gander! ;)"
  },
  "source": "meta"
}