{
  "id": 181873,
  "title": "Understanding Submission and some basic insights from the data",
  "url": "/competitions/rsna-str-pulmonary-embolism-detection/discussion/181873",
  "author_name": "prk007",
  "post_date": "2020-09-10T13:25:56.738000",
  "votes": 3,
  "comment_count": 2,
  "views": 0,
  "content": "<p>From the data we can say that we n have studies, and each study has x no. of dicom files, I found that each study =&gt; each patient's whole lung CT scan and each image from a study(dicom image) =&gt; a slice from patients CT scan.</p>\n<p>Now coming to predictions</p>\n<p>There are 2 levels of predictions that we have to submit:</p>\n<ul>\n<li><p>Patient level prediction</p></li>\n<li><p>Slice level prediction</p>\n<p>so the 1st set of predictions will be at patient level and the rest are at slice level.(make sure you read the data page to get more clarity on how to submit the preds)</p></li>\n</ul>\n<p>Out of all the given tabular data, the labels 'QA Contrast', 'QA Motion', 'Flow Artifact' label, 'True filling defect not PE' labels are just for information but using 'True filling defect not PE' can be used to predict non PE patients.</p>\n<p>refer to <a href=\"url\" target=\"_blank\">https://www.kaggle.com/prk007/insights-from-tabular-and-image-data</a></p>",
  "messages": [
    {
      "id": 1005404,
      "postDate": "2020-09-10T13:25:56.740Z",
      "content": "<p>From the data we can say that we n have studies, and each study has x no. of dicom files, I found that each study =&gt; each patient's whole lung CT scan and each image from a study(dicom image) =&gt; a slice from patients CT scan.</p>\n<p>Now coming to predictions</p>\n<p>There are 2 levels of predictions that we have to submit:</p>\n<ul>\n<li><p>Patient level prediction</p></li>\n<li><p>Slice level prediction</p>\n<p>so the 1st set of predictions will be at patient level and the rest are at slice level.(make sure you read the data page to get more clarity on how to submit the preds)</p></li>\n</ul>\n<p>Out of all the given tabular data, the labels 'QA Contrast', 'QA Motion', 'Flow Artifact' label, 'True filling defect not PE' labels are just for information but using 'True filling defect not PE' can be used to predict non PE patients.</p>\n<p>refer to <a href=\"url\" target=\"_blank\">https://www.kaggle.com/prk007/insights-from-tabular-and-image-data</a></p>",
      "rawMarkdown": "From the data we can say that we n have studies, and each study has x no. of dicom files, I found that each study => each patient's whole lung CT scan and each image from a study(dicom image) => a slice from patients CT scan.\n\nNow coming to predictions\n\nThere are 2 levels of predictions that we have to submit:\n - Patient level prediction\n - Slice level prediction\n\n so the 1st set of predictions will be at patient level and the rest are at slice level.(make sure you read the data page to get more clarity on how to submit the preds)\n\nOut of all the given tabular data, the labels 'QA Contrast', 'QA Motion', 'Flow Artifact' label, 'True filling defect not PE' labels are just for information but using 'True filling defect not PE' can be used to predict non PE patients.\n\nrefer to [https://www.kaggle.com/prk007/insights-from-tabular-and-image-data](url)\n",
      "votes": 3
    },
    {
      "id": 1006715,
      "postDate": "2020-09-11T14:00:00.463Z",
      "content": "<p>Thanks for clearing up the confusion of the submission method. </p>",
      "rawMarkdown": "Thanks for clearing up the confusion of the submission method. ",
      "votes": 1,
      "replies": [
        {
          "id": 1006869,
          "postDate": "2020-09-11T15:48:19.943Z",
          "content": "<p>u r welcome</p>",
          "rawMarkdown": "u r welcome"
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 1006715,
      "author_name": "Redwan Sony",
      "author_url": "",
      "post_date": "2020-09-11T14:00:00.463000",
      "content": "<p>Thanks for clearing up the confusion of the submission method. </p>",
      "votes": 1,
      "replies": [
        {
          "id": 1006869,
          "author_name": "prk007",
          "author_url": "",
          "post_date": "2020-09-11T15:48:19.943000",
          "content": "<p>u r welcome</p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1005404": "From the data we can say that we n have studies, and each study has x no. of dicom files, I found that each study => each patient's whole lung CT scan and each image from a study(dicom image) => a slice from patients CT scan.\n\nNow coming to predictions\n\nThere are 2 levels of predictions that we have to submit:\n - Patient level prediction\n - Slice level prediction\n\n so the 1st set of predictions will be at patient level and the rest are at slice level.(make sure you read the data page to get more clarity on how to submit the preds)\n\nOut of all the given tabular data, the labels 'QA Contrast', 'QA Motion', 'Flow Artifact' label, 'True filling defect not PE' labels are just for information but using 'True filling defect not PE' can be used to predict non PE patients.\n\nrefer to [https://www.kaggle.com/prk007/insights-from-tabular-and-image-data](url)\n",
    "1006715": "Thanks for clearing up the confusion of the submission method. "
  }
}