{
  "id": 183836,
  "title": "Train PNG(512x512)",
  "url": "/competitions/rsna-str-pulmonary-embolism-detection/discussion/183836",
  "author_name": "SumanSudhir",
  "post_date": "2020-09-18T07:57:41.158000",
  "votes": 9,
  "comment_count": 8,
  "views": 0,
  "content": "<p>Have generated a 3-channel RGB dataset of size 512x512 using three different windows suggested by <a href=\"https://www.kaggle.com/vaillant\" target=\"_blank\">@vaillant</a> <a href=\"https://www.kaggle.com/c/rsna-str-pulmonary-embolism-detection/discussion/182930\" target=\"_blank\">here</a> from the provided DICOM images.</p>\n<ul>\n<li><strong>RED</strong> channel: <strong>MEDIASTINAL</strong> window [40,400]</li>\n<li><strong>GREEN</strong> channel: <strong>PE</strong> window [100,700]</li>\n<li><strong>BLUE</strong> channel: <strong>LUNG</strong> Window[-600,1500]<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2149138%2F0ccaac09367d54d1a87d1a92d7b04e0e%2Fpreprocessing.png?generation=1600415091234100&amp;alt=media\" alt=\"\"></li>\n</ul>\n<p>I will try to upload the dataset if anyone needs it btw the size of the generated data is 507 GB. I am not sure Kaggle will allow to upload it.</p>",
  "messages": [
    {
      "id": 1015468,
      "postDate": "2020-09-18T07:57:41.160Z",
      "content": "<p>Have generated a 3-channel RGB dataset of size 512x512 using three different windows suggested by <a href=\"https://www.kaggle.com/vaillant\" target=\"_blank\">@vaillant</a> <a href=\"https://www.kaggle.com/c/rsna-str-pulmonary-embolism-detection/discussion/182930\" target=\"_blank\">here</a> from the provided DICOM images.</p>\n<ul>\n<li><strong>RED</strong> channel: <strong>MEDIASTINAL</strong> window [40,400]</li>\n<li><strong>GREEN</strong> channel: <strong>PE</strong> window [100,700]</li>\n<li><strong>BLUE</strong> channel: <strong>LUNG</strong> Window[-600,1500]<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2149138%2F0ccaac09367d54d1a87d1a92d7b04e0e%2Fpreprocessing.png?generation=1600415091234100&amp;alt=media\" alt=\"\"></li>\n</ul>\n<p>I will try to upload the dataset if anyone needs it btw the size of the generated data is 507 GB. I am not sure Kaggle will allow to upload it.</p>",
      "rawMarkdown": "Have generated a 3-channel RGB dataset of size 512x512 using three different windows suggested by @vaillant [here](https://www.kaggle.com/c/rsna-str-pulmonary-embolism-detection/discussion/182930) from the provided DICOM images.\n\n- **RED** channel: **MEDIASTINAL** window [40,400]\n- **GREEN** channel: **PE** window [100,700]\n- **BLUE** channel: **LUNG** Window[-600,1500]\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2149138%2F0ccaac09367d54d1a87d1a92d7b04e0e%2Fpreprocessing.png?generation=1600415091234100&alt=media)\n\nI will try to upload the dataset if anyone needs it btw the size of the generated data is 507 GB. I am not sure Kaggle will allow to upload it.",
      "votes": 9
    },
    {
      "id": 1033574,
      "postDate": "2020-10-01T05:22:53.420Z",
      "content": "<p>Why you all use these three window  [40,400], [100,700] and [-600,1500] ?<br>\nHow about other window?</p>",
      "rawMarkdown": "Why you all use these three window  [40,400], [100,700] and [-600,1500] ?\nHow about other window?",
      "votes": 1,
      "replies": [
        {
          "id": 1033589,
          "postDate": "2020-10-01T05:39:06.470Z",
          "content": "<p>Window value is taken from this RSNA paper <a href=\"https://pubs.rsna.org/doi/pdf/10.1148/rg.245045008\" target=\"_blank\">https://pubs.rsna.org/doi/pdf/10.1148/rg.245045008</a></p>",
          "rawMarkdown": "Window value is taken from this RSNA paper https://pubs.rsna.org/doi/pdf/10.1148/rg.245045008",
          "votes": 1
        },
        {
          "id": 1043650,
          "postDate": "2020-10-09T06:02:12.957Z",
          "content": "<p>Thank you!</p>",
          "rawMarkdown": "Thank you!"
        }
      ]
    },
    {
      "id": 1048567,
      "postDate": "2020-10-13T15:50:07.053Z",
      "content": "<p>did you try uploading the dataset :) and is it helping in lb score ?  </p>",
      "rawMarkdown": "did you try uploading the dataset :) and is it helping in lb score ?  ",
      "replies": [
        {
          "id": 1049119,
          "postDate": "2020-10-14T06:16:20.313Z",
          "content": "<p>I didn't try to upload it as its size is above 500 GB. Here are image generated by a similar approach <a href=\"https://www.kaggle.com/c/rsna-str-pulmonary-embolism-detection/discussion/185742\" target=\"_blank\">link1</a> <a href=\"https://www.kaggle.com/c/rsna-str-pulmonary-embolism-detection/discussion/182930\" target=\"_blank\">link2</a></p>",
          "rawMarkdown": "I didn't try to upload it as its size is above 500 GB. Here are image generated by a similar approach [link1](https://www.kaggle.com/c/rsna-str-pulmonary-embolism-detection/discussion/185742) [link2](https://www.kaggle.com/c/rsna-str-pulmonary-embolism-detection/discussion/182930)"
        }
      ]
    },
    {
      "id": 1016707,
      "postDate": "2020-09-19T07:13:52.173Z",
      "content": "<p>thanks <a href=\"https://www.kaggle.com/sudhiriitb\" target=\"_blank\">@sudhiriitb</a> , can you share how to run parallelism on top of Ian's DICOM img code?🙏🙏</p>",
      "rawMarkdown": "thanks @sudhiriitb , can you share how to run parallelism on top of Ian's DICOM img code?🙏🙏\n",
      "replies": [
        {
          "id": 1017812,
          "postDate": "2020-09-19T09:27:12.903Z",
          "content": "<p><a href=\"https://www.kaggle.com/fiyeroleung\" target=\"_blank\">@fiyeroleung</a> I have added parallelism to his code.<br>\n<a href=\"https://gist.github.com/SumanSudhir/7f3b72510eb2b8cff080a225334f5dc4\" target=\"_blank\">Here is the link to his new code</a> </p>",
          "rawMarkdown": "@fiyeroleung I have added parallelism to his code.\n[Here is the link to his new code](https://gist.github.com/SumanSudhir/7f3b72510eb2b8cff080a225334f5dc4) ",
          "votes": 2
        },
        {
          "id": 1017855,
          "postDate": "2020-09-19T09:51:41.553Z",
          "content": "<p>Thanks so much!  <a href=\"https://www.kaggle.com/sudhiriitb\" target=\"_blank\">@sudhiriitb</a> <br>\nI made a stupid mistake no wonder it doesn't work.<br>\nYour code works like a champ and save me many hours!</p>",
          "rawMarkdown": "Thanks so much!  @sudhiriitb \nI made a stupid mistake no wonder it doesn't work.\nYour code works like a champ and save me many hours!",
          "votes": 2
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 1033574,
      "author_name": "HoneJie Li",
      "author_url": "",
      "post_date": "2020-10-01T05:22:53.420000",
      "content": "<p>Why you all use these three window  [40,400], [100,700] and [-600,1500] ?<br>\nHow about other window?</p>",
      "votes": 1,
      "replies": [
        {
          "id": 1033589,
          "author_name": "SumanSudhir",
          "author_url": "",
          "post_date": "2020-10-01T05:39:06.470000",
          "content": "<p>Window value is taken from this RSNA paper <a href=\"https://pubs.rsna.org/doi/pdf/10.1148/rg.245045008\" target=\"_blank\">https://pubs.rsna.org/doi/pdf/10.1148/rg.245045008</a></p>",
          "votes": 1,
          "replies": []
        },
        {
          "id": 1043650,
          "author_name": "HoneJie Li",
          "author_url": "",
          "post_date": "2020-10-09T06:02:12.957000",
          "content": "<p>Thank you!</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 1048567,
      "author_name": "Shubham Thapa",
      "author_url": "",
      "post_date": "2020-10-13T15:50:07.053000",
      "content": "<p>did you try uploading the dataset :) and is it helping in lb score ?  </p>",
      "votes": 0,
      "replies": [
        {
          "id": 1049119,
          "author_name": "SumanSudhir",
          "author_url": "",
          "post_date": "2020-10-14T06:16:20.313000",
          "content": "<p>I didn't try to upload it as its size is above 500 GB. Here are image generated by a similar approach <a href=\"https://www.kaggle.com/c/rsna-str-pulmonary-embolism-detection/discussion/185742\" target=\"_blank\">link1</a> <a href=\"https://www.kaggle.com/c/rsna-str-pulmonary-embolism-detection/discussion/182930\" target=\"_blank\">link2</a></p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 1016707,
      "author_name": "Salaryman",
      "author_url": "",
      "post_date": "2020-09-19T07:13:52.173000",
      "content": "<p>thanks <a href=\"https://www.kaggle.com/sudhiriitb\" target=\"_blank\">@sudhiriitb</a> , can you share how to run parallelism on top of Ian's DICOM img code?🙏🙏</p>",
      "votes": 0,
      "replies": [
        {
          "id": 1017812,
          "author_name": "SumanSudhir",
          "author_url": "",
          "post_date": "2020-09-19T09:27:12.903000",
          "content": "<p><a href=\"https://www.kaggle.com/fiyeroleung\" target=\"_blank\">@fiyeroleung</a> I have added parallelism to his code.<br>\n<a href=\"https://gist.github.com/SumanSudhir/7f3b72510eb2b8cff080a225334f5dc4\" target=\"_blank\">Here is the link to his new code</a> </p>",
          "votes": 2,
          "replies": []
        },
        {
          "id": 1017855,
          "author_name": "Salaryman",
          "author_url": "",
          "post_date": "2020-09-19T09:51:41.553000",
          "content": "<p>Thanks so much!  <a href=\"https://www.kaggle.com/sudhiriitb\" target=\"_blank\">@sudhiriitb</a> <br>\nI made a stupid mistake no wonder it doesn't work.<br>\nYour code works like a champ and save me many hours!</p>",
          "votes": 2,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1015468": "Have generated a 3-channel RGB dataset of size 512x512 using three different windows suggested by @vaillant [here](https://www.kaggle.com/c/rsna-str-pulmonary-embolism-detection/discussion/182930) from the provided DICOM images.\n\n- **RED** channel: **MEDIASTINAL** window [40,400]\n- **GREEN** channel: **PE** window [100,700]\n- **BLUE** channel: **LUNG** Window[-600,1500]\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2149138%2F0ccaac09367d54d1a87d1a92d7b04e0e%2Fpreprocessing.png?generation=1600415091234100&alt=media)\n\nI will try to upload the dataset if anyone needs it btw the size of the generated data is 507 GB. I am not sure Kaggle will allow to upload it.",
    "1033574": "Why you all use these three window  [40,400], [100,700] and [-600,1500] ?\nHow about other window?",
    "1048567": "did you try uploading the dataset :) and is it helping in lb score ?  ",
    "1016707": "thanks @sudhiriitb , can you share how to run parallelism on top of Ian's DICOM img code?🙏🙏\n"
  }
}