{
  "id": 253818,
  "title": "3D medical image pretrained segmentation nets",
  "url": "/competitions/rsna-miccai-brain-tumor-radiogenomic-classification/discussion/253818",
  "author_name": "",
  "post_date": "2021-07-18T19:37:47.766713100Z",
  "votes": 10,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Found this article on arxiv recently by researchers from Tencent (the colossal Chinese tech company):<br>\n<a href=\"https://arxiv.org/pdf/1904.00625.pdf\" target=\"_blank\">Med3D: Transfer Learning for 3D Medical Image Analysis</a></p>\n<p>They've collected together 3D MRI &amp; CT segmentation datasets from eight different competitions (see below), and used it to pretrain a segmentation network they call Med3D. </p>\n<p><img src=\"https://i.postimg.cc/K8jH40MG/Screen-Shot-2021-07-19-at-5-25-30-am.png\" alt=\"\"></p>\n<p>They then demonstrate its abilities via transfer learning on some other challenges like LiTS.</p>\n<p>Pre-trained models and some code is at this repo: <a href=\"https://github.com/Tencent/MedicalNet\" target=\"_blank\">https://github.com/Tencent/MedicalNet</a></p>",
  "messages": [
    {
      "id": "1392543",
      "postDate": "07/18/2021 19:37:47",
      "content": "<p>Found this article on arxiv recently by researchers from Tencent (the colossal Chinese tech company):<br>\n<a href=\"https://arxiv.org/pdf/1904.00625.pdf\" target=\"_blank\">Med3D: Transfer Learning for 3D Medical Image Analysis</a></p>\n<p>They've collected together 3D MRI &amp; CT segmentation datasets from eight different competitions (see below), and used it to pretrain a segmentation network they call Med3D. </p>\n<p><img src=\"https://i.postimg.cc/K8jH40MG/Screen-Shot-2021-07-19-at-5-25-30-am.png\" alt=\"\"></p>\n<p>They then demonstrate its abilities via transfer learning on some other challenges like LiTS.</p>\n<p>Pre-trained models and some code is at this repo: <a href=\"https://github.com/Tencent/MedicalNet\" target=\"_blank\">https://github.com/Tencent/MedicalNet</a></p>",
      "rawMarkdown": "Found this article on arxiv recently by researchers from Tencent (the colossal Chinese tech company):\n[Med3D: Transfer Learning for 3D Medical Image Analysis](https://arxiv.org/pdf/1904.00625.pdf)\n\nThey've collected together 3D MRI & CT segmentation datasets from eight different competitions (see below), and used it to pretrain a segmentation network they call Med3D. \n\n![](https://i.postimg.cc/K8jH40MG/Screen-Shot-2021-07-19-at-5-25-30-am.png)\n\nThey then demonstrate its abilities via transfer learning on some other challenges like LiTS.\n\nPre-trained models and some code is at this repo: https://github.com/Tencent/MedicalNet",
      "votes": null
    },
    {
      "id": "1405960",
      "postDate": "07/31/2021 11:08:58",
      "content": "<p><a href=\"https://www.kaggle.com/reubenschmidt\" target=\"_blank\">@reubenschmidt</a> thanks for providing the Github link! I think it will be useful to have this in the future.  <br>\nHowever, because I am new to CV and medical imaging, how can the information you provided be used for such a competition? Thanks, again.</p>",
      "rawMarkdown": "reubenschmidt thanks for providing the Github link! I think it will be useful to have this in the future.  \nHowever, because I am new to CV and medical imaging, how can the information you provided be used for such a competition? Thanks, again.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1405960,
      "author_name": "elenaeb",
      "author_url": "",
      "post_date": "07/31/2021 11:08:58",
      "content": "<p><a href=\"https://www.kaggle.com/reubenschmidt\" target=\"_blank\">@reubenschmidt</a> thanks for providing the Github link! I think it will be useful to have this in the future.  <br>\nHowever, because I am new to CV and medical imaging, how can the information you provided be used for such a competition? Thanks, again.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1392543": "Found this article on arxiv recently by researchers from Tencent (the colossal Chinese tech company):\n[Med3D: Transfer Learning for 3D Medical Image Analysis](https://arxiv.org/pdf/1904.00625.pdf)\n\nThey've collected together 3D MRI & CT segmentation datasets from eight different competitions (see below), and used it to pretrain a segmentation network they call Med3D. \n\n![](https://i.postimg.cc/K8jH40MG/Screen-Shot-2021-07-19-at-5-25-30-am.png)\n\nThey then demonstrate its abilities via transfer learning on some other challenges like LiTS.\n\nPre-trained models and some code is at this repo: https://github.com/Tencent/MedicalNet",
    "1405960": "reubenschmidt thanks for providing the Github link! I think it will be useful to have this in the future.  \nHowever, because I am new to CV and medical imaging, how can the information you provided be used for such a competition? Thanks, again."
  },
  "source": "meta"
}