{
  "id": 276639,
  "title": "NVIDIA: Top Spots in MICCAI 2021 Brain Tumor Segmentation Challenge",
  "url": "/competitions/rsna-miccai-brain-tumor-radiogenomic-classification/discussion/276639",
  "author_name": "",
  "post_date": "2021-10-05T16:49:06.200570100Z",
  "votes": 25,
  "comment_count": 2,
  "views": 0,
  "content": "<p>NVIDIA data scientists placed three of the top 10 spots in a <a href=\"http://www.braintumorsegmentation.org/\" target=\"_blank\">brain tumor segmentation challenge</a> validation phase at the prestigious <a href=\"https://miccai2021.org/en/\" target=\"_blank\">MICCAI 2021</a> medical imaging conference.</p>\n<p>In essence, </p>\n<blockquote>\n  <p>NVIDIA developers ranked No. 1, No. 2, and No. 7 in the challenge’s validation phase, each creating different types of AI model approaches for tumor segmentation — including an <strong>optimized U-Net</strong> model, a <strong>SegResNet</strong> model with automatic hyperparameter optimization, and a <strong>Swin UNETR</strong> model which is a transformer-based approach for computer vision.</p>\n</blockquote>\n<p>Interesting solutions, read the full post: <a href=\"https://developer.nvidia.com/blog/nvidia-data-scientists-take-top-spots-in-miccai-2021-brain-tumor-segmentation-challenge/?ncid=so-nvsh-894437-vt14&amp;dysig_tid=7f9e5959b54c45438f50736f64235021#cid=ix03_so-nvsh_en-us\" target=\"_blank\">NVIDIA BraTS Solutions 2021</a>. </p>\n<hr>\n<h2>1. Optimized U-Net for Brain Tumor Segmentation – Rank #1</h2>\n<p><img src=\"https://user-images.githubusercontent.com/17668390/136066525-319b14c1-ee9d-4aa7-ae0f-df9107d75cab.png\" alt=\"BRaTS-fig-1\"></p>\n<h2>2. SegResNet: Redundancy Reduction in Semantic Segmentation of 3D Brain MRIs – Rank #2</h2>\n<p><img src=\"https://user-images.githubusercontent.com/17668390/136066620-4d1c0a5c-5b72-40e1-9817-abfb79c17dd1.png\" alt=\"BRaTS-fig-2\"></p>\n<h2>3. Swin UNETR: Shifted Window Transformers for 3D Semantic Segmentation of Brain Tumors – Rank #7</h2>\n<p><img src=\"https://user-images.githubusercontent.com/17668390/136066678-8fb19691-6b72-463e-b9ba-b80281c214ae.png\" alt=\"BRaTS-fig-3\"></p>",
  "messages": [
    {
      "id": "1535294",
      "postDate": "10/05/2021 16:49:06",
      "content": "<p>NVIDIA data scientists placed three of the top 10 spots in a <a href=\"http://www.braintumorsegmentation.org/\" target=\"_blank\">brain tumor segmentation challenge</a> validation phase at the prestigious <a href=\"https://miccai2021.org/en/\" target=\"_blank\">MICCAI 2021</a> medical imaging conference.</p>\n<p>In essence, </p>\n<blockquote>\n  <p>NVIDIA developers ranked No. 1, No. 2, and No. 7 in the challenge’s validation phase, each creating different types of AI model approaches for tumor segmentation — including an <strong>optimized U-Net</strong> model, a <strong>SegResNet</strong> model with automatic hyperparameter optimization, and a <strong>Swin UNETR</strong> model which is a transformer-based approach for computer vision.</p>\n</blockquote>\n<p>Interesting solutions, read the full post: <a href=\"https://developer.nvidia.com/blog/nvidia-data-scientists-take-top-spots-in-miccai-2021-brain-tumor-segmentation-challenge/?ncid=so-nvsh-894437-vt14&amp;dysig_tid=7f9e5959b54c45438f50736f64235021#cid=ix03_so-nvsh_en-us\" target=\"_blank\">NVIDIA BraTS Solutions 2021</a>. </p>\n<hr>\n<h2>1. Optimized U-Net for Brain Tumor Segmentation – Rank #1</h2>\n<p><img src=\"https://user-images.githubusercontent.com/17668390/136066525-319b14c1-ee9d-4aa7-ae0f-df9107d75cab.png\" alt=\"BRaTS-fig-1\"></p>\n<h2>2. SegResNet: Redundancy Reduction in Semantic Segmentation of 3D Brain MRIs – Rank #2</h2>\n<p><img src=\"https://user-images.githubusercontent.com/17668390/136066620-4d1c0a5c-5b72-40e1-9817-abfb79c17dd1.png\" alt=\"BRaTS-fig-2\"></p>\n<h2>3. Swin UNETR: Shifted Window Transformers for 3D Semantic Segmentation of Brain Tumors – Rank #7</h2>\n<p><img src=\"https://user-images.githubusercontent.com/17668390/136066678-8fb19691-6b72-463e-b9ba-b80281c214ae.png\" alt=\"BRaTS-fig-3\"></p>",
      "rawMarkdown": "NVIDIA data scientists placed three of the top 10 spots in a [brain tumor segmentation challenge](http://www.braintumorsegmentation.org/) validation phase at the prestigious [MICCAI 2021](https://miccai2021.org/en/) medical imaging conference.\n\nIn essence, \n\n> NVIDIA developers ranked No. 1, No. 2, and No. 7 in the challenge’s validation phase, each creating different types of AI model approaches for tumor segmentation — including an **optimized U-Net** model, a **SegResNet** model with automatic hyperparameter optimization, and a **Swin UNETR** model which is a transformer-based approach for computer vision.\n\nInteresting solutions, read the full post: [NVIDIA BraTS Solutions 2021](https://developer.nvidia.com/blog/nvidia-data-scientists-take-top-spots-in-miccai-2021-brain-tumor-segmentation-challenge/?ncid=so-nvsh-894437-vt14&dysig_tid=7f9e5959b54c45438f50736f64235021#cid=ix03_so-nvsh_en-us). \n\n--- \n\n## 1. Optimized U-Net for Brain Tumor Segmentation – Rank #1\n\n![BRaTS-fig-1](https://user-images.githubusercontent.com/17668390/136066525-319b14c1-ee9d-4aa7-ae0f-df9107d75cab.png)\n\n\n## 2. SegResNet: Redundancy Reduction in Semantic Segmentation of 3D Brain MRIs – Rank #2\n\n![BRaTS-fig-2](https://user-images.githubusercontent.com/17668390/136066620-4d1c0a5c-5b72-40e1-9817-abfb79c17dd1.png)\n\n## 3. Swin UNETR: Shifted Window Transformers for 3D Semantic Segmentation of Brain Tumors – Rank #7\n\n![BRaTS-fig-3](https://user-images.githubusercontent.com/17668390/136066678-8fb19691-6b72-463e-b9ba-b80281c214ae.png)",
      "votes": null
    },
    {
      "id": "1541807",
      "postDate": "10/11/2021 21:10:19",
      "content": "<p>Very interesting approaches! Have you implemented any of them? </p>",
      "rawMarkdown": "Very interesting approaches! Have you implemented any of them?",
      "votes": null
    },
    {
      "id": "1546094",
      "postDate": "10/15/2021 19:33:54",
      "content": "<p><a href=\"https://www.kaggle.com/tegzes\" target=\"_blank\">@tegzes</a> <br>\nUnfortunately no, couldn't manage time for it. <br>\nHowever, if possible I will be trying to implement the Swin UNETR later.  </p>",
      "rawMarkdown": "tegzes \nUnfortunately no, couldn't manage time for it. \nHowever, if possible I will be trying to implement the Swin UNETR later.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1541807,
      "author_name": "tegzes",
      "author_url": "",
      "post_date": "10/11/2021 21:10:19",
      "content": "<p>Very interesting approaches! Have you implemented any of them? </p>",
      "votes": null,
      "replies": [
        {
          "id": 1546094,
          "author_name": "ipythonx",
          "author_url": "",
          "post_date": "10/15/2021 19:33:54",
          "content": "<p><a href=\"https://www.kaggle.com/tegzes\" target=\"_blank\">@tegzes</a> <br>\nUnfortunately no, couldn't manage time for it. <br>\nHowever, if possible I will be trying to implement the Swin UNETR later.  </p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1535294": "NVIDIA data scientists placed three of the top 10 spots in a [brain tumor segmentation challenge](http://www.braintumorsegmentation.org/) validation phase at the prestigious [MICCAI 2021](https://miccai2021.org/en/) medical imaging conference.\n\nIn essence, \n\n> NVIDIA developers ranked No. 1, No. 2, and No. 7 in the challenge’s validation phase, each creating different types of AI model approaches for tumor segmentation — including an **optimized U-Net** model, a **SegResNet** model with automatic hyperparameter optimization, and a **Swin UNETR** model which is a transformer-based approach for computer vision.\n\nInteresting solutions, read the full post: [NVIDIA BraTS Solutions 2021](https://developer.nvidia.com/blog/nvidia-data-scientists-take-top-spots-in-miccai-2021-brain-tumor-segmentation-challenge/?ncid=so-nvsh-894437-vt14&dysig_tid=7f9e5959b54c45438f50736f64235021#cid=ix03_so-nvsh_en-us). \n\n--- \n\n## 1. Optimized U-Net for Brain Tumor Segmentation – Rank #1\n\n![BRaTS-fig-1](https://user-images.githubusercontent.com/17668390/136066525-319b14c1-ee9d-4aa7-ae0f-df9107d75cab.png)\n\n\n## 2. SegResNet: Redundancy Reduction in Semantic Segmentation of 3D Brain MRIs – Rank #2\n\n![BRaTS-fig-2](https://user-images.githubusercontent.com/17668390/136066620-4d1c0a5c-5b72-40e1-9817-abfb79c17dd1.png)\n\n## 3. Swin UNETR: Shifted Window Transformers for 3D Semantic Segmentation of Brain Tumors – Rank #7\n\n![BRaTS-fig-3](https://user-images.githubusercontent.com/17668390/136066678-8fb19691-6b72-463e-b9ba-b80281c214ae.png)",
    "1541807": "Very interesting approaches! Have you implemented any of them?",
    "1546094": "tegzes \nUnfortunately no, couldn't manage time for it. \nHowever, if possible I will be trying to implement the Swin UNETR later."
  },
  "source": "meta"
}