{
  "id": 47620,
  "title": "Separate mask for each nuclei or whole mask for the image?",
  "url": "/competitions/data-science-bowl-2018/discussion/47620",
  "author_name": "",
  "post_date": "2018-01-17T03:27:59.014158500Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>After applying U-net, I get a whole image mask. Should I split it into separate nuclei mask?</p>",
  "messages": [
    {
      "id": "269632",
      "postDate": "01/17/2018 03:27:59",
      "content": "<p>After applying U-net, I get a whole image mask. Should I split it into separate nuclei mask?</p>",
      "rawMarkdown": "After applying U-net, I get a whole image mask. Should I split it into separate nuclei mask?",
      "votes": null
    },
    {
      "id": "269750",
      "postDate": "01/17/2018 08:01:00",
      "content": "<p>yes, that is what the skimage.measure.label function does in this kernel (<a href=\"https://www.kaggle.com/kmader/nuclei-overview-to-submission\">https://www.kaggle.com/kmader/nuclei-overview-to-submission</a>). It takes the mask and performs connected component labeling in order to show the different masks and generate them independently. Cells that are touching get classified together so doing a watershed would probably be better.</p>",
      "rawMarkdown": "yes, that is what the skimage.measure.label function does in this kernel (https://www.kaggle.com/kmader/nuclei-overview-to-submission). It takes the mask and performs connected component labeling in order to show the different masks and generate them independently. Cells that are touching get classified together so doing a watershed would probably be better.",
      "votes": null
    },
    {
      "id": "271521",
      "postDate": "01/20/2018 16:44:38",
      "content": "<p>thank you!</p>",
      "rawMarkdown": "thank you!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 269750,
      "author_name": "kmader",
      "author_url": "",
      "post_date": "01/17/2018 08:01:00",
      "content": "<p>yes, that is what the skimage.measure.label function does in this kernel (<a href=\"https://www.kaggle.com/kmader/nuclei-overview-to-submission\">https://www.kaggle.com/kmader/nuclei-overview-to-submission</a>). It takes the mask and performs connected component labeling in order to show the different masks and generate them independently. Cells that are touching get classified together so doing a watershed would probably be better.</p>",
      "votes": null,
      "replies": [
        {
          "id": 271521,
          "author_name": "xwzy66",
          "author_url": "",
          "post_date": "01/20/2018 16:44:38",
          "content": "<p>thank you!</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "269632": "After applying U-net, I get a whole image mask. Should I split it into separate nuclei mask?",
    "269750": "yes, that is what the skimage.measure.label function does in this kernel (https://www.kaggle.com/kmader/nuclei-overview-to-submission). It takes the mask and performs connected component labeling in order to show the different masks and generate them independently. Cells that are touching get classified together so doing a watershed would probably be better.",
    "271521": "thank you!"
  },
  "source": "meta"
}