{
  "id": 36595,
  "title": "Healthy images",
  "url": "/competitions/diabetic-retinopathy-detection/discussion/36595",
  "author_name": "",
  "post_date": "2017-07-18T22:27:59.770091500Z",
  "votes": null,
  "comment_count": 3,
  "views": 0,
  "content": "<p>How can I differentiate between healthy and sick images? I read that there is a classification but I don't see where I can see that </p>",
  "messages": [
    {
      "id": "204541",
      "postDate": "07/18/2017 22:27:59",
      "content": "<p>How can I differentiate between healthy and sick images? I read that there is a classification but I don't see where I can see that </p>",
      "rawMarkdown": "How can I differentiate between healthy and sick images? I read that there is a classification but I don't see where I can see that",
      "votes": null
    },
    {
      "id": "209619",
      "postDate": "08/02/2017 20:37:29",
      "content": "<p>Healthy images are rated 0 in the training images package.</p>",
      "rawMarkdown": "Healthy images are rated 0 in the training images package.",
      "votes": null
    },
    {
      "id": "299009",
      "postDate": "03/20/2018 09:39:55",
      "content": "<p>This is a multi-label classification problem, but you can actually convert it into a binary classification one! \nImagine that you have your labels as an array like: labels = [0,1,2,0,0,0,4,0,0]</p>\n\n<p>You can convert the array into only two labels, where 0 is translated to <em>No Disease</em> and everything else is <em>Disease</em>. If you're using Python, you can use <strong><em>List Comprehension</em></strong>:</p>\n\n<p><code>labels = [1 if i != 0 else 0 for i in labels]</code></p>\n\n<p>Result: [0,1,2,0,0,0,4,0,0] --&gt; [0,1,1,0,0,0,1,0,0]</p>\n\n<p>The relation between image and its label is in the <em>trainLabels.csv</em> file :)</p>",
      "rawMarkdown": "This is a multi-label classification problem, but you can actually convert it into a binary classification one! \nImagine that you have your labels as an array like: labels = [0,1,2,0,0,0,4,0,0]\n\nYou can convert the array into only two labels, where 0 is translated to *No Disease* and everything else is *Disease*. If you're using Python, you can use ***List Comprehension***:\n\n`labels = [1 if i != 0 else 0 for i in labels]`\n\nResult: [0,1,2,0,0,0,4,0,0] --&gt; [0,1,1,0,0,0,1,0,0]\n\nThe relation between image and its label is in the *trainLabels.csv* file :)",
      "votes": null
    },
    {
      "id": "307518",
      "postDate": "04/01/2018 22:58:12",
      "content": "<p>That aprouch is not usefull for ophthalmologists because first stages of retinopathy do not need laser treatment, just blood sugar control.</p>",
      "rawMarkdown": "That aprouch is not usefull for ophthalmologists because first stages of retinopathy do not need laser treatment, just blood sugar control.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 209619,
      "author_name": "stephanehenry",
      "author_url": "",
      "post_date": "08/02/2017 20:37:29",
      "content": "<p>Healthy images are rated 0 in the training images package.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 299009,
      "author_name": "mafaldafalcaotvf",
      "author_url": "",
      "post_date": "03/20/2018 09:39:55",
      "content": "<p>This is a multi-label classification problem, but you can actually convert it into a binary classification one! \nImagine that you have your labels as an array like: labels = [0,1,2,0,0,0,4,0,0]</p>\n\n<p>You can convert the array into only two labels, where 0 is translated to <em>No Disease</em> and everything else is <em>Disease</em>. If you're using Python, you can use <strong><em>List Comprehension</em></strong>:</p>\n\n<p><code>labels = [1 if i != 0 else 0 for i in labels]</code></p>\n\n<p>Result: [0,1,2,0,0,0,4,0,0] --&gt; [0,1,1,0,0,0,1,0,0]</p>\n\n<p>The relation between image and its label is in the <em>trainLabels.csv</em> file :)</p>",
      "votes": null,
      "replies": [
        {
          "id": 307518,
          "author_name": "masterdkh",
          "author_url": "",
          "post_date": "04/01/2018 22:58:12",
          "content": "<p>That aprouch is not usefull for ophthalmologists because first stages of retinopathy do not need laser treatment, just blood sugar control.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "204541": "How can I differentiate between healthy and sick images? I read that there is a classification but I don't see where I can see that",
    "209619": "Healthy images are rated 0 in the training images package.",
    "299009": "This is a multi-label classification problem, but you can actually convert it into a binary classification one! \nImagine that you have your labels as an array like: labels = [0,1,2,0,0,0,4,0,0]\n\nYou can convert the array into only two labels, where 0 is translated to *No Disease* and everything else is *Disease*. If you're using Python, you can use ***List Comprehension***:\n\n`labels = [1 if i != 0 else 0 for i in labels]`\n\nResult: [0,1,2,0,0,0,4,0,0] --&gt; [0,1,1,0,0,0,1,0,0]\n\nThe relation between image and its label is in the *trainLabels.csv* file :)",
    "307518": "That aprouch is not usefull for ophthalmologists because first stages of retinopathy do not need laser treatment, just blood sugar control."
  },
  "source": "meta"
}