{
  "id": 145669,
  "title": "What is the purpose of the multi-resolution tiff files?",
  "url": "/competitions/prostate-cancer-grade-assessment/discussion/145669",
  "author_name": "Matt",
  "post_date": "2020-04-24T03:34:55.041000",
  "votes": 3,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Why does each tiff file contain the WSI at several different resolutions? Is there any downside to just using the highest resolution (except for computational efficiency)?</p>",
  "messages": [
    {
      "id": 818783,
      "postDate": "2020-04-24T05:57:59.267Z",
      "content": "<p>The main point is indeed computational efficiency, which allows for example viewing these images in custom-built viewer software (e.g. QuPath, ASAP, Cytomine). In some cancer detection approaches, the low-resolution levels are used to first pick suspicious regions, and only these regions are then further analyzed at high-resolution.</p>\n\n<p>A thing to consider may be memory usage vs. resolution - using a lower resolution may allow, for example, feeding larger image regions as input to DNNs without running out of memory, thus capturing more tissue context in each input tile. But in principle, given unlimited computational/memory constraints, I doubt there would be any downsides with always using the maximal resolution.</p>",
      "rawMarkdown": "The main point is indeed computational efficiency, which allows for example viewing these images in custom-built viewer software (e.g. QuPath, ASAP, Cytomine). In some cancer detection approaches, the low-resolution levels are used to first pick suspicious regions, and only these regions are then further analyzed at high-resolution.\n\nA thing to consider may be memory usage vs. resolution - using a lower resolution may allow, for example, feeding larger image regions as input to DNNs without running out of memory, thus capturing more tissue context in each input tile. But in principle, given unlimited computational/memory constraints, I doubt there would be any downsides with always using the maximal resolution.",
      "votes": 9,
      "replies": [
        {
          "id": 818984,
          "postDate": "2020-04-24T09:05:58.733Z",
          "content": "<p>Thank you so much for this detailed and thoughtful answer!</p>",
          "rawMarkdown": "Thank you so much for this detailed and thoughtful answer!"
        }
      ]
    },
    {
      "id": 818679,
      "postDate": "2020-04-24T03:34:55.040Z",
      "content": "<p>Why does each tiff file contain the WSI at several different resolutions? Is there any downside to just using the highest resolution (except for computational efficiency)?</p>",
      "rawMarkdown": "Why does each tiff file contain the WSI at several different resolutions? Is there any downside to just using the highest resolution (except for computational efficiency)?",
      "votes": 3
    }
  ],
  "comments": [
    {
      "id": 818783,
      "author_name": "Kimmo Kartasalo",
      "author_url": "",
      "post_date": "2020-04-24T05:57:59.267000",
      "content": "<p>The main point is indeed computational efficiency, which allows for example viewing these images in custom-built viewer software (e.g. QuPath, ASAP, Cytomine). In some cancer detection approaches, the low-resolution levels are used to first pick suspicious regions, and only these regions are then further analyzed at high-resolution.</p>\n\n<p>A thing to consider may be memory usage vs. resolution - using a lower resolution may allow, for example, feeding larger image regions as input to DNNs without running out of memory, thus capturing more tissue context in each input tile. But in principle, given unlimited computational/memory constraints, I doubt there would be any downsides with always using the maximal resolution.</p>",
      "votes": 9,
      "replies": [
        {
          "id": 818984,
          "author_name": "Matt",
          "author_url": "",
          "post_date": "2020-04-24T09:05:58.733000",
          "content": "<p>Thank you so much for this detailed and thoughtful answer!</p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "818783": "The main point is indeed computational efficiency, which allows for example viewing these images in custom-built viewer software (e.g. QuPath, ASAP, Cytomine). In some cancer detection approaches, the low-resolution levels are used to first pick suspicious regions, and only these regions are then further analyzed at high-resolution.\n\nA thing to consider may be memory usage vs. resolution - using a lower resolution may allow, for example, feeding larger image regions as input to DNNs without running out of memory, thus capturing more tissue context in each input tile. But in principle, given unlimited computational/memory constraints, I doubt there would be any downsides with always using the maximal resolution.",
    "818679": "Why does each tiff file contain the WSI at several different resolutions? Is there any downside to just using the highest resolution (except for computational efficiency)?"
  }
}