{
  "id": 146651,
  "title": "How can I efficiently store image patches?",
  "url": "/competitions/prostate-cancer-grade-assessment/discussion/146651",
  "author_name": "Matt",
  "post_date": "2020-04-28T00:36:23.037000",
  "votes": 4,
  "comment_count": 0,
  "views": 0,
  "content": "<p>I've been trying to extract patches from whole slide images but I've had trouble storing them in a reasonable size. I extract square patches 512px in width from the highest resolution slide images. The patches do not overlap and I filter off any that contain no/little biopsy.</p>\n\n<p>I've tried many different methods to save: saving as PNG, saving as numpy (compressed and uncompressed), saving with HDF5, etc. but everytime the patches end up taking much more space than even the original tiff. How can I efficiently store patches so that it doesn't take a terabyte? 😆 </p>",
  "messages": [
    {
      "id": 823857,
      "postDate": "2020-04-28T00:36:23.037Z",
      "content": "<p>I've been trying to extract patches from whole slide images but I've had trouble storing them in a reasonable size. I extract square patches 512px in width from the highest resolution slide images. The patches do not overlap and I filter off any that contain no/little biopsy.</p>\n\n<p>I've tried many different methods to save: saving as PNG, saving as numpy (compressed and uncompressed), saving with HDF5, etc. but everytime the patches end up taking much more space than even the original tiff. How can I efficiently store patches so that it doesn't take a terabyte? 😆 </p>",
      "rawMarkdown": "I've been trying to extract patches from whole slide images but I've had trouble storing them in a reasonable size. I extract square patches 512px in width from the highest resolution slide images. The patches do not overlap and I filter off any that contain no/little biopsy.\n\nI've tried many different methods to save: saving as PNG, saving as numpy (compressed and uncompressed), saving with HDF5, etc. but everytime the patches end up taking much more space than even the original tiff. How can I efficiently store patches so that it doesn't take a terabyte? 😆 ",
      "votes": 4
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "823857": "I've been trying to extract patches from whole slide images but I've had trouble storing them in a reasonable size. I extract square patches 512px in width from the highest resolution slide images. The patches do not overlap and I filter off any that contain no/little biopsy.\n\nI've tried many different methods to save: saving as PNG, saving as numpy (compressed and uncompressed), saving with HDF5, etc. but everytime the patches end up taking much more space than even the original tiff. How can I efficiently store patches so that it doesn't take a terabyte? 😆 "
  }
}