{
  "id": 369341,
  "title": "More PNG/JPG Datasets to Get Started",
  "url": "/competitions/rsna-breast-cancer-detection/discussion/369341",
  "author_name": "Theo Viel",
  "post_date": "2022-11-29T22:12:07.662000",
  "votes": 48,
  "comment_count": 7,
  "views": 0,
  "content": "<p>Similar <a href=\"https://www.kaggle.com/radek1\" target=\"_blank\">@radek1</a>'s <a href=\"https://www.kaggle.com/competitions/rsna-breast-cancer-detection/discussion/369282\" target=\"_blank\">thread</a>, I've also started the (boring) task of converting the dicom images to pngs/jpgs. My code was made public a few hours ago already, after fixing some issues I'm starting to upload the datasets !</p>\n<ul>\n<li><strong>Code :</strong> <a href=\"https://www.kaggle.com/code/theoviel/dicom-resized-png-jpg/\" target=\"_blank\">Dicom -&gt; Resized PNG/JPG</a>. You can regenerate the data with the parameters of your choice. </li>\n<li><a href=\"https://www.kaggle.com/datasets/theoviel/rsna-breast-cancer-256-pngs\" target=\"_blank\">256x256 pngs</a> - to train your first models, or if you don't have a lot of compute power</li>\n<li><a href=\"https://www.kaggle.com/datasets/theoviel/rsna-breast-cancer-512-pngs\" target=\"_blank\">512x512 pngs</a> - to build more competitive models.</li>\n<li><a href=\"https://www.kaggle.com/datasets/theoviel/rsna-breast-cancer-768-pngs\" target=\"_blank\">768x768 pngs</a> - perhaps bigger images are better ?</li>\n<li><a href=\"https://www.kaggle.com/datasets/theoviel/rsna-breast-cancer-1024-pngs\" target=\"_blank\">1024x1024 pngs</a> - if you don't know what to do with your compute</li>\n</ul>",
  "messages": [
    {
      "id": 2049039,
      "postDate": "2022-11-29T22:12:07.663Z",
      "content": "<p>Similar <a href=\"https://www.kaggle.com/radek1\" target=\"_blank\">@radek1</a>'s <a href=\"https://www.kaggle.com/competitions/rsna-breast-cancer-detection/discussion/369282\" target=\"_blank\">thread</a>, I've also started the (boring) task of converting the dicom images to pngs/jpgs. My code was made public a few hours ago already, after fixing some issues I'm starting to upload the datasets !</p>\n<ul>\n<li><strong>Code :</strong> <a href=\"https://www.kaggle.com/code/theoviel/dicom-resized-png-jpg/\" target=\"_blank\">Dicom -&gt; Resized PNG/JPG</a>. You can regenerate the data with the parameters of your choice. </li>\n<li><a href=\"https://www.kaggle.com/datasets/theoviel/rsna-breast-cancer-256-pngs\" target=\"_blank\">256x256 pngs</a> - to train your first models, or if you don't have a lot of compute power</li>\n<li><a href=\"https://www.kaggle.com/datasets/theoviel/rsna-breast-cancer-512-pngs\" target=\"_blank\">512x512 pngs</a> - to build more competitive models.</li>\n<li><a href=\"https://www.kaggle.com/datasets/theoviel/rsna-breast-cancer-768-pngs\" target=\"_blank\">768x768 pngs</a> - perhaps bigger images are better ?</li>\n<li><a href=\"https://www.kaggle.com/datasets/theoviel/rsna-breast-cancer-1024-pngs\" target=\"_blank\">1024x1024 pngs</a> - if you don't know what to do with your compute</li>\n</ul>",
      "rawMarkdown": "\nSimilar @radek1's [thread](https://www.kaggle.com/competitions/rsna-breast-cancer-detection/discussion/369282), I've also started the (boring) task of converting the dicom images to pngs/jpgs. My code was made public a few hours ago already, after fixing some issues I'm starting to upload the datasets !\n\n- **Code :** [Dicom -> Resized PNG/JPG](https://www.kaggle.com/code/theoviel/dicom-resized-png-jpg/). You can regenerate the data with the parameters of your choice. \n-  [256x256 pngs](https://www.kaggle.com/datasets/theoviel/rsna-breast-cancer-256-pngs) - to train your first models, or if you don't have a lot of compute power\n- [512x512 pngs](https://www.kaggle.com/datasets/theoviel/rsna-breast-cancer-512-pngs) - to build more competitive models.\n- [768x768 pngs](https://www.kaggle.com/datasets/theoviel/rsna-breast-cancer-768-pngs) - perhaps bigger images are better ?\n-  [1024x1024 pngs](https://www.kaggle.com/datasets/theoviel/rsna-breast-cancer-1024-pngs) - if you don't know what to do with your compute",
      "votes": 48
    },
    {
      "id": 2054729,
      "postDate": "2022-12-04T12:03:21.817Z",
      "content": "<p>Thanks it helps to start quickly.<br>\nBTW: 822_1942326353 is not a breast</p>",
      "rawMarkdown": "Thanks it helps to start quickly.\nBTW: 822_1942326353 is not a breast",
      "votes": 3
    },
    {
      "id": 2049044,
      "postDate": "2022-11-29T22:24:48.520Z",
      "content": "<p>Thanks for sharing !</p>\n<p>Edit : the 256x256 link is not working.</p>",
      "rawMarkdown": "Thanks for sharing !\n\nEdit : the 256x256 link is not working.",
      "votes": 1,
      "replies": [
        {
          "id": 2049047,
          "postDate": "2022-11-29T22:29:13.563Z",
          "content": "<p>It's not ready yet, but will be in a few minutes :)</p>",
          "rawMarkdown": "It's not ready yet, but will be in a few minutes :)",
          "votes": 1
        }
      ]
    },
    {
      "id": 2056101,
      "postDate": "2022-12-05T18:24:06.673Z",
      "content": "<p>I recommend the highest resolution you can handle since many mammography findings are small, particularly calcifications</p>",
      "rawMarkdown": "I recommend the highest resolution you can handle since many mammography findings are small, particularly calcifications",
      "votes": 2
    },
    {
      "id": 2054728,
      "postDate": "2022-12-04T11:54:53.063Z",
      "content": "<p>do you have any experience with what resolution is the best?</p>",
      "rawMarkdown": "do you have any experience with what resolution is the best?",
      "replies": [
        {
          "id": 2134337,
          "postDate": "2023-02-07T22:49:53.613Z",
          "content": "<p>Original resolution and original bit depth is best according to my own tests---that is for learning/classifying breast cancer subcategories. Luckily there is usually quite a bit of background to crop. 👍</p>",
          "rawMarkdown": "Original resolution and original bit depth is best according to my own tests---that is for learning/classifying breast cancer subcategories. Luckily there is usually quite a bit of background to crop. 👍"
        }
      ]
    },
    {
      "id": 2051135,
      "postDate": "2022-12-01T08:31:23.207Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 2054729,
      "author_name": "MPWARE",
      "author_url": "",
      "post_date": "2022-12-04T12:03:21.817000",
      "content": "<p>Thanks it helps to start quickly.<br>\nBTW: 822_1942326353 is not a breast</p>",
      "votes": 3,
      "replies": []
    },
    {
      "id": 2049044,
      "author_name": "Vincent Schuler",
      "author_url": "",
      "post_date": "2022-11-29T22:24:48.520000",
      "content": "<p>Thanks for sharing !</p>\n<p>Edit : the 256x256 link is not working.</p>",
      "votes": 1,
      "replies": [
        {
          "id": 2049047,
          "author_name": "Theo Viel",
          "author_url": "",
          "post_date": "2022-11-29T22:29:13.563000",
          "content": "<p>It's not ready yet, but will be in a few minutes :)</p>",
          "votes": 1,
          "replies": []
        }
      ]
    },
    {
      "id": 2056101,
      "author_name": "Hari T.",
      "author_url": "",
      "post_date": "2022-12-05T18:24:06.673000",
      "content": "<p>I recommend the highest resolution you can handle since many mammography findings are small, particularly calcifications</p>",
      "votes": 2,
      "replies": []
    },
    {
      "id": 2054728,
      "author_name": "Jirka",
      "author_url": "",
      "post_date": "2022-12-04T11:54:53.063000",
      "content": "<p>do you have any experience with what resolution is the best?</p>",
      "votes": 0,
      "replies": [
        {
          "id": 2134337,
          "author_name": "Antti Isosalo",
          "author_url": "",
          "post_date": "2023-02-07T22:49:53.613000",
          "content": "<p>Original resolution and original bit depth is best according to my own tests---that is for learning/classifying breast cancer subcategories. Luckily there is usually quite a bit of background to crop. 👍</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 2051135,
      "author_name": "",
      "author_url": "",
      "post_date": "2022-12-01T08:31:23.207000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2049039": "\nSimilar @radek1's [thread](https://www.kaggle.com/competitions/rsna-breast-cancer-detection/discussion/369282), I've also started the (boring) task of converting the dicom images to pngs/jpgs. My code was made public a few hours ago already, after fixing some issues I'm starting to upload the datasets !\n\n- **Code :** [Dicom -> Resized PNG/JPG](https://www.kaggle.com/code/theoviel/dicom-resized-png-jpg/). You can regenerate the data with the parameters of your choice. \n-  [256x256 pngs](https://www.kaggle.com/datasets/theoviel/rsna-breast-cancer-256-pngs) - to train your first models, or if you don't have a lot of compute power\n- [512x512 pngs](https://www.kaggle.com/datasets/theoviel/rsna-breast-cancer-512-pngs) - to build more competitive models.\n- [768x768 pngs](https://www.kaggle.com/datasets/theoviel/rsna-breast-cancer-768-pngs) - perhaps bigger images are better ?\n-  [1024x1024 pngs](https://www.kaggle.com/datasets/theoviel/rsna-breast-cancer-1024-pngs) - if you don't know what to do with your compute",
    "2054729": "Thanks it helps to start quickly.\nBTW: 822_1942326353 is not a breast",
    "2049044": "Thanks for sharing !\n\nEdit : the 256x256 link is not working.",
    "2056101": "I recommend the highest resolution you can handle since many mammography findings are small, particularly calcifications",
    "2054728": "do you have any experience with what resolution is the best?",
    "2051135": ""
  }
}