{
  "id": 663749,
  "title": "Strategies for Obtaining Signal Segmentation Masks from ECG Images",
  "url": "/competitions/physionet-ecg-image-digitization/discussion/663749",
  "author_name": "Luca",
  "post_date": "2025-12-20T06:21:27.517000",
  "votes": 3,
  "comment_count": 0,
  "views": 0,
  "content": "<p>After reviewing the current discussion threads, I haven't seen much detail on the methodology for generating segmentation masks of the signal. I’m looking for effective ways to generate segmentation masks for training (on either original or unwarped PNGs).</p>\n<p>Currently, I'm considering:</p>\n<ul>\n<li>Back-projecting GT signals onto image pixel space.</li>\n<li>Re-simulate using the ECG simulator to generate synthetic images with masks.</li>\n</ul>\n<p>Is there a more direct \"standard\" approach I'm missing? Would love to brainstorm alternative workflows or hear others' thoughts on handling this. Thanks.</p>",
  "messages": [
    {
      "id": 3379569,
      "postDate": "2025-12-20T06:21:27.517Z",
      "content": "<p>After reviewing the current discussion threads, I haven't seen much detail on the methodology for generating segmentation masks of the signal. I’m looking for effective ways to generate segmentation masks for training (on either original or unwarped PNGs).</p>\n<p>Currently, I'm considering:</p>\n<ul>\n<li>Back-projecting GT signals onto image pixel space.</li>\n<li>Re-simulate using the ECG simulator to generate synthetic images with masks.</li>\n</ul>\n<p>Is there a more direct \"standard\" approach I'm missing? Would love to brainstorm alternative workflows or hear others' thoughts on handling this. Thanks.</p>",
      "rawMarkdown": "After reviewing the current discussion threads, I haven't seen much detail on the methodology for generating segmentation masks of the signal. I’m looking for effective ways to generate segmentation masks for training (on either original or unwarped PNGs).\n\nCurrently, I'm considering:\n- Back-projecting GT signals onto image pixel space.\n- Re-simulate using the ECG simulator to generate synthetic images with masks.\n\nIs there a more direct \"standard\" approach I'm missing? Would love to brainstorm alternative workflows or hear others' thoughts on handling this. Thanks.",
      "votes": 3
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "3379569": "After reviewing the current discussion threads, I haven't seen much detail on the methodology for generating segmentation masks of the signal. I’m looking for effective ways to generate segmentation masks for training (on either original or unwarped PNGs).\n\nCurrently, I'm considering:\n- Back-projecting GT signals onto image pixel space.\n- Re-simulate using the ECG simulator to generate synthetic images with masks.\n\nIs there a more direct \"standard\" approach I'm missing? Would love to brainstorm alternative workflows or hear others' thoughts on handling this. Thanks."
  }
}