{
  "id": 613348,
  "title": "Is this direction reasonable?",
  "url": "/competitions/physionet-ecg-image-digitization/discussion/613348",
  "author_name": "",
  "post_date": "2025-10-26T05:52:23.219508600Z",
  "votes": 1,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Hello. I'm struggling with ECG analysis. I would like to get your opinions. <br>\nThe following is the analysis procedure I am considering.</p>\n<ol>\n<li>Extract waveforms from ECG images with noise using segmentation.</li>\n<li>Convert the extracted waveforms into data.<br>\nIs it reasonable to perform the analysis in this order? Please, let me know!!</li>\n</ol>",
  "messages": [
    {
      "id": "3307098",
      "postDate": "10/26/2025 05:52:23",
      "content": "<p>Hello. I'm struggling with ECG analysis. I would like to get your opinions. <br>\nThe following is the analysis procedure I am considering.</p>\n<ol>\n<li>Extract waveforms from ECG images with noise using segmentation.</li>\n<li>Convert the extracted waveforms into data.<br>\nIs it reasonable to perform the analysis in this order? Please, let me know!!</li>\n</ol>",
      "rawMarkdown": "Hello. I'm struggling with ECG analysis. I would like to get your opinions. \nThe following is the analysis procedure I am considering.\n1. Extract waveforms from ECG images with noise using segmentation.\n2. Convert the extracted waveforms into data.\nIs it reasonable to perform the analysis in this order? Please, let me know!!",
      "votes": null
    },
    {
      "id": "3307530",
      "postDate": "10/27/2025 08:16:05",
      "content": "<p>Yes, you're right. However, it's necessary to build a model for segmentation first, and then perform sequence reconstruction on the segmented electrocardiograms to convert them into tabular data.</p>",
      "rawMarkdown": "Yes, you're right. However, it's necessary to build a model for segmentation first, and then perform sequence reconstruction on the segmented electrocardiograms to convert them into tabular data.",
      "votes": null
    },
    {
      "id": "3307599",
      "postDate": "10/27/2025 11:44:32",
      "content": "<p>Thank you for your reply!<br>\nI’m glad to know that my approach isn’t wrong.<br>\nI have one question, though…<br>\nFor segmentation, ground truth data is required — does that mean I have to create it myself?</p>",
      "rawMarkdown": "Thank you for your reply!\nI’m glad to know that my approach isn’t wrong.\nI have one question, though…\nFor segmentation, ground truth data is required — does that mean I have to create it myself?",
      "votes": null
    },
    {
      "id": "3309745",
      "postDate": "11/01/2025 09:45:27",
      "content": "<p>yes, or perhaps use a pre-trained model</p>",
      "rawMarkdown": "yes, or perhaps use a pre-trained model",
      "votes": null
    },
    {
      "id": "3312064",
      "postDate": "11/06/2025 11:16:58",
      "content": "<p>Isn't the annotation work insanely tough!?</p>",
      "rawMarkdown": "Isn't the annotation work insanely tough!?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3307530,
      "author_name": "qifeihhh666",
      "author_url": "",
      "post_date": "10/27/2025 08:16:05",
      "content": "<p>Yes, you're right. However, it's necessary to build a model for segmentation first, and then perform sequence reconstruction on the segmented electrocardiograms to convert them into tabular data.</p>",
      "votes": null,
      "replies": [
        {
          "id": 3307599,
          "author_name": "nakaosyoya",
          "author_url": "",
          "post_date": "10/27/2025 11:44:32",
          "content": "<p>Thank you for your reply!<br>\nI’m glad to know that my approach isn’t wrong.<br>\nI have one question, though…<br>\nFor segmentation, ground truth data is required — does that mean I have to create it myself?</p>",
          "votes": null,
          "replies": [
            {
              "id": 3309745,
              "author_name": "eliasstenhede",
              "author_url": "",
              "post_date": "11/01/2025 09:45:27",
              "content": "<p>yes, or perhaps use a pre-trained model</p>",
              "votes": null,
              "replies": [
                {
                  "id": 3312064,
                  "author_name": "nakaosyoya",
                  "author_url": "",
                  "post_date": "11/06/2025 11:16:58",
                  "content": "<p>Isn't the annotation work insanely tough!?</p>",
                  "votes": null,
                  "replies": []
                }
              ]
            }
          ]
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "3307098": "Hello. I'm struggling with ECG analysis. I would like to get your opinions. \nThe following is the analysis procedure I am considering.\n1. Extract waveforms from ECG images with noise using segmentation.\n2. Convert the extracted waveforms into data.\nIs it reasonable to perform the analysis in this order? Please, let me know!!",
    "3307530": "Yes, you're right. However, it's necessary to build a model for segmentation first, and then perform sequence reconstruction on the segmented electrocardiograms to convert them into tabular data.",
    "3307599": "Thank you for your reply!\nI’m glad to know that my approach isn’t wrong.\nI have one question, though…\nFor segmentation, ground truth data is required — does that mean I have to create it myself?",
    "3309745": "yes, or perhaps use a pre-trained model",
    "3312064": "Isn't the annotation work insanely tough!?"
  },
  "source": "meta"
}