{
  "id": 276809,
  "title": "💥which one is better, train model with all modality as input or train model in diffrent step?💥",
  "url": "/competitions/rsna-miccai-brain-tumor-radiogenomic-classification/discussion/276809",
  "author_name": "",
  "post_date": "2021-10-06T13:19:12.604573600Z",
  "votes": 2,
  "comment_count": 4,
  "views": 0,
  "content": "<p>I trained 3D and 2D models. I use all modality images at once or train models in 4 step with different modality . which one is better and for predict what is the best modality to use as input?</p>",
  "messages": [
    {
      "id": "1536104",
      "postDate": "10/06/2021 13:19:12",
      "content": "<p>I trained 3D and 2D models. I use all modality images at once or train models in 4 step with different modality . which one is better and for predict what is the best modality to use as input?</p>",
      "rawMarkdown": "I trained 3D and 2D models. I use all modality images at once or train models in 4 step with different modality . which one is better and for predict what is the best modality to use as input?",
      "votes": null
    },
    {
      "id": "1536487",
      "postDate": "10/06/2021 18:20:22",
      "content": "<p>In my case, I tried 2D as well as 3D models, 2D works better in my case and as mentioned in other discussion topics, FLAIR and T2 seems to give good results.</p>\n<p>May I ask whether you have utilized the same model in 4 steps for different four modalities?</p>",
      "rawMarkdown": "In my case, I tried 2D as well as 3D models, 2D works better in my case and as mentioned in other discussion topics, FLAIR and T2 seems to give good results.\n\nMay I ask whether you have utilized the same model in 4 steps for different four modalities?",
      "votes": null
    },
    {
      "id": "1538364",
      "postDate": "10/08/2021 10:46:30",
      "content": "<p>I trained model 4 time with different modality as input and save weight of each model </p>",
      "rawMarkdown": "I trained model 4 time with different modality as input and save weight of each model",
      "votes": null
    },
    {
      "id": "1539732",
      "postDate": "10/09/2021 17:54:00",
      "content": "<p>Training in different modalities and merging them gave me better results.</p>",
      "rawMarkdown": "Training in different modalities and merging them gave me better results.",
      "votes": null
    },
    {
      "id": "1541297",
      "postDate": "10/11/2021 11:41:53",
      "content": "<p>How did you merge the result?</p>",
      "rawMarkdown": "How did you merge the result?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1536487,
      "author_name": "wfarzana",
      "author_url": "",
      "post_date": "10/06/2021 18:20:22",
      "content": "<p>In my case, I tried 2D as well as 3D models, 2D works better in my case and as mentioned in other discussion topics, FLAIR and T2 seems to give good results.</p>\n<p>May I ask whether you have utilized the same model in 4 steps for different four modalities?</p>",
      "votes": null,
      "replies": [
        {
          "id": 1538364,
          "author_name": "givkashi",
          "author_url": "",
          "post_date": "10/08/2021 10:46:30",
          "content": "<p>I trained model 4 time with different modality as input and save weight of each model </p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1539732,
      "author_name": "gunesevitan",
      "author_url": "",
      "post_date": "10/09/2021 17:54:00",
      "content": "<p>Training in different modalities and merging them gave me better results.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1541297,
          "author_name": "givkashi",
          "author_url": "",
          "post_date": "10/11/2021 11:41:53",
          "content": "<p>How did you merge the result?</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1536104": "I trained 3D and 2D models. I use all modality images at once or train models in 4 step with different modality . which one is better and for predict what is the best modality to use as input?",
    "1536487": "In my case, I tried 2D as well as 3D models, 2D works better in my case and as mentioned in other discussion topics, FLAIR and T2 seems to give good results.\n\nMay I ask whether you have utilized the same model in 4 steps for different four modalities?",
    "1538364": "I trained model 4 time with different modality as input and save weight of each model",
    "1539732": "Training in different modalities and merging them gave me better results.",
    "1541297": "How did you merge the result?"
  },
  "source": "meta"
}