{ "id": 530418, "title": "How to handle multiple input images for single study", "url": "/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/530418", "author_name": "Pragya Mishra", "post_date": "2024-08-26T14:38:14.486000", "votes": 0, "comment_count": 5, "views": 0, "content": "

How do I perform prediction if there are multiple images for a single study_id , I have no idea how to handle it. Can anyone suggest how this can be achieved?

", "messages": [ { "id": 2971220, "postDate": "2024-08-27T01:14:29.217Z", "content": "

Here are three possible approaches:

\n
    \n

  1. Independent Predictions with Aggregation
    \nMake independent predictions and then aggregate the results.

    2. \n

  2. Sequence Modeling
    \nDesign a sequence model and treat the images like a time series.

    3. \n

  3. Ensemble Modeling
    \nUse an ensemble model by training different models for different inputs and then integrating them.

    4. \n
\n

I hope this provides you with some useful guidance.

", "rawMarkdown": "Here are three possible approaches:\n\n1. **Independent Predictions with Aggregation** \nMake independent predictions and then aggregate the results.\n\n2. **Sequence Modeling**\nDesign a sequence model and treat the images like a time series.\n\n3. **Ensemble Modeling**\nUse an ensemble model by training different models for different inputs and then integrating them.\n\nI hope this provides you with some useful guidance.", "votes": 4, "replies": [ { "id": 2973775, "postDate": "2024-08-29T22:21:36.343Z", "rawMarkdown": "", "isDeleted": true }, { "id": 2980760, "postDate": "2024-09-06T06:34:45.147Z", "content": "

Thanks thats really helpful I will try them out.

", "rawMarkdown": "Thanks thats really helpful I will try them out." } ] }, { "id": 2992911, "postDate": "2024-09-19T08:34:57.247Z", "content": "

Hey @pragyamishra23 ,
\nmany people have tried directly merging folders and then sorting them with ImagePositionPatient in Metadata

", "rawMarkdown": "Hey @pragyamishra23 ,\nmany people have tried directly merging folders and then sorting them with ImagePositionPatient in Metadata" }, { "id": 2973829, "postDate": "2024-08-30T01:48:07.823Z", "content": "

For Tensorflow Implementation

\n

you can build a model with multiple input and multiple outputs using the functional API

", "rawMarkdown": "**For Tensorflow Implementation**\n\nyou can build a model with multiple input and multiple outputs using the functional API" }, { "id": 2970840, "postDate": "2024-08-26T14:38:14.487Z", "content": "

How do I perform prediction if there are multiple images for a single study_id , I have no idea how to handle it. Can anyone suggest how this can be achieved?

", "rawMarkdown": "How do I perform prediction if there are multiple images for a single study_id , I have no idea how to handle it. Can anyone suggest how this can be achieved?" } ], "comments": [ { "id": 2971220, "author_name": "ylxy123", "author_url": "", "post_date": "2024-08-27T01:14:29.217000", "content": "

Here are three possible approaches:

\n
    \n

  1. Independent Predictions with Aggregation
    \nMake independent predictions and then aggregate the results.

    2. \n

  2. Sequence Modeling
    \nDesign a sequence model and treat the images like a time series.

    3. \n

  3. Ensemble Modeling
    \nUse an ensemble model by training different models for different inputs and then integrating them.

    4. \n
\n

I hope this provides you with some useful guidance.

", "votes": 4, "replies": [ { "id": 2973775, "author_name": "", "author_url": "", "post_date": "2024-08-29T22:21:36.343000", "content": "", "votes": 0, "replies": [] }, { "id": 2980760, "author_name": "Pragya Mishra", "author_url": "", "post_date": "2024-09-06T06:34:45.147000", "content": "

Thanks thats really helpful I will try them out.

", "votes": 0, "replies": [] } ] }, { "id": 2992911, "author_name": "Rohit Chaudhari", "author_url": "", "post_date": "2024-09-19T08:34:57.247000", "content": "

Hey @pragyamishra23 ,
\nmany people have tried directly merging folders and then sorting them with ImagePositionPatient in Metadata

", "votes": 0, "replies": [] }, { "id": 2973829, "author_name": "James Ezechinyere", "author_url": "", "post_date": "2024-08-30T01:48:07.823000", "content": "

For Tensorflow Implementation

\n

you can build a model with multiple input and multiple outputs using the functional API

", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "2971220": "Here are three possible approaches:\n\n1. **Independent Predictions with Aggregation** \nMake independent predictions and then aggregate the results.\n\n2. **Sequence Modeling**\nDesign a sequence model and treat the images like a time series.\n\n3. **Ensemble Modeling**\nUse an ensemble model by training different models for different inputs and then integrating them.\n\nI hope this provides you with some useful guidance.", "2992911": "Hey @pragyamishra23 ,\nmany people have tried directly merging folders and then sorting them with ImagePositionPatient in Metadata", "2973829": "**For Tensorflow Implementation**\n\nyou can build a model with multiple input and multiple outputs using the functional API", "2970840": "How do I perform prediction if there are multiple images for a single study_id , I have no idea how to handle it. Can anyone suggest how this can be achieved?" } }