{ "id": 341140, "title": "RadImageNet instead of ImageNet for Transfer Learning in Medical Image!", "url": "/competitions/rsna-2022-cervical-spine-fracture-detection/discussion/341140", "author_name": "Innat", "post_date": "2022-08-01T12:01:13.857000", "votes": 16, "comment_count": 3, "views": 0, "content": "

Rad-ImageNet

\n

Website Paper GitHub

\n

\"drawing\"

\n
\n

RadImageNet is a large radiologic database of annotated medical images from multiple modalities and of multiple pathologies. RadImageNet, consisting of 5 million annotated medical images consisting of CT, MRI, and ultrasound of musculoskeletal, neurologic, oncologic, gastrointestinal, endocrine, and pulmonary pathologies over 450,000 patients. The database is unprecedented in scale and breadth in the medical imaging field, constituting a more appropriate basis for medical imaging transfer learning applications. The RadImageNet database is an open-access medical imaging database. This study was designed to improve transfer learning performance on downstream medical imaging applications. The RadImageNet dataset are available by request at https://www.radimagenet.com/

\n
\n

\"Slide2\"

\n
\n

Note: You probably should not get and upload this data to kaggle to make it publicly accessible. There is some license issue for that, which we should honor.

\n
", "messages": [ { "id": 1880000, "postDate": "2022-08-01T12:01:13.857Z", "content": "

Rad-ImageNet

\n

Website Paper GitHub

\n

\"drawing\"

\n
\n

RadImageNet is a large radiologic database of annotated medical images from multiple modalities and of multiple pathologies. RadImageNet, consisting of 5 million annotated medical images consisting of CT, MRI, and ultrasound of musculoskeletal, neurologic, oncologic, gastrointestinal, endocrine, and pulmonary pathologies over 450,000 patients. The database is unprecedented in scale and breadth in the medical imaging field, constituting a more appropriate basis for medical imaging transfer learning applications. The RadImageNet database is an open-access medical imaging database. This study was designed to improve transfer learning performance on downstream medical imaging applications. The RadImageNet dataset are available by request at https://www.radimagenet.com/

\n
\n

\"Slide2\"

\n
\n

Note: You probably should not get and upload this data to kaggle to make it publicly accessible. There is some license issue for that, which we should honor.

\n
", "rawMarkdown": "## Rad-ImageNet\n\n[Website](https://www.radimagenet.com/) [Paper](https://www.researchgate.net/publication/352302343_RadImageNet_A_Large-scale_Radiologic_Dataset_for_Enhancing_Deep_Learning_Transfer_Learning_Research) [GitHub](https://github.com/BMEII-AI/RadImageNet)\n\n\"drawing\"\n\n> **RadImageNet** is a large radiologic database of annotated medical images from multiple modalities and of multiple pathologies. RadImageNet, consisting of **5 million** annotated medical images consisting of CT, MRI, and ultrasound of musculoskeletal, neurologic, oncologic, gastrointestinal, endocrine, and pulmonary pathologies over **450,000** patients. The database is unprecedented in scale and breadth in the medical imaging field, constituting a more appropriate basis for medical imaging transfer learning applications. The RadImageNet database is an open-access medical imaging database. This study was designed to improve transfer learning performance on downstream medical imaging applications. The RadImageNet dataset are available by request at https://www.radimagenet.com/\n\n![Slide2](https://user-images.githubusercontent.com/17668390/182142468-988376c5-5828-40ce-85e8-c9b2dc8eea8f.jpg)\n\n> **Note**: You probably should not get and upload this data to kaggle to make it **publicly accessible**. There is some license issue for that, which we should honor. ", "votes": 16 }, { "id": 1885689, "postDate": "2022-08-05T10:26:11.793Z", "content": "

RadImageNet 2D Pretrained Weights.

\n

2D pretrained_no_top.
\n3D version. (WIP)

", "rawMarkdown": "RadImageNet 2D Pretrained Weights. \n\n[2D pretrained_no_top.](https://www.kaggle.com/datasets/ipythonx/notop-wg-radimagenet)\n3D version. ([WIP](https://github.com/BMEII-AI/RadImageNet/issues/2))\n\n" }, { "id": 1888825, "postDate": "2022-08-07T19:46:47.320Z", "rawMarkdown": "", "votes": -1, "isDeleted": true, "replies": [ { "id": 1889933, "postDate": "2022-08-08T13:02:41.407Z", "rawMarkdown": "", "isDeleted": true } ] } ], "comments": [ { "id": 1885689, "author_name": "Innat", "author_url": "", "post_date": "2022-08-05T10:26:11.793000", "content": "

RadImageNet 2D Pretrained Weights.

\n

2D pretrained_no_top.
\n3D version. (WIP)

", "votes": 0, "replies": [] }, { "id": 1888825, "author_name": "", "author_url": "", "post_date": "2022-08-07T19:46:47.320000", "content": "", "votes": -1, "replies": [ { "id": 1889933, "author_name": "", "author_url": "", "post_date": "2022-08-08T13:02:41.407000", "content": "", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "1880000": "## Rad-ImageNet\n\n[Website](https://www.radimagenet.com/) [Paper](https://www.researchgate.net/publication/352302343_RadImageNet_A_Large-scale_Radiologic_Dataset_for_Enhancing_Deep_Learning_Transfer_Learning_Research) [GitHub](https://github.com/BMEII-AI/RadImageNet)\n\n\"drawing\"\n\n> **RadImageNet** is a large radiologic database of annotated medical images from multiple modalities and of multiple pathologies. RadImageNet, consisting of **5 million** annotated medical images consisting of CT, MRI, and ultrasound of musculoskeletal, neurologic, oncologic, gastrointestinal, endocrine, and pulmonary pathologies over **450,000** patients. The database is unprecedented in scale and breadth in the medical imaging field, constituting a more appropriate basis for medical imaging transfer learning applications. The RadImageNet database is an open-access medical imaging database. This study was designed to improve transfer learning performance on downstream medical imaging applications. The RadImageNet dataset are available by request at https://www.radimagenet.com/\n\n![Slide2](https://user-images.githubusercontent.com/17668390/182142468-988376c5-5828-40ce-85e8-c9b2dc8eea8f.jpg)\n\n> **Note**: You probably should not get and upload this data to kaggle to make it **publicly accessible**. There is some license issue for that, which we should honor. ", "1885689": "RadImageNet 2D Pretrained Weights. \n\n[2D pretrained_no_top.](https://www.kaggle.com/datasets/ipythonx/notop-wg-radimagenet)\n3D version. ([WIP](https://github.com/BMEII-AI/RadImageNet/issues/2))\n\n", "1888825": "" } }