{ "id": 344342, "title": "Identifying vertebrae C1-C7 in each image", "url": "/competitions/rsna-2022-cervical-spine-fracture-detection/discussion/344342", "author_name": "", "post_date": "2022-08-14T22:53:17.416111200Z", "votes": 7, "comment_count": 6, "views": 0, "content": "

I haven't seen much progress on finding which vertebrae is in each image yet so I have just released my NOTEBOOK. The idea is that we can identify the targets from the unique values in the segmentation masks. I have collected this info and stored it in this DATASET.

\n

\n

The next step would be to build a supervised model (using metadata/images) to predict which vertebrae is in each image for all the other patients in the train (& test) set which don't have segmentation masks. The challenging part is preserving the monotonicity (C1->C2->etc). I haven't figured this out yet so I'm sharing this resource to hopefully kickstart some progress.

", "messages": [ { "id": "1898870", "postDate": "08/14/2022 22:53:17", "content": "

I haven't seen much progress on finding which vertebrae is in each image yet so I have just released my NOTEBOOK. The idea is that we can identify the targets from the unique values in the segmentation masks. I have collected this info and stored it in this DATASET.

\n

\n

The next step would be to build a supervised model (using metadata/images) to predict which vertebrae is in each image for all the other patients in the train (& test) set which don't have segmentation masks. The challenging part is preserving the monotonicity (C1->C2->etc). I haven't figured this out yet so I'm sharing this resource to hopefully kickstart some progress.

", "rawMarkdown": "I haven't seen much progress on **finding which vertebrae is in each image** yet so I have just released my [NOTEBOOK](https://www.kaggle.com/code/samuelcortinhas/extracting-vertebrae-c1-c7). The idea is that we can identify the targets from the **unique values in the segmentation masks**. I have collected this info and stored it in this [DATASET](https://www.kaggle.com/datasets/samuelcortinhas/rsna-2022-spine-fracture-detection-metadata/code). \n\n\n\nThe **next step** would be to **build a supervised model** (using metadata/images) to predict which vertebrae is in each image for **all the other patients** in the train (& test) set which **don't have segmentation masks**. The **challenging** part is **preserving the monotonicity** (C1->C2->etc). I haven't figured this out yet so I'm sharing this resource to hopefully kickstart some progress.", "votes": null }, { "id": "1898913", "postDate": "08/14/2022 23:57:09", "content": "

A simple enough supervised model gets like 85% accuracy (in my tests), won't making it 2.5d solve your problem with monotonicity most of the time, if that does not work, you can always just make a 3d model.

", "rawMarkdown": "A simple enough supervised model gets like 85% accuracy (in my tests), won't making it 2.5d solve your problem with monotonicity most of the time, if that does not work, you can always just make a 3d model.", "votes": null }, { "id": "1899530", "postDate": "08/15/2022 10:27:22", "content": "

@harshitsheoran thanks for the suggestion. I trained a random forest classifier and got 95% accuracy. The code has been added it to my notebook linked above.

", "rawMarkdown": "harshitsheoran thanks for the suggestion. I trained a random forest classifier and got 95% accuracy. The code has been added it to my notebook linked above.", "votes": null }, { "id": "1900969", "postDate": "08/16/2022 11:42:32", "content": "

Excellent work and great initiative in sharing your progress!
\nDid you try to train a model end2end with this and submit it? Is it improving it's performance?
\nAgain, nice work!

\n

The Devastator.

", "rawMarkdown": "Excellent work and great initiative in sharing your progress!\nDid you try to train a model end2end with this and submit it? Is it improving it's performance? \nAgain, nice work! \n\nThe Devastator.", "votes": null }, { "id": "1901595", "postDate": "08/16/2022 19:16:39", "content": "

I am in the process of doing this. Will let you how it goes.

", "rawMarkdown": "I am in the process of doing this. Will let you how it goes.", "votes": null }, { "id": "1904524", "postDate": "08/18/2022 08:52:19", "content": "

Big Data Leak! Not good result if the leak is solved, leak is in train_test_split, you are not splitting it by StudyInstanceUID, instead taking really similar values in train and validation

", "rawMarkdown": "Big Data Leak! Not good result if the leak is solved, leak is in train_test_split, you are not splitting it by StudyInstanceUID, instead taking really similar values in train and validation", "votes": null }, { "id": "1913262", "postDate": "08/25/2022 09:03:43", "content": "

Thank you for spotting this! I've fixed it now.

", "rawMarkdown": "Thank you for spotting this! I've fixed it now.", "votes": null } ], "comments": [ { "id": 1898913, "author_name": "harshitsheoran", "author_url": "", "post_date": "08/14/2022 23:57:09", "content": "

A simple enough supervised model gets like 85% accuracy (in my tests), won't making it 2.5d solve your problem with monotonicity most of the time, if that does not work, you can always just make a 3d model.

", "votes": null, "replies": [ { "id": 1899530, "author_name": "samuelcortinhas", "author_url": "", "post_date": "08/15/2022 10:27:22", "content": "

@harshitsheoran thanks for the suggestion. I trained a random forest classifier and got 95% accuracy. The code has been added it to my notebook linked above.

", "votes": null, "replies": [] } ] }, { "id": 1900969, "author_name": "thedevastator", "author_url": "", "post_date": "08/16/2022 11:42:32", "content": "

Excellent work and great initiative in sharing your progress!
\nDid you try to train a model end2end with this and submit it? Is it improving it's performance?
\nAgain, nice work!

\n

The Devastator.

", "votes": null, "replies": [ { "id": 1901595, "author_name": "samuelcortinhas", "author_url": "", "post_date": "08/16/2022 19:16:39", "content": "

I am in the process of doing this. Will let you how it goes.

", "votes": null, "replies": [] } ] }, { "id": 1904524, "author_name": "harshitsheoran", "author_url": "", "post_date": "08/18/2022 08:52:19", "content": "

Big Data Leak! Not good result if the leak is solved, leak is in train_test_split, you are not splitting it by StudyInstanceUID, instead taking really similar values in train and validation

", "votes": null, "replies": [ { "id": 1913262, "author_name": "samuelcortinhas", "author_url": "", "post_date": "08/25/2022 09:03:43", "content": "

Thank you for spotting this! I've fixed it now.

", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "1898870": "I haven't seen much progress on **finding which vertebrae is in each image** yet so I have just released my [NOTEBOOK](https://www.kaggle.com/code/samuelcortinhas/extracting-vertebrae-c1-c7). The idea is that we can identify the targets from the **unique values in the segmentation masks**. I have collected this info and stored it in this [DATASET](https://www.kaggle.com/datasets/samuelcortinhas/rsna-2022-spine-fracture-detection-metadata/code). \n\n\n\nThe **next step** would be to **build a supervised model** (using metadata/images) to predict which vertebrae is in each image for **all the other patients** in the train (& test) set which **don't have segmentation masks**. The **challenging** part is **preserving the monotonicity** (C1->C2->etc). I haven't figured this out yet so I'm sharing this resource to hopefully kickstart some progress.", "1898913": "A simple enough supervised model gets like 85% accuracy (in my tests), won't making it 2.5d solve your problem with monotonicity most of the time, if that does not work, you can always just make a 3d model.", "1899530": "harshitsheoran thanks for the suggestion. I trained a random forest classifier and got 95% accuracy. The code has been added it to my notebook linked above.", "1900969": "Excellent work and great initiative in sharing your progress!\nDid you try to train a model end2end with this and submit it? Is it improving it's performance? \nAgain, nice work! \n\nThe Devastator.", "1901595": "I am in the process of doing this. Will let you how it goes.", "1904524": "Big Data Leak! Not good result if the leak is solved, leak is in train_test_split, you are not splitting it by StudyInstanceUID, instead taking really similar values in train and validation", "1913262": "Thank you for spotting this! I've fixed it now." }, "source": "meta" }