{ "id": 436096, "title": "total-segmentator-on-rsna-2023-abdominal-trauma", "url": "/competitions/rsna-2023-abdominal-trauma-detection/discussion/436096", "author_name": "hengck23", "post_date": "2023-09-01T03:35:52.522000", "votes": 16, "comment_count": 21, "views": 0, "content": "

There are some dataset error, please see discussion below
\ni temporary put it at google share link:
\nhttps://drive.google.com/drive/folders/11tCW2_huVJTWryfNgMFaLuP3Ll0XasNe?usp=sharing

\n
\n

It takes quite sometime to run total-segmentator.
\nHere is the results!

\n

public dataset: https://www.kaggle.com/datasets/hengck23/total-segmentator-on-rsna-2023-abdominal-trauma

\n

Apply total segmentator[1] on rsna 2023 abdominal trauma dataset[2].
\nThe command used is based on public notebook[3]

\n
!TotalSegmentator \\\n-i nputtrain_images \\\n-o temp/masks \\\n-ot  \\\n-rs spleen kidney_left kidney_right liver esophagus colon duodenum small_bowel stomach\n
\n

NOTE: there are probably error (about 5%?) in the total segmentator results. Please do a check before using this dataset!!!

", "messages": [ { "id": 2417978, "postDate": "2023-09-01T03:35:52.523Z", "content": "

There are some dataset error, please see discussion below
\ni temporary put it at google share link:
\nhttps://drive.google.com/drive/folders/11tCW2_huVJTWryfNgMFaLuP3Ll0XasNe?usp=sharing

\n
\n

It takes quite sometime to run total-segmentator.
\nHere is the results!

\n

public dataset: https://www.kaggle.com/datasets/hengck23/total-segmentator-on-rsna-2023-abdominal-trauma

\n

Apply total segmentator[1] on rsna 2023 abdominal trauma dataset[2].
\nThe command used is based on public notebook[3]

\n
!TotalSegmentator \\\n-i nputtrain_images \\\n-o temp/masks \\\n-ot  \\\n-rs spleen kidney_left kidney_right liver esophagus colon duodenum small_bowel stomach\n
\n

NOTE: there are probably error (about 5%?) in the total segmentator results. Please do a check before using this dataset!!!

", "rawMarkdown": "There are some dataset error, please see discussion below\ni temporary put it at google share link:\nhttps://drive.google.com/drive/folders/11tCW2_huVJTWryfNgMFaLuP3Ll0XasNe?usp=sharing\n\n---\n\nIt takes quite sometime to run total-segmentator.\nHere is the results!\n\npublic dataset: https://www.kaggle.com/datasets/hengck23/total-segmentator-on-rsna-2023-abdominal-trauma\n\nApply total segmentator[1] on rsna 2023 abdominal trauma dataset[2].\nThe command used is based on public notebook[3]\n\n```\n!TotalSegmentator \\\n-i /kaggle/input/rsna-2023-abdominal-trauma-detection/train_images/10104/27573 \\\n-o /kaggle/temp/masks \\\n-ot 'nifti' \\\n-rs spleen kidney_left kidney_right liver esophagus colon duodenum small_bowel stomach\n\n```\n\n# NOTE: there are probably error (about 5%?) in the total segmentator results. Please do a check before using this dataset!!! #\n", "votes": 16 }, { "id": 2418089, "postDate": "2023-09-01T06:06:57.240Z", "content": "

The nii file seems to be empty.

", "rawMarkdown": "The nii file seems to be empty.", "votes": 3, "replies": [ { "id": 2442460, "postDate": "2023-09-17T04:09:14.577Z", "content": "

I noticed that too. Why is that ?

", "rawMarkdown": "I noticed that too. Why is that ?" } ] }, { "id": 2422861, "postDate": "2023-09-04T09:47:04.413Z", "content": "

Some segmentation(patients id 6203-9983) are missing. @hengck23

", "rawMarkdown": "Some segmentation(patients id 6203-9983) are missing. @hengck23 ", "votes": 1, "replies": [ { "id": 2424072, "postDate": "2023-09-05T03:19:36.097Z", "content": "

\n

i think i did not run all due to some error in my code.

", "rawMarkdown": "~~i think it has been updated. check the file size in google and the one you have downloaded~~\n\ni think i did not run all due to some error in my code.\n", "votes": 1 } ] }, { "id": 2418125, "postDate": "2023-09-01T06:48:25.870Z", "content": "

i temporary put it at google share link:
\nhttps://drive.google.com/drive/folders/11tCW2_huVJTWryfNgMFaLuP3Ll0XasNe?usp=sharing

", "rawMarkdown": "i temporary put it at google share link: \nhttps://drive.google.com/drive/folders/11tCW2_huVJTWryfNgMFaLuP3Ll0XasNe?usp=sharing", "votes": 1 }, { "id": 2418116, "postDate": "2023-09-01T06:41:06.327Z", "content": "

i think it is this issue:
\nhttps://www.kaggle.com/discussions/questions-and-answers/336968
\nhttps://stackoverflow.com/questions/76737983/upload-nii-files-to-kaggle

\n

my original files are in nii.gz format

\n

anyone has a solution?

", "rawMarkdown": "i think it is this issue:\nhttps://www.kaggle.com/discussions/questions-and-answers/336968\nhttps://stackoverflow.com/questions/76737983/upload-nii-files-to-kaggle\n\nmy original files are in nii.gz format\n\nanyone has a solution?", "votes": 1, "replies": [ { "id": 2418587, "postDate": "2023-09-01T11:52:50.497Z", "content": "

Hi,

\n

You could add a password to the compressed file and extract it in the notebook manually, 7za and zip are installed in the Kaggle notebooks, so with zip for example:

\n

\"\"

\n

Here is another paper/model related to TotalSegmentator in case you want to create another dataset.

\n

https://github.com/Ziyan-Huang/STU-Net

\n

Edit:

\n

Note the capital P in unzip.

", "rawMarkdown": "Hi,\n\nYou could add a password to the compressed file and extract it in the notebook manually, 7za and zip are installed in the Kaggle notebooks, so with zip for example:\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F14573416%2Fff2528f638b4e843cb9a3b6a1c576fcd%2FScreenshot%20from%202023-09-01%2013-49-00.png?generation=1693568962025002&alt=media)\n\nHere is another paper/model related to TotalSegmentator in case you want to create another dataset.\n\n[https://github.com/Ziyan-Huang/STU-Net](url)\n\nEdit:\n\nNote the capital P in unzip.", "votes": 1, "replies": [ { "id": 2432365, "postDate": "2023-09-10T19:09:26.140Z", "content": "

Thank you))

", "rawMarkdown": "Thank you))" } ] }, { "id": 2424042, "postDate": "2023-09-05T02:45:12.987Z", "content": "

rename and remove .gz suffix

", "rawMarkdown": "rename and remove `.gz` suffix", "votes": 1, "replies": [ { "id": 2424256, "postDate": "2023-09-05T06:51:18.507Z", "content": "

@hengck23 hey I was able to get to upload correctly, I downloaded your files from the drive link. I renamed all the files by removing the .gz suffix. This suffix means that data is compressed so Kaggle tries to decompress (im guessing using gunzip) but fails and creates an empty directory.

\n

I will probably delete this dataset soon; I am creating my own using the same tool and I don't want to take credit for your work.

", "rawMarkdown": "@hengck23 hey I was able to get to [upload correctly](https://www.kaggle.com/datasets/outwrest/rsna-segmentations), I downloaded your files from the drive link. I renamed all the files by removing the `.gz` suffix. This suffix means that data is compressed so Kaggle tries to decompress (im guessing using `gunzip`) but fails and creates an empty directory.\n\nI will probably delete this dataset soon; I am creating my own using the same tool and I don't want to take credit for your work. ", "votes": 1 } ] }, { "id": 2432200, "postDate": "2023-09-10T16:53:14.793Z", "rawMarkdown": "", "isDeleted": true } ] }, { "id": 2430698, "postDate": "2023-09-09T13:56:10.423Z", "content": "

Did you test how long it will take to run segmentator on test set?

", "rawMarkdown": "Did you test how long it will take to run segmentator on test set?", "replies": [ { "id": 2430941, "postDate": "2023-09-09T17:17:48.663Z", "content": "

no. i don't think you can run it with 9hr.
\nand i don't think fine segmentation is required.

", "rawMarkdown": "no. i don't think you can run it with 9hr.\nand i don't think fine segmentation is required.", "replies": [ { "id": 2430955, "postDate": "2023-09-09T17:24:15.927Z", "content": "

Then we can't train the model that use segmentation as input？Cuz we don't have segmentation of test images

", "rawMarkdown": "Then we can't train the model that use segmentation as input？Cuz we don't have segmentation of test images" }, { "id": 2431117, "postDate": "2023-09-09T19:41:29.780Z", "content": "

you still can use it in training, e.g. to predict bounding box, bounding section or regularization in aux loss.
\nor you can simplify the segmentation since maybe only coarse segmentation is required

", "rawMarkdown": "you still can use it in training, e.g. to predict bounding box, bounding section or regularization in aux loss.\nor you can simplify the segmentation since maybe only coarse segmentation is required", "votes": 1 }, { "id": 2442332, "postDate": "2023-09-16T22:34:35.847Z", "content": "

I just read this disscusion , and im not sure that i get what you say , if we cant do the segmantion during the infernce so how it will be helpful to use it during training, oue model would not overfit for examples that are segmanted ?

", "rawMarkdown": "I just read this disscusion , and im not sure that i get what you say , if we cant do the segmantion during the infernce so how it will be helpful to use it during training, oue model would not overfit for examples that are segmanted ? " } ] }, { "id": 2432185, "postDate": "2023-09-10T16:45:26.270Z", "content": "

Hello
\nFor one image with shape (512, 512, 803), segmentation - spleen, kidney_left, kidney_right, liver.

\n

LowRes mask 90.07948136329651 sec
\nHighRes mask 137.6063380241394 sec

", "rawMarkdown": "Hello \nFor one image with shape (512, 512, 803), segmentation - spleen, kidney_left, kidney_right, liver.\n\nLowRes mask 90.07948136329651 sec\nHighRes mask 137.6063380241394 sec" } ] }, { "id": 2449237, "postDate": "2023-09-21T04:51:17.683Z", "rawMarkdown": "", "isDeleted": true, "replies": [ { "id": 2449256, "postDate": "2023-09-21T05:09:38.043Z", "content": "

NOTE: there are probably error (about 5%?) in the total segmentator results. Please do a check before using this dataset!!!

", "rawMarkdown": "# NOTE: there are probably error (about 5%?) in the total segmentator results. Please do a check before using this dataset!!!", "votes": 1 } ] }, { "id": 2418050, "postDate": "2023-09-01T05:06:51.130Z", "content": "

Thanks a lot for sharing @hengck23

", "rawMarkdown": "Thanks a lot for sharing @hengck23 ", "votes": 1 }, { "id": 2432191, "postDate": "2023-09-10T16:49:56.677Z", "content": "

Thank you so much

", "rawMarkdown": "Thank you so much" } ], "comments": [ { "id": 2418089, "author_name": "fate", "author_url": "", "post_date": "2023-09-01T06:06:57.240000", "content": "

The nii file seems to be empty.

", "votes": 3, "replies": [ { "id": 2442460, "author_name": "Franklin Shih0617", "author_url": "", "post_date": "2023-09-17T04:09:14.577000", "content": "

I noticed that too. Why is that ?

", "votes": 0, "replies": [] } ] }, { "id": 2422861, "author_name": "fate", "author_url": "", "post_date": "2023-09-04T09:47:04.413000", "content": "

Some segmentation(patients id 6203-9983) are missing. @hengck23

", "votes": 1, "replies": [ { "id": 2424072, "author_name": "hengck23", "author_url": "", "post_date": "2023-09-05T03:19:36.097000", "content": "

\n

i think i did not run all due to some error in my code.

", "votes": 1, "replies": [] } ] }, { "id": 2418125, "author_name": "hengck23", "author_url": "", "post_date": "2023-09-01T06:48:25.870000", "content": "

i temporary put it at google share link:
\nhttps://drive.google.com/drive/folders/11tCW2_huVJTWryfNgMFaLuP3Ll0XasNe?usp=sharing

", "votes": 1, "replies": [] }, { "id": 2418116, "author_name": "hengck23", "author_url": "", "post_date": "2023-09-01T06:41:06.327000", "content": "

i think it is this issue:
\nhttps://www.kaggle.com/discussions/questions-and-answers/336968
\nhttps://stackoverflow.com/questions/76737983/upload-nii-files-to-kaggle

\n

my original files are in nii.gz format

\n

anyone has a solution?

", "votes": 1, "replies": [ { "id": 2418587, "author_name": "Antonio Félix", "author_url": "", "post_date": "2023-09-01T11:52:50.497000", "content": "

Hi,

\n

You could add a password to the compressed file and extract it in the notebook manually, 7za and zip are installed in the Kaggle notebooks, so with zip for example:

\n

\"\"

\n

Here is another paper/model related to TotalSegmentator in case you want to create another dataset.

\n

https://github.com/Ziyan-Huang/STU-Net

\n

Edit:

\n

Note the capital P in unzip.

", "votes": 1, "replies": [ { "id": 2432365, "author_name": "Max Semakov", "author_url": "", "post_date": "2023-09-10T19:09:26.140000", "content": "

Thank you))

", "votes": 0, "replies": [] } ] }, { "id": 2424042, "author_name": "outwrest", "author_url": "", "post_date": "2023-09-05T02:45:12.987000", "content": "

rename and remove .gz suffix

", "votes": 1, "replies": [ { "id": 2424256, "author_name": "outwrest", "author_url": "", "post_date": "2023-09-05T06:51:18.507000", "content": "

@hengck23 hey I was able to get to upload correctly, I downloaded your files from the drive link. I renamed all the files by removing the .gz suffix. This suffix means that data is compressed so Kaggle tries to decompress (im guessing using gunzip) but fails and creates an empty directory.

\n

I will probably delete this dataset soon; I am creating my own using the same tool and I don't want to take credit for your work.

", "votes": 1, "replies": [] } ] }, { "id": 2432200, "author_name": "", "author_url": "", "post_date": "2023-09-10T16:53:14.793000", "content": "", "votes": 0, "replies": [] } ] }, { "id": 2430698, "author_name": "NorthM344", "author_url": "", "post_date": "2023-09-09T13:56:10.423000", "content": "

Did you test how long it will take to run segmentator on test set?

", "votes": 0, "replies": [ { "id": 2430941, "author_name": "hengck23", "author_url": "", "post_date": "2023-09-09T17:17:48.663000", "content": "

no. i don't think you can run it with 9hr.
\nand i don't think fine segmentation is required.

", "votes": 0, "replies": [ { "id": 2430955, "author_name": "NorthM344", "author_url": "", "post_date": "2023-09-09T17:24:15.927000", "content": "

Then we can't train the model that use segmentation as input？Cuz we don't have segmentation of test images

", "votes": 0, "replies": [] }, { "id": 2431117, "author_name": "hengck23", "author_url": "", "post_date": "2023-09-09T19:41:29.780000", "content": "

you still can use it in training, e.g. to predict bounding box, bounding section or regularization in aux loss.
\nor you can simplify the segmentation since maybe only coarse segmentation is required

", "votes": 1, "replies": [] }, { "id": 2442332, "author_name": "SneakyWave12", "author_url": "", "post_date": "2023-09-16T22:34:35.847000", "content": "

I just read this disscusion , and im not sure that i get what you say , if we cant do the segmantion during the infernce so how it will be helpful to use it during training, oue model would not overfit for examples that are segmanted ?

", "votes": 0, "replies": [] } ] }, { "id": 2432185, "author_name": "Max Semakov", "author_url": "", "post_date": "2023-09-10T16:45:26.270000", "content": "

Hello
\nFor one image with shape (512, 512, 803), segmentation - spleen, kidney_left, kidney_right, liver.

\n

LowRes mask 90.07948136329651 sec
\nHighRes mask 137.6063380241394 sec

", "votes": 0, "replies": [] } ] }, { "id": 2449237, "author_name": "", "author_url": "", "post_date": "2023-09-21T04:51:17.683000", "content": "", "votes": 0, "replies": [ { "id": 2449256, "author_name": "hengck23", "author_url": "", "post_date": "2023-09-21T05:09:38.043000", "content": "

NOTE: there are probably error (about 5%?) in the total segmentator results. Please do a check before using this dataset!!!

", "votes": 1, "replies": [] } ] }, { "id": 2418050, "author_name": "Rishabh Jain", "author_url": "", "post_date": "2023-09-01T05:06:51.130000", "content": "

Thanks a lot for sharing @hengck23

", "votes": 1, "replies": [] }, { "id": 2432191, "author_name": "Max Semakov", "author_url": "", "post_date": "2023-09-10T16:49:56.677000", "content": "

Thank you so much

", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "2417978": "There are some dataset error, please see discussion below\ni temporary put it at google share link:\nhttps://drive.google.com/drive/folders/11tCW2_huVJTWryfNgMFaLuP3Ll0XasNe?usp=sharing\n\n---\n\nIt takes quite sometime to run total-segmentator.\nHere is the results!\n\npublic dataset: https://www.kaggle.com/datasets/hengck23/total-segmentator-on-rsna-2023-abdominal-trauma\n\nApply total segmentator[1] on rsna 2023 abdominal trauma dataset[2].\nThe command used is based on public notebook[3]\n\n```\n!TotalSegmentator \\\n-i /kaggle/input/rsna-2023-abdominal-trauma-detection/train_images/10104/27573 \\\n-o /kaggle/temp/masks \\\n-ot 'nifti' \\\n-rs spleen kidney_left kidney_right liver esophagus colon duodenum small_bowel stomach\n\n```\n\n# NOTE: there are probably error (about 5%?) in the total segmentator results. Please do a check before using this dataset!!! #\n", "2418089": "The nii file seems to be empty.", "2422861": "Some segmentation(patients id 6203-9983) are missing. @hengck23 ", "2418125": "i temporary put it at google share link: \nhttps://drive.google.com/drive/folders/11tCW2_huVJTWryfNgMFaLuP3Ll0XasNe?usp=sharing", "2418116": "i think it is this issue:\nhttps://www.kaggle.com/discussions/questions-and-answers/336968\nhttps://stackoverflow.com/questions/76737983/upload-nii-files-to-kaggle\n\nmy original files are in nii.gz format\n\nanyone has a solution?", "2430698": "Did you test how long it will take to run segmentator on test set?", "2449237": "", "2418050": "Thanks a lot for sharing @hengck23 ", "2432191": "Thank you so much" } }