I uploaded decompressed dicom images and a notebook created from the official dataset with python3.6 and gdcm.
\nI removed only one file: ID00052637202186188008618/4.dcm, GDCM and pydicom failed in its decompression because its pixel array was collapsed to be read as jpeg.
I uploaded decompressed dicom images and a notebook created from the official dataset with python3.6 and gdcm.
\nI removed only one file: ID00052637202186188008618/4.dcm, GDCM and pydicom failed in its decompression because its pixel array was collapsed to be read as jpeg.
Use simpleITK.
", "rawMarkdown": "Use simpleITK." }, { "id": 1002753, "postDate": "2020-09-08T11:59:41.370Z", "content": "Why do you want to read them without GDCM?
", "rawMarkdown": "Why do you want to read them without GDCM?", "replies": [ { "id": 1003808, "postDate": "2020-09-09T09:08:03.120Z", "content": "I intended to make this an answer for this thread.
\nWithout the Internet, we have to read them without GDCM, don't we? or Plz tell me if there is a good way to install GDCM without the Internet.
\nI think Vtk can be a good alternative choice, but I'm more familiar with pydicom.
I didn't use Vtk, so I can't tell you about that one.
\nYou can't install it without internet. But you need GDCM only for the training. You won't need for predictions. So you can use internet in a notebook for training and use another one for the prediction.
There is this comment from a competition host in the thread you linked.
\n\n", "rawMarkdown": "I didn't use Vtk, so I can't tell you about that one. \nYou can't install it without internet. But you need GDCM only for the training. You won't need for predictions. So you can use internet in a notebook for training and use another one for the prediction.\n\nThere is this comment from a competition host in the thread you linked.\n> One solution could be to install the package, train a model, then upload the pretrained model. Another one could be to save the scans that are not readable without GDCM (two scans) in npy files, and then load them using np load." } ] }, { "id": 1003804, "postDate": "2020-09-09T09:06:52.060Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 1005026, "author_name": "bz6102365", "author_url": "", "post_date": "2020-09-10T08:05:13.923000", "content": "One solution could be to install the package, train a model, then upload the pretrained model. Another one could be to save the scans that are not readable without GDCM (two scans) in npy files, and then load them using np load.
\n
Use simpleITK.
", "votes": 0, "replies": [] }, { "id": 1002753, "author_name": "Yohann Wattiez", "author_url": "", "post_date": "2020-09-08T11:59:41.370000", "content": "Why do you want to read them without GDCM?
", "votes": 0, "replies": [ { "id": 1003808, "author_name": "HiroJam", "author_url": "", "post_date": "2020-09-09T09:08:03.120000", "content": "I intended to make this an answer for this thread.
\nWithout the Internet, we have to read them without GDCM, don't we? or Plz tell me if there is a good way to install GDCM without the Internet.
\nI think Vtk can be a good alternative choice, but I'm more familiar with pydicom.
I didn't use Vtk, so I can't tell you about that one.
\nYou can't install it without internet. But you need GDCM only for the training. You won't need for predictions. So you can use internet in a notebook for training and use another one for the prediction.
There is this comment from a competition host in the thread you linked.
\n\n", "votes": 0, "replies": [] } ] }, { "id": 1003804, "author_name": "", "author_url": "", "post_date": "2020-09-09T09:06:52.060000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "1002650": "I uploaded [decompressed dicom images](https://www.kaggle.com/hirojam/osic-pulmonary-fibrosis-progression-decompressed) and [a notebook](https://www.kaggle.com/hirojam/use-of-decompressed-images) created from the official dataset with python3.6 and gdcm.\nI removed only one file: ID00052637202186188008618/4.dcm, GDCM and pydicom failed in its decompression because its pixel array was collapsed to be read as jpeg.", "1005026": "Use simpleITK.", "1002753": "Why do you want to read them without GDCM?", "1003804": "" } }One solution could be to install the package, train a model, then upload the pretrained model. Another one could be to save the scans that are not readable without GDCM (two scans) in npy files, and then load them using np load.
\n