I think it would be easier to create a multi-input model by using dicom metadata to associate the xyz coordinates of the two axes, but I don't know how to handle the metadata.
\nIf anyone has any ideas please let me know
I think it would be easier to create a multi-input model by using dicom metadata to associate the xyz coordinates of the two axes, but I don't know how to handle the metadata.
\nIf anyone has any ideas please let me know
I am not sure if this will help in associating the xyz coordinates of the two axes, but I am able to group slices inside of a series folder such that I am able to create a volume and an affine matrix using combine_slices function from dicom-numpy package.
This affine can be used to get the orientation, origin and spacing of the volume generated. Will that help your case? If so, I can make my draft notebook publicly available (the slice grouping code is too long to fit inside a single comment).
", "rawMarkdown": "I am not sure if this will help in associating the xyz coordinates of the two axes, but I am able to group slices inside of a series folder such that I am able to create a volume and an affine matrix using `combine_slices` function from `dicom-numpy` package.\n\nThis affine can be used to get the orientation, origin and spacing of the volume generated. Will that help your case? If so, I can make my draft notebook publicly available (the slice grouping code is too long to fit inside a single comment).", "votes": null }, { "id": "2858826", "postDate": "06/06/2024 16:49:56", "content": "I made a notebook showing how to cross-reference sagittal and axial images using ImagePositionPatient in the DICOM metadata:
However, this doesn't seem to be totally accurate in my experience when mapping axial to sagittal images (the example in my notebook shows a gap which I do not think is correct).
", "rawMarkdown": "I made a notebook showing how to cross-reference sagittal and axial images using `ImagePositionPatient` in the DICOM metadata: \n\nhttps://www.kaggle.com/code/vaillant/cross-reference-images-in-different-mri-planes?scriptVersionId=181874384\n\nHowever, this doesn't seem to be totally accurate in my experience when mapping axial to sagittal images (the example in my notebook shows a gap which I do not think is correct).", "votes": null }, { "id": "2858866", "postDate": "06/06/2024 17:26:09", "content": "Is it possible to align them because scans of different orientation are taken at different time with different posture?
", "rawMarkdown": "Is it possible to align them because scans of different orientation are taken at different time with different posture?", "votes": null }, { "id": "2862087", "postDate": "06/08/2024 14:48:49", "content": "Thanks for sharing! It is true that some axial images are mapped to areas other than L1~S1, and it seems impossible to make the correspondence as is.
", "rawMarkdown": "Thanks for sharing! It is true that some axial images are mapped to areas other than L1~S1, and it seems impossible to make the correspondence as is.", "votes": null }, { "id": "2866449", "postDate": "06/11/2024 10:53:17", "content": "if you could make this available that would be fantastic
", "rawMarkdown": "if you could make this available that would be fantastic", "votes": null } ], "comments": [ { "id": 2855797, "author_name": "coderrkj", "author_url": "", "post_date": "06/05/2024 03:32:50", "content": "I am not sure if this will help in associating the xyz coordinates of the two axes, but I am able to group slices inside of a series folder such that I am able to create a volume and an affine matrix using combine_slices function from dicom-numpy package.
This affine can be used to get the orientation, origin and spacing of the volume generated. Will that help your case? If so, I can make my draft notebook publicly available (the slice grouping code is too long to fit inside a single comment).
", "votes": null, "replies": [ { "id": 2866449, "author_name": "homiecal", "author_url": "", "post_date": "06/11/2024 10:53:17", "content": "if you could make this available that would be fantastic
", "votes": null, "replies": [] } ] }, { "id": 2858826, "author_name": "vaillant", "author_url": "", "post_date": "06/06/2024 16:49:56", "content": "I made a notebook showing how to cross-reference sagittal and axial images using ImagePositionPatient in the DICOM metadata:
However, this doesn't seem to be totally accurate in my experience when mapping axial to sagittal images (the example in my notebook shows a gap which I do not think is correct).
", "votes": null, "replies": [ { "id": 2858866, "author_name": "manaidu", "author_url": "", "post_date": "06/06/2024 17:26:09", "content": "Is it possible to align them because scans of different orientation are taken at different time with different posture?
", "votes": null, "replies": [] }, { "id": 2862087, "author_name": "abebe9849", "author_url": "", "post_date": "06/08/2024 14:48:49", "content": "Thanks for sharing! It is true that some axial images are mapped to areas other than L1~S1, and it seems impossible to make the correspondence as is.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "2855487": "I think it would be easier to create a multi-input model by using dicom metadata to associate the xyz coordinates of the two axes, but I don't know how to handle the metadata. \nIf anyone has any ideas please let me know", "2855797": "I am not sure if this will help in associating the xyz coordinates of the two axes, but I am able to group slices inside of a series folder such that I am able to create a volume and an affine matrix using `combine_slices` function from `dicom-numpy` package.\n\nThis affine can be used to get the orientation, origin and spacing of the volume generated. Will that help your case? If so, I can make my draft notebook publicly available (the slice grouping code is too long to fit inside a single comment).", "2858826": "I made a notebook showing how to cross-reference sagittal and axial images using `ImagePositionPatient` in the DICOM metadata: \n\nhttps://www.kaggle.com/code/vaillant/cross-reference-images-in-different-mri-planes?scriptVersionId=181874384\n\nHowever, this doesn't seem to be totally accurate in my experience when mapping axial to sagittal images (the example in my notebook shows a gap which I do not think is correct).", "2858866": "Is it possible to align them because scans of different orientation are taken at different time with different posture?", "2862087": "Thanks for sharing! It is true that some axial images are mapped to areas other than L1~S1, and it seems impossible to make the correspondence as is.", "2866449": "if you could make this available that would be fantastic" }, "source": "meta" }