I don‘t have any medical background, and I feel a little confused over the presented data. My apprehension about the images is that patients take CTs at different times (series folders). However, what is the relationship between images in the same series folder? I think this is critical for us to decide how to use these images. Thank you if you are willing to share!
", "messages": [ { "id": 2414409, "postDate": "2023-08-29T16:02:37.293Z", "content": "Meta data in the dicom files should have date and time - think I have seen a shared notebook or discussion where the code to look at that data was given.
\nMost likely all the multiple sets of folders are for CT scans where the patient has dye injected. The pinned discussion topic covers this pretty well.
", "rawMarkdown": "Meta data in the dicom files should have date and time - think I have seen a shared notebook or discussion where the code to look at that data was given.\n\nMost likely all the multiple sets of folders are for CT scans where the patient has dye injected. The [pinned discussion topic](https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection/discussion/427226) covers this pretty well.\n\n", "replies": [ { "id": 2414737, "postDate": "2023-08-29T23:45:24.140Z", "content": "Thank you, but I think this post https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection/discussion/427226 is not able to solve my problem. This post just tells me that images from different series folders are taken in different phases, but what about the images in the same series folder? Why the images from the same series folder are different and why do we need these different images?
", "rawMarkdown": "Thank you, but I think this post [https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection/discussion/427226](url) is not able to solve my problem. This post just tells me that images from different series folders are taken in different phases, but what about the images in the same series folder? Why the images from the same series folder are different and why do we need these different images?", "replies": [ { "id": 2415082, "postDate": "2023-08-30T05:25:33.647Z", "content": "OK - I am confused on what your question is….
\nAs example - train_images has folder for patient 19, which has sub folder 14374.
\nDoes your question deal will images 229.dcm, 230.dcm, etc to 901.dcm ?
", "rawMarkdown": "OK - I am confused on what your question is....\n\nAs example - train_images has folder for patient 19, which has sub folder 14374.\n\nDoes your question deal will images 229.dcm, 230.dcm, etc to 901.dcm ?" }, { "id": 2415483, "postDate": "2023-08-30T11:08:15.827Z", "content": "Yes. That's right! According to the data description, these are the images of different instance IDs. So, my question is what is the meaning of the difference between these images?
", "rawMarkdown": "Yes. That's right! According to the data description, these are the images of different instance IDs. So, my question is what is the meaning of the difference between these images?" }, { "id": 2415730, "postDate": "2023-08-30T15:32:15.233Z", "content": "Some Medical context if you’re interested
\n\n\nThree discussions that can get towards answering your question. The difference between the images located in the same folder is the position of the slice of the human patients body as axial slices. In that view there will not be a patient who's spleen can be seen in the same image as their kidneys. Some of the shared notebooks use every 10th image to represent the patient (stride = 10) entire body (reduce the size of the data). I have decent compute resources so using every slice in my model training.
\nThe supplied test is very misleading with only a single image, but it's a simple place holder for the real test set (public/private) that we never see.
", "rawMarkdown": "[Some Medical context if you’re interested ](https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection/discussion/432900)\n\n[Understanding how Patient IDs, Series IDs, and Instance Numbers are related together in CT Scan Datasets](https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection/discussion/431475)\n\n[CT Scans and Contrast Phases](https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection/discussion/427226)\n\nThree discussions that can get towards answering your question. The difference between the images located in the same folder is the position of the slice of the human patients body as axial slices. In that view there will not be a patient who's spleen can be seen in the same image as their kidneys. Some of the shared notebooks use every 10th image to represent the patient (stride = 10) entire body (reduce the size of the data). I have decent compute resources so using every slice in my model training.\n\nThe supplied test is very misleading with only a single image, but it's a simple place holder for the real test set (public/private) that we never see.\n\n" } ] } ] }, { "id": 2413478, "postDate": "2023-08-29T00:26:36.297Z", "content": "I don‘t have any medical background, and I feel a little confused over the presented data. My apprehension about the images is that patients take CTs at different times (series folders). However, what is the relationship between images in the same series folder? I think this is critical for us to decide how to use these images. Thank you if you are willing to share!
", "rawMarkdown": "I don‘t have any medical background, and I feel a little confused over the presented data. My apprehension about the images is that patients take CTs at different times (series folders). However, what is the relationship between images in the same series folder? I think this is critical for us to decide how to use these images. Thank you if you are willing to share!" } ], "comments": [ { "id": 2414409, "author_name": "PC Jimmmy", "author_url": "", "post_date": "2023-08-29T16:02:37.293000", "content": "Meta data in the dicom files should have date and time - think I have seen a shared notebook or discussion where the code to look at that data was given.
\nMost likely all the multiple sets of folders are for CT scans where the patient has dye injected. The pinned discussion topic covers this pretty well.
", "votes": 0, "replies": [ { "id": 2414737, "author_name": "Trunway", "author_url": "", "post_date": "2023-08-29T23:45:24.140000", "content": "Thank you, but I think this post https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection/discussion/427226 is not able to solve my problem. This post just tells me that images from different series folders are taken in different phases, but what about the images in the same series folder? Why the images from the same series folder are different and why do we need these different images?
", "votes": 0, "replies": [ { "id": 2415082, "author_name": "PC Jimmmy", "author_url": "", "post_date": "2023-08-30T05:25:33.647000", "content": "OK - I am confused on what your question is….
\nAs example - train_images has folder for patient 19, which has sub folder 14374.
\nDoes your question deal will images 229.dcm, 230.dcm, etc to 901.dcm ?
", "votes": 0, "replies": [] }, { "id": 2415483, "author_name": "Trunway", "author_url": "", "post_date": "2023-08-30T11:08:15.827000", "content": "Yes. That's right! According to the data description, these are the images of different instance IDs. So, my question is what is the meaning of the difference between these images?
", "votes": 0, "replies": [] }, { "id": 2415730, "author_name": "PC Jimmmy", "author_url": "", "post_date": "2023-08-30T15:32:15.233000", "content": "Some Medical context if you’re interested
\n\n\nThree discussions that can get towards answering your question. The difference between the images located in the same folder is the position of the slice of the human patients body as axial slices. In that view there will not be a patient who's spleen can be seen in the same image as their kidneys. Some of the shared notebooks use every 10th image to represent the patient (stride = 10) entire body (reduce the size of the data). I have decent compute resources so using every slice in my model training.
\nThe supplied test is very misleading with only a single image, but it's a simple place holder for the real test set (public/private) that we never see.
", "votes": 0, "replies": [] } ] } ] } ], "raw_markdown_by_id": { "2414409": "Meta data in the dicom files should have date and time - think I have seen a shared notebook or discussion where the code to look at that data was given.\n\nMost likely all the multiple sets of folders are for CT scans where the patient has dye injected. The [pinned discussion topic](https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection/discussion/427226) covers this pretty well.\n\n", "2413478": "I don‘t have any medical background, and I feel a little confused over the presented data. My apprehension about the images is that patients take CTs at different times (series folders). However, what is the relationship between images in the same series folder? I think this is critical for us to decide how to use these images. Thank you if you are willing to share!" } }