this was a bit confusing to me so i want to share with you my understanding of it, it might be your problem as well:
\npatient ID: a specific number given to each patient.
\nSeries ID: each series corresponds to one CT scan. each CT scan contain a set of related images. These images are typically taken in sequence as the scanner moves through the body, creating a series of cross-sectional slices. so a series ID is a specific and unique number assigned to each individual CT scan.
\nInstance Number: as i said each CT scan contain series of related images, Instance Number is a unique number for each image in each series. Just as a book has pages numbered in order, these numbers indicate the position of each image within a series.
\n Patient\n (ID: ABC123)\n |||\n Series Series\n (ID: 001) (ID: 002)\n ||\n ||\n Instance Instance\n (Num: 001) (Num: 001)\n ||\n ||\n Image1 Image1\nIn this diagram:
\n.There is a patient with unique patient ID (ABC123).
\n .this patient has undergone multiple CT scans, which are represented by different series IDs (001 and 002).
\n .Within each CT scan series, there are individual images represented by instance numbers (001, 002, …).
\n .These images are the actual CT scan images taken during the procedure.
so, patient ABC123 has had two CT scan procedures with series IDs 001 and 002. Within the first CT scan series, there are images with instance numbers 001, 002, and so on.
\nsummery:
\n.A patient has one unique patient ID.
\n .A patient can have multiple series IDs, each corresponding to a different medical procedure or examination.
\n .Each series ID contains multiple instance numbers, each representing an individual image within that series.
Hope this will be useful :)
", "messages": [ { "id": 2388914, "postDate": "2023-08-13T17:16:07.310Z", "content": "this was a bit confusing to me so i want to share with you my understanding of it, it might be your problem as well:
\npatient ID: a specific number given to each patient.
\nSeries ID: each series corresponds to one CT scan. each CT scan contain a set of related images. These images are typically taken in sequence as the scanner moves through the body, creating a series of cross-sectional slices. so a series ID is a specific and unique number assigned to each individual CT scan.
\nInstance Number: as i said each CT scan contain series of related images, Instance Number is a unique number for each image in each series. Just as a book has pages numbered in order, these numbers indicate the position of each image within a series.
\n Patient\n (ID: ABC123)\n |||\n Series Series\n (ID: 001) (ID: 002)\n ||\n ||\n Instance Instance\n (Num: 001) (Num: 001)\n ||\n ||\n Image1 Image1\nIn this diagram:
\n.There is a patient with unique patient ID (ABC123).
\n .this patient has undergone multiple CT scans, which are represented by different series IDs (001 and 002).
\n .Within each CT scan series, there are individual images represented by instance numbers (001, 002, …).
\n .These images are the actual CT scan images taken during the procedure.
so, patient ABC123 has had two CT scan procedures with series IDs 001 and 002. Within the first CT scan series, there are images with instance numbers 001, 002, and so on.
\nsummery:
\n.A patient has one unique patient ID.
\n .A patient can have multiple series IDs, each corresponding to a different medical procedure or examination.
\n .Each series ID contains multiple instance numbers, each representing an individual image within that series.
Hope this will be useful :)
", "rawMarkdown": "this was a bit confusing to me so i want to share with you my understanding of it, it might be your problem as well:\n\npatient ID: a specific number given to each patient.\n\nSeries ID: each series corresponds to one CT scan. each CT scan contain a set of related images. These images are typically taken in sequence as the scanner moves through the body, creating a series of cross-sectional slices. so a series ID is a specific and unique number assigned to each individual CT scan.\n\nInstance Number: as i said each CT scan contain series of related images, Instance Number is a unique number for each image in each series. Just as a book has pages numbered in order, these numbers indicate the position of each image within a series. \n\n\n Patient\n (ID: ABC123)\n |\n +---+---+\n | |\n Series Series\n (ID: 001) (ID: 002)\n | |\n | |\n Instance Instance\n (Num: 001) (Num: 001)\n | |\n | |\n Image1 Image1\n\n \n\nIn this diagram:\n\n .There is a patient with unique patient ID (ABC123).\n .this patient has undergone multiple CT scans, which are represented by different series IDs (001 and 002).\n .Within each CT scan series, there are individual images represented by instance numbers (001, 002, ...).\n .These images are the actual CT scan images taken during the procedure.\n\n\nso, patient ABC123 has had two CT scan procedures with series IDs 001 and 002. Within the first CT scan series, there are images with instance numbers 001, 002, and so on.\n\nsummery:\n\n .A patient has one unique patient ID.\n .A patient can have multiple series IDs, each corresponding to a different medical procedure or examination.\n .Each series ID contains multiple instance numbers, each representing an individual image within that series.\n\n\n\nHope this will be useful :)\n", "votes": 8 }, { "id": 2390649, "postDate": "2023-08-14T17:05:49.297Z", "content": "Thanks for the clarification it is very useful but I have a question:
\nare we supposed to use the whole series in training? since the part of interest can be represented in small group of instances?
I think it is very necessary to first select the instance of interest, maybe crop the organ from instances is useful too. Introducing a prior will make the model perform better. In order to do this, see discussion, and use TotalSegmentor in it will be a good choice.
", "rawMarkdown": "I think it is very necessary to first select the instance of interest, maybe crop the organ from instances is useful too. Introducing a prior will make the model perform better. In order to do this, see [discussion](https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection/discussion/), and use TotalSegmentor in it will be a good choice." } ] } ], "comments": [ { "id": 2390649, "author_name": "Areej Malkawi", "author_url": "", "post_date": "2023-08-14T17:05:49.297000", "content": "Thanks for the clarification it is very useful but I have a question:
\nare we supposed to use the whole series in training? since the part of interest can be represented in small group of instances?
I think it is very necessary to first select the instance of interest, maybe crop the organ from instances is useful too. Introducing a prior will make the model perform better. In order to do this, see discussion, and use TotalSegmentor in it will be a good choice.
", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "2388914": "this was a bit confusing to me so i want to share with you my understanding of it, it might be your problem as well:\n\npatient ID: a specific number given to each patient.\n\nSeries ID: each series corresponds to one CT scan. each CT scan contain a set of related images. These images are typically taken in sequence as the scanner moves through the body, creating a series of cross-sectional slices. so a series ID is a specific and unique number assigned to each individual CT scan.\n\nInstance Number: as i said each CT scan contain series of related images, Instance Number is a unique number for each image in each series. Just as a book has pages numbered in order, these numbers indicate the position of each image within a series. \n\n\n Patient\n (ID: ABC123)\n |\n +---+---+\n | |\n Series Series\n (ID: 001) (ID: 002)\n | |\n | |\n Instance Instance\n (Num: 001) (Num: 001)\n | |\n | |\n Image1 Image1\n\n \n\nIn this diagram:\n\n .There is a patient with unique patient ID (ABC123).\n .this patient has undergone multiple CT scans, which are represented by different series IDs (001 and 002).\n .Within each CT scan series, there are individual images represented by instance numbers (001, 002, ...).\n .These images are the actual CT scan images taken during the procedure.\n\n\nso, patient ABC123 has had two CT scan procedures with series IDs 001 and 002. Within the first CT scan series, there are images with instance numbers 001, 002, and so on.\n\nsummery:\n\n .A patient has one unique patient ID.\n .A patient can have multiple series IDs, each corresponding to a different medical procedure or examination.\n .Each series ID contains multiple instance numbers, each representing an individual image within that series.\n\n\n\nHope this will be useful :)\n", "2390649": "Thanks for the clarification it is very useful but I have a question:\nare we supposed to use the whole series in training? since the part of interest can be represented in small group of instances? " } }