![\"Sample](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F9833631%2F8daff7a94fc3300c7faa45b8809ece3a%2Fsample_normal.png?generation=1720276638819372&alt=media\")
Hello,
\nI just wanted to clarify if there are much more normal/mild regions than were annotated. For example, in the sample below, we can observe normal/mild level of left neural foraminal narrowing in L5/S1 given that there is little fat effacement within the bounding box region. However, we can make a similar observation about the unannotated foramina above it. Those are probably normal/mild too. If so, are the annotated normal/mild regions somehow special than the unannotated ones? Or are the annotated regions just pedagogical samples of what normal foramina are supposed to look like?
\nIn contrast, were all the annotated moderate/severe regions all that were observed to be moderate/severe? Meaning, was the best effort made to ensure that all moderate/severe regions were annotated?
\n
Hello,
\nI just wanted to clarify if there are much more normal/mild regions than were annotated. For example, in the sample below, we can observe normal/mild level of left neural foraminal narrowing in L5/S1 given that there is little fat effacement within the bounding box region. However, we can make a similar observation about the unannotated foramina above it. Those are probably normal/mild too. If so, are the annotated normal/mild regions somehow special than the unannotated ones? Or are the annotated regions just pedagogical samples of what normal foramina are supposed to look like?
\nIn contrast, were all the annotated moderate/severe regions all that were observed to be moderate/severe? Meaning, was the best effort made to ensure that all moderate/severe regions were annotated?
\n
This is just my understanding: most of series ids gets 5 annotation representing the 5 levels of lumbar spine. So there are many totally healthy annotated region. You can check for yourself by combining the label_coordinates file and the train csv file.
", "rawMarkdown": "This is just my understanding: most of series ids gets 5 annotation representing the 5 levels of lumbar spine. So there are many totally healthy annotated region. You can check for yourself by combining the label_coordinates file and the train csv file.", "votes": null } ], "comments": [ { "id": 2911152, "author_name": "llleeeoooh", "author_url": "", "post_date": "07/08/2024 06:01:34", "content": "This is just my understanding: most of series ids gets 5 annotation representing the 5 levels of lumbar spine. So there are many totally healthy annotated region. You can check for yourself by combining the label_coordinates file and the train csv file.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2908786": "Hello,\n\nI just wanted to clarify if there are much more normal/mild regions than were annotated. For example, in the sample below, we can observe normal/mild level of left neural foraminal narrowing in L5/S1 given that there is little fat effacement within the bounding box region. However, we can make a similar observation about the unannotated foramina above it. Those are probably normal/mild too. If so, are the **annotated** normal/mild regions somehow special than the **unannotated** ones? Or are the annotated regions just pedagogical samples of what normal foramina are supposed to look like?\n\nIn contrast, were all the annotated moderate/severe regions all that were observed to be moderate/severe? Meaning, was the best effort made to ensure that all moderate/severe regions were annotated?\n\n", "2911152": "This is just my understanding: most of series ids gets 5 annotation representing the 5 levels of lumbar spine. So there are many totally healthy annotated region. You can check for yourself by combining the label_coordinates file and the train csv file." }, "source": "meta" }