![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F9533096%2Fe6119d40589812fe1ac8b6d1ae6b20ea%2FExample.JPG?generation=1663423633788328&alt=media\")
As shown in the below image the upper half seems like the healthy one and the below one contains some clots so, to train a model which can identify clots (of the given two categories), aren't we supposed to train the model using the clotted part only? This is the case in many images where the two similar figures are there but one has some clots and the other does not.
\n
I have gone through many notebooks but I am unable to get any clarification regarding the same.
", "messages": [ { "id": 1946780, "postDate": "2022-09-20T05:21:17.373Z", "content": "As shown in the below image the upper half seems like the healthy one and the below one contains some clots so, to train a model which can identify clots (of the given two categories), aren't we supposed to train the model using the clotted part only? This is the case in many images where the two similar figures are there but one has some clots and the other does not.
\n
I have gone through many notebooks but I am unable to get any clarification regarding the same.
", "rawMarkdown": "As shown in the below image the upper half seems like the healthy one and the below one contains some clots so, to train a model which can identify clots (of the given two categories), aren't we supposed to train the model using the clotted part only? This is the case in many images where the two similar figures are there but one has some clots and the other does not.\n\n\n\nI have gone through many notebooks but I am unable to get any clarification regarding the same.", "votes": 6 }, { "id": 1965020, "postDate": "2022-10-01T04:34:08.390Z", "content": "I read in some paper that whole slide images could have noise in terms of a few things, including tissue slice folds ( I guess when kept in cold storage, not sure). I believe the sharp dark lines could be that, they are too distinct but not present in all images.
\nThis is just a guess, I remembered from my initial literature search, so sharing here. I think the clots have to be the color contrast patterns given the discussion around the stains and various textures - again, just a guess :)
", "rawMarkdown": "I read in some paper that whole slide images could have noise in terms of a few things, including tissue slice folds ( I guess when kept in cold storage, not sure). I believe the sharp dark lines could be that, they are too distinct but not present in all images.\n\nThis is just a guess, I remembered from my initial literature search, so sharing here. I think the clots have to be the color contrast patterns given the discussion around the stains and various textures - again, just a guess :)", "votes": 1 }, { "id": 1947017, "postDate": "2022-09-20T08:03:19.740Z", "content": "I am not sure, but all the objects in the image are blood clots, I think.🤔
", "rawMarkdown": "I am not sure, but all the objects in the image are blood clots, I think.🤔", "votes": 2, "replies": [ { "id": 1948345, "postDate": "2022-09-21T02:49:39.197Z", "content": "If you are sure about that may be the two similar figures represent different levels of severity.
", "rawMarkdown": "If you are sure about that may be the two similar figures represent different levels of severity." }, { "id": 1949981, "postDate": "2022-09-22T03:49:57.753Z", "content": "According to the introduced literature, they were originated from the formalin-fixed paraffin-embedded clot material. So, I suppose the similar figures may be the adjacent parts from the same clot.🤔
", "rawMarkdown": "According to the introduced literature, they were originated from the formalin-fixed paraffin-embedded clot material. So, I suppose the similar figures may be the adjacent parts from the same clot.🤔" } ] }, { "id": 1948165, "postDate": "2022-09-20T21:45:19.120Z", "content": "I am not sure
", "rawMarkdown": "I am not sure" }, { "id": 1949993, "postDate": "2022-09-22T04:08:22.123Z", "rawMarkdown": "", "votes": 1, "isDeleted": true }, { "id": 1947027, "postDate": "2022-09-20T08:10:01.413Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 1965020, "author_name": "tdiceman", "author_url": "", "post_date": "2022-10-01T04:34:08.390000", "content": "I read in some paper that whole slide images could have noise in terms of a few things, including tissue slice folds ( I guess when kept in cold storage, not sure). I believe the sharp dark lines could be that, they are too distinct but not present in all images.
\nThis is just a guess, I remembered from my initial literature search, so sharing here. I think the clots have to be the color contrast patterns given the discussion around the stains and various textures - again, just a guess :)
", "votes": 1, "replies": [] }, { "id": 1947017, "author_name": "Hiroshi Sakiyama", "author_url": "", "post_date": "2022-09-20T08:03:19.740000", "content": "I am not sure, but all the objects in the image are blood clots, I think.🤔
", "votes": 2, "replies": [ { "id": 1948345, "author_name": "Abhishek Maurya", "author_url": "", "post_date": "2022-09-21T02:49:39.197000", "content": "If you are sure about that may be the two similar figures represent different levels of severity.
", "votes": 0, "replies": [] }, { "id": 1949981, "author_name": "Hiroshi Sakiyama", "author_url": "", "post_date": "2022-09-22T03:49:57.753000", "content": "According to the introduced literature, they were originated from the formalin-fixed paraffin-embedded clot material. So, I suppose the similar figures may be the adjacent parts from the same clot.🤔
", "votes": 0, "replies": [] } ] }, { "id": 1948165, "author_name": "zhehao liang", "author_url": "", "post_date": "2022-09-20T21:45:19.120000", "content": "I am not sure
", "votes": 0, "replies": [] }, { "id": 1949993, "author_name": "", "author_url": "", "post_date": "2022-09-22T04:08:22.123000", "content": "", "votes": 1, "replies": [] }, { "id": 1947027, "author_name": "", "author_url": "", "post_date": "2022-09-20T08:10:01.413000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "1946780": "As shown in the below image the upper half seems like the healthy one and the below one contains some clots so, to train a model which can identify clots (of the given two categories), aren't we supposed to train the model using the clotted part only? This is the case in many images where the two similar figures are there but one has some clots and the other does not.\n\n\n\nI have gone through many notebooks but I am unable to get any clarification regarding the same.", "1965020": "I read in some paper that whole slide images could have noise in terms of a few things, including tissue slice folds ( I guess when kept in cold storage, not sure). I believe the sharp dark lines could be that, they are too distinct but not present in all images.\n\nThis is just a guess, I remembered from my initial literature search, so sharing here. I think the clots have to be the color contrast patterns given the discussion around the stains and various textures - again, just a guess :)", "1947017": "I am not sure, but all the objects in the image are blood clots, I think.🤔", "1948165": "I am not sure", "1949993": "", "1947027": "" } }