I've generated a pdf file (attached) with a sample of original training set images and the full masks placed side by side. I am observing that most of the masks in the whole set are for the cells themselves rather than the nuclei. So my question is, are we supposed to find the nuclei as described in the brief or should we find the cells? Or perhaps, when there ain't any clear nuclei visible we should mask the cells and when clearly visible nuclei are present then we should mask them only?
\n\nFrom the provided training set though, if we see the sample figures 1 and 5 in the pdf, they have clearly visible nuclei and the nuclei only are masked. But in figure 15, I think I can see atleast 2 nuclei and still the cells have been masked in those instances. So I am confused about the requirements.
\n\nI am new to all this though, may be I am missing something.
", "messages": [ { "id": "308989", "postDate": "04/04/2018 13:19:06", "content": "I've generated a pdf file (attached) with a sample of original training set images and the full masks placed side by side. I am observing that most of the masks in the whole set are for the cells themselves rather than the nuclei. So my question is, are we supposed to find the nuclei as described in the brief or should we find the cells? Or perhaps, when there ain't any clear nuclei visible we should mask the cells and when clearly visible nuclei are present then we should mask them only?
\n\nFrom the provided training set though, if we see the sample figures 1 and 5 in the pdf, they have clearly visible nuclei and the nuclei only are masked. But in figure 15, I think I can see atleast 2 nuclei and still the cells have been masked in those instances. So I am confused about the requirements.
\n\nI am new to all this though, may be I am missing something.
", "rawMarkdown": "I've generated a pdf file (attached) with a sample of original training set images and the full masks placed side by side. I am observing that most of the masks in the whole set are for the cells themselves rather than the nuclei. So my question is, are we supposed to find the nuclei as described in the brief or should we find the cells? Or perhaps, when there ain't any clear nuclei visible we should mask the cells and when clearly visible nuclei are present then we should mask them only? \n\nFrom the provided training set though, if we see the sample figures 1 and 5 in the pdf, they have clearly visible nuclei and the nuclei only are masked. But in figure 15, I think I can see atleast 2 nuclei and still the cells have been masked in those instances. So I am confused about the requirements.\n\nI am new to all this though, may be I am missing something.", "votes": null }, { "id": "308994", "postDate": "04/04/2018 13:24:35", "content": "Hi there, I checked through the PDF and indeed everything marked is a nucleus, not a full cell. It takes some practice and experience to know what to look for, but hopefully this is reassuring.
\n\nIn Fig 15, I wonder if you are seeing the nucleoli, which are dark holes within each nucleus and thinking that the nucleoli are nuclei?
", "rawMarkdown": "Hi there, I checked through the PDF and indeed everything marked is a nucleus, not a full cell. It takes some practice and experience to know what to look for, but hopefully this is reassuring.\n\nIn Fig 15, I wonder if you are seeing the nucleoli, which are dark holes within each nucleus and thinking that the nucleoli are nuclei?", "votes": null }, { "id": "309001", "postDate": "04/04/2018 13:31:11", "content": "Thanks for the prompt answer! Yes as I don't have much knowledge about this domain, I was wrongly thinking that the masked areas are cells, my bad. Thanks for the clarification, surely it's reassuring and it becomes a tad bit easier too I feel.
\n\nOkay now I understand that in the fig 15 those small dots inside the two nuclei are 'nucleoli' actually, thanks again, regards!
", "rawMarkdown": "Thanks for the prompt answer! Yes as I don't have much knowledge about this domain, I was wrongly thinking that the masked areas are cells, my bad. Thanks for the clarification, surely it's reassuring and it becomes a tad bit easier too I feel.\n\nOkay now I understand that in the fig 15 those small dots inside the two nuclei are 'nucleoli' actually, thanks again, regards!", "votes": null }, { "id": "309017", "postDate": "04/04/2018 13:48:04", "content": "Hi Anne! Further n this regard, another question comes to my mind now. Do biologists have to work hard to differentiate between nuclei and nucleoli too? If so, are there a large number of these cases or they are not much of a concern? In case they are a real concern, are there any particular properties (e.g. shape and size etc.) of each of them which can help us to differentiate between the nuclei and nucleoli? Thanks.
", "rawMarkdown": "Hi Anne! Further n this regard, another question comes to my mind now. Do biologists have to work hard to differentiate between nuclei and nucleoli too? If so, are there a large number of these cases or they are not much of a concern? In case they are a real concern, are there any particular properties (e.g. shape and size etc.) of each of them which can help us to differentiate between the nuclei and nucleoli? Thanks.", "votes": null }, { "id": "309031", "postDate": "04/04/2018 14:19:30", "content": "I'd say it's pretty obvious to a biologist which is which. I suppose nucleoli are usually the same color as background, that is a tip. Another clue is that cells are not often nucleus-shaped - that is nuclei are usually fairly oval-shaped whereas cell borders usually have more irregular edges (but there are exceptions!)
", "rawMarkdown": "I'd say it's pretty obvious to a biologist which is which. I suppose nucleoli are usually the same color as background, that is a tip. Another clue is that cells are not often nucleus-shaped - that is nuclei are *usually* fairly oval-shaped whereas cell borders usually have more irregular edges (but there are exceptions!)", "votes": null }, { "id": "309036", "postDate": "04/04/2018 14:36:02", "content": "Okay, I get it. I think we don't have to worry too much about the Nucleoli then. Thank you so much for sharing your knowledge and giving us the insight!
", "rawMarkdown": "Okay, I get it. I think we don't have to worry too much about the Nucleoli then. Thank you so much for sharing your knowledge and giving us the insight!", "votes": null }, { "id": "309836", "postDate": "04/06/2018 03:37:36", "content": "Hi Anne! I've another question and I am asking it in this post itself as there is already a discussion and my question is simple too.
\n\nWithin an image, would the Nuclei always be more or less the same size on an average (this is what I've observed in the training set mostly) ? Or, can there be cases where within an image, some nuclei are large whereas some are very small ?
\n\nThanks!
", "rawMarkdown": "Hi Anne! I've another question and I am asking it in this post itself as there is already a discussion and my question is simple too. \n\nWithin an image, would the Nuclei always be more or less the same size on an average (this is what I've observed in the training set mostly) ? Or, can there be cases where within an image, some nuclei are large whereas some are very small ? \n\nThanks!", "votes": null }, { "id": "311121", "postDate": "04/09/2018 12:47:19", "content": "Generally, all nuclei in an image are in the same size range. But, there are major exceptions such as micronuclei which we chose to annotate/include in this challenge. They are much tinier than the typical nucleus in the image.
", "rawMarkdown": "Generally, all nuclei in an image are in the same size range. But, there are major exceptions such as micronuclei which we chose to annotate/include in this challenge. They are much tinier than the typical nucleus in the image.", "votes": null }, { "id": "311142", "postDate": "04/09/2018 13:39:05", "content": "Thanks for the information Anne! So what I understand is that there can be a mix of micronuclei and typical nuclei within the same image. It'd be very useful for me if you could provide a few image IDs from the train data for these type of images?
\n\nWhat I am trying to gauge and confirm from experts like you is that other from the different types of conditions in which these images are taken, what general inherent features different types of nuclei can have that can be useful in searching them out. Thanks.
", "rawMarkdown": "Thanks for the information Anne! So what I understand is that there can be a mix of micronuclei and typical nuclei within the same image. It'd be very useful for me if you could provide a few image IDs from the train data for these type of images? \n\nWhat I am trying to gauge and confirm from experts like you is that other from the different types of conditions in which these images are taken, what general inherent features different types of nuclei can have that can be useful in searching them out. Thanks.", "votes": null } ], "comments": [ { "id": 308994, "author_name": "drannecarpenter", "author_url": "", "post_date": "04/04/2018 13:24:35", "content": "Hi there, I checked through the PDF and indeed everything marked is a nucleus, not a full cell. It takes some practice and experience to know what to look for, but hopefully this is reassuring.
\n\nIn Fig 15, I wonder if you are seeing the nucleoli, which are dark holes within each nucleus and thinking that the nucleoli are nuclei?
", "votes": null, "replies": [ { "id": 309001, "author_name": "kishupro", "author_url": "", "post_date": "04/04/2018 13:31:11", "content": "Thanks for the prompt answer! Yes as I don't have much knowledge about this domain, I was wrongly thinking that the masked areas are cells, my bad. Thanks for the clarification, surely it's reassuring and it becomes a tad bit easier too I feel.
\n\nOkay now I understand that in the fig 15 those small dots inside the two nuclei are 'nucleoli' actually, thanks again, regards!
", "votes": null, "replies": [] }, { "id": 309017, "author_name": "kishupro", "author_url": "", "post_date": "04/04/2018 13:48:04", "content": "Hi Anne! Further n this regard, another question comes to my mind now. Do biologists have to work hard to differentiate between nuclei and nucleoli too? If so, are there a large number of these cases or they are not much of a concern? In case they are a real concern, are there any particular properties (e.g. shape and size etc.) of each of them which can help us to differentiate between the nuclei and nucleoli? Thanks.
", "votes": null, "replies": [] }, { "id": 309031, "author_name": "drannecarpenter", "author_url": "", "post_date": "04/04/2018 14:19:30", "content": "I'd say it's pretty obvious to a biologist which is which. I suppose nucleoli are usually the same color as background, that is a tip. Another clue is that cells are not often nucleus-shaped - that is nuclei are usually fairly oval-shaped whereas cell borders usually have more irregular edges (but there are exceptions!)
", "votes": null, "replies": [] }, { "id": 309036, "author_name": "kishupro", "author_url": "", "post_date": "04/04/2018 14:36:02", "content": "Okay, I get it. I think we don't have to worry too much about the Nucleoli then. Thank you so much for sharing your knowledge and giving us the insight!
", "votes": null, "replies": [] }, { "id": 309836, "author_name": "kishupro", "author_url": "", "post_date": "04/06/2018 03:37:36", "content": "Hi Anne! I've another question and I am asking it in this post itself as there is already a discussion and my question is simple too.
\n\nWithin an image, would the Nuclei always be more or less the same size on an average (this is what I've observed in the training set mostly) ? Or, can there be cases where within an image, some nuclei are large whereas some are very small ?
\n\nThanks!
", "votes": null, "replies": [] }, { "id": 311121, "author_name": "drannecarpenter", "author_url": "", "post_date": "04/09/2018 12:47:19", "content": "Generally, all nuclei in an image are in the same size range. But, there are major exceptions such as micronuclei which we chose to annotate/include in this challenge. They are much tinier than the typical nucleus in the image.
", "votes": null, "replies": [] }, { "id": 311142, "author_name": "kishupro", "author_url": "", "post_date": "04/09/2018 13:39:05", "content": "Thanks for the information Anne! So what I understand is that there can be a mix of micronuclei and typical nuclei within the same image. It'd be very useful for me if you could provide a few image IDs from the train data for these type of images?
\n\nWhat I am trying to gauge and confirm from experts like you is that other from the different types of conditions in which these images are taken, what general inherent features different types of nuclei can have that can be useful in searching them out. Thanks.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "308989": "I've generated a pdf file (attached) with a sample of original training set images and the full masks placed side by side. I am observing that most of the masks in the whole set are for the cells themselves rather than the nuclei. So my question is, are we supposed to find the nuclei as described in the brief or should we find the cells? Or perhaps, when there ain't any clear nuclei visible we should mask the cells and when clearly visible nuclei are present then we should mask them only? \n\nFrom the provided training set though, if we see the sample figures 1 and 5 in the pdf, they have clearly visible nuclei and the nuclei only are masked. But in figure 15, I think I can see atleast 2 nuclei and still the cells have been masked in those instances. So I am confused about the requirements.\n\nI am new to all this though, may be I am missing something.", "308994": "Hi there, I checked through the PDF and indeed everything marked is a nucleus, not a full cell. It takes some practice and experience to know what to look for, but hopefully this is reassuring.\n\nIn Fig 15, I wonder if you are seeing the nucleoli, which are dark holes within each nucleus and thinking that the nucleoli are nuclei?", "309001": "Thanks for the prompt answer! Yes as I don't have much knowledge about this domain, I was wrongly thinking that the masked areas are cells, my bad. Thanks for the clarification, surely it's reassuring and it becomes a tad bit easier too I feel.\n\nOkay now I understand that in the fig 15 those small dots inside the two nuclei are 'nucleoli' actually, thanks again, regards!", "309017": "Hi Anne! Further n this regard, another question comes to my mind now. Do biologists have to work hard to differentiate between nuclei and nucleoli too? If so, are there a large number of these cases or they are not much of a concern? In case they are a real concern, are there any particular properties (e.g. shape and size etc.) of each of them which can help us to differentiate between the nuclei and nucleoli? Thanks.", "309031": "I'd say it's pretty obvious to a biologist which is which. I suppose nucleoli are usually the same color as background, that is a tip. Another clue is that cells are not often nucleus-shaped - that is nuclei are *usually* fairly oval-shaped whereas cell borders usually have more irregular edges (but there are exceptions!)", "309036": "Okay, I get it. I think we don't have to worry too much about the Nucleoli then. Thank you so much for sharing your knowledge and giving us the insight!", "309836": "Hi Anne! I've another question and I am asking it in this post itself as there is already a discussion and my question is simple too. \n\nWithin an image, would the Nuclei always be more or less the same size on an average (this is what I've observed in the training set mostly) ? Or, can there be cases where within an image, some nuclei are large whereas some are very small ? \n\nThanks!", "311121": "Generally, all nuclei in an image are in the same size range. But, there are major exceptions such as micronuclei which we chose to annotate/include in this challenge. They are much tinier than the typical nucleus in the image.", "311142": "Thanks for the information Anne! So what I understand is that there can be a mix of micronuclei and typical nuclei within the same image. It'd be very useful for me if you could provide a few image IDs from the train data for these type of images? \n\nWhat I am trying to gauge and confirm from experts like you is that other from the different types of conditions in which these images are taken, what general inherent features different types of nuclei can have that can be useful in searching them out. Thanks." }, "source": "meta" }