I just wanted to understand that mentioned in the description of the dataset the BLUE channel refers to nucleus, GREEN to proteins of interest, microtubules in RED and endoplasmic reticulum in YELLOW. Are these classes mutually exclusive, when I say mutually exclusive I mean that all the classes related to the nucleus(
\nI just wanted to understand that mentioned in the description of the dataset the BLUE channel refers to nucleus, GREEN to proteins of interest, microtubules in RED and endoplasmic reticulum in YELLOW. Are these classes mutually exclusive, when I say mutually exclusive I mean that all the classes related to the nucleus(
\nThe example classes in your list will all show as overlapping with the blue nuclei channel, but we are looking for the pattern that is being shown in the green channel. So for example, you might see the green overlap with the nucleoli that is visible in the blue and therefore you can conclude that the protein localizes to the nucleoli.
\nUse the red, blue, and yellow as reference markers and the green as the target marker.
", "rawMarkdown": "The example classes in your list will all show as overlapping with the blue nuclei channel, but we are looking for the pattern that is being shown in the green channel. So for example, you might see the green overlap with the nucleoli that is visible in the blue and therefore you can conclude that the protein localizes to the nucleoli.\n\nUse the red, blue, and yellow as reference markers and the green as the target marker.", "votes": 2, "replies": [ { "id": 1278952, "postDate": "2021-04-20T12:54:10.067Z", "content": "So a combination of Red-green, Blue-green and yellow-green would be a good starting point if I dont want to play with 4-Channels together. Also could you please point out or give a reference, of the 18 classes which class maps to which channel.
\neg: BLUE (Nucleoplasm,Nuclear membrane,Nucleoli,Nucleoli fibrillar center,Nuclear speckles,Nuclear bodies). Is there any other class apart from the above can be seen in the Blue Channel?
\nYELLOW(Endoplasmic reticulum) is there any other cell attribute that can be seen in the Yellow channel and same thing for RED.
\nAny reference point would also help.
\nThanks
Casper can confirm this, but not all classes can be identified by looking at one specific channel. The main channel is green, which gives the most information about the protein. The other channels could be used as references. Recently, there have been some notebooks that only used green channel and achieved a decent score.
", "rawMarkdown": "Casper can confirm this, but not all classes can be identified by looking at one specific channel. The main channel is green, which gives the most information about the protein. The other channels could be used as references. Recently, there have been some notebooks that only used green channel and achieved a decent score. ", "votes": 2 }, { "id": 1284354, "postDate": "2021-04-25T20:06:20.757Z", "content": "Ah, yes, this is entirely true. I should maybe have been more clear. The 3 reference channels are enough to understand the structure of the cells but it doesn't show a perfect overlap with most classes. When annotating manually, one has to use their knowledge of cellular structure in conjunction with where the green is showing to properly annotate. It is not enough to just look at one channel + green.
", "rawMarkdown": "Ah, yes, this is entirely true. I should maybe have been more clear. The 3 reference channels are enough to understand the structure of the cells but it doesn't show a perfect overlap with most classes. When annotating manually, one has to use their knowledge of cellular structure in conjunction with where the green is showing to properly annotate. It is not enough to just look at one channel + green.", "votes": 1 } ] }, { "id": 1278808, "postDate": "2021-04-20T09:41:40.020Z", "content": "Yes, it means that you can classify all the proteins belonging to those six channels just by looking at the blue and green images. This is because all those six categories have some relationship with the nucleus, which is visible in the blue channel. So, you don't need to know about the microtubules or ER information. But, since this is a multi label problem, looking at all channels will be useful to make a prediction. An image might have label of nucleoplasm and ER. In that case, you would need the help of blue and yellow channels to classify it.
", "rawMarkdown": "Yes, it means that you can classify all the proteins belonging to those six channels just by looking at the blue and green images. This is because all those six categories have some relationship with the nucleus, which is visible in the blue channel. So, you don't need to know about the microtubules or ER information. But, since this is a multi label problem, looking at all channels will be useful to make a prediction. An image might have label of nucleoplasm and ER. In that case, you would need the help of blue and yellow channels to classify it. ", "votes": 2, "replies": [ { "id": 1278816, "postDate": "2021-04-20T09:55:31.523Z", "content": "Got it….Thanks for the clarification
", "rawMarkdown": "Got it....Thanks for the clarification", "votes": 2 } ] } ], "comments": [ { "id": 1278930, "author_name": "Casper Winsnes", "author_url": "", "post_date": "2021-04-20T12:34:44.620000", "content": "The example classes in your list will all show as overlapping with the blue nuclei channel, but we are looking for the pattern that is being shown in the green channel. So for example, you might see the green overlap with the nucleoli that is visible in the blue and therefore you can conclude that the protein localizes to the nucleoli.
\nUse the red, blue, and yellow as reference markers and the green as the target marker.
", "votes": 2, "replies": [ { "id": 1278952, "author_name": "Chandan Verma", "author_url": "", "post_date": "2021-04-20T12:54:10.067000", "content": "So a combination of Red-green, Blue-green and yellow-green would be a good starting point if I dont want to play with 4-Channels together. Also could you please point out or give a reference, of the 18 classes which class maps to which channel.
\neg: BLUE (Nucleoplasm,Nuclear membrane,Nucleoli,Nucleoli fibrillar center,Nuclear speckles,Nuclear bodies). Is there any other class apart from the above can be seen in the Blue Channel?
\nYELLOW(Endoplasmic reticulum) is there any other cell attribute that can be seen in the Yellow channel and same thing for RED.
\nAny reference point would also help.
\nThanks
Casper can confirm this, but not all classes can be identified by looking at one specific channel. The main channel is green, which gives the most information about the protein. The other channels could be used as references. Recently, there have been some notebooks that only used green channel and achieved a decent score.
", "votes": 2, "replies": [] }, { "id": 1284354, "author_name": "Casper Winsnes", "author_url": "", "post_date": "2021-04-25T20:06:20.757000", "content": "Ah, yes, this is entirely true. I should maybe have been more clear. The 3 reference channels are enough to understand the structure of the cells but it doesn't show a perfect overlap with most classes. When annotating manually, one has to use their knowledge of cellular structure in conjunction with where the green is showing to properly annotate. It is not enough to just look at one channel + green.
", "votes": 1, "replies": [] } ] }, { "id": 1278808, "author_name": "novice03", "author_url": "", "post_date": "2021-04-20T09:41:40.020000", "content": "Yes, it means that you can classify all the proteins belonging to those six channels just by looking at the blue and green images. This is because all those six categories have some relationship with the nucleus, which is visible in the blue channel. So, you don't need to know about the microtubules or ER information. But, since this is a multi label problem, looking at all channels will be useful to make a prediction. An image might have label of nucleoplasm and ER. In that case, you would need the help of blue and yellow channels to classify it.
", "votes": 2, "replies": [ { "id": 1278816, "author_name": "Chandan Verma", "author_url": "", "post_date": "2021-04-20T09:55:31.523000", "content": "Got it….Thanks for the clarification
", "votes": 2, "replies": [] } ] } ], "raw_markdown_by_id": { "1278767": "I just wanted to understand that mentioned in the description of the dataset the BLUE channel refers to nucleus, GREEN to proteins of interest, microtubules in RED and endoplasmic reticulum in YELLOW. Are these classes mutually exclusive, when I say mutually exclusive I mean that all the classes related to the nucleus(\n0. Nucleoplasm\n1. Nuclear membrane\n2. Nucleoli\n3. Nucleoli fibrillar center\n4. Nuclear speckles\n5. Nuclear bodies) will be represented by the Blue channel only and not any other. Same with Red and yellow channel as well. Could anyone help clarify is my understanding correct?", "1278930": "The example classes in your list will all show as overlapping with the blue nuclei channel, but we are looking for the pattern that is being shown in the green channel. So for example, you might see the green overlap with the nucleoli that is visible in the blue and therefore you can conclude that the protein localizes to the nucleoli.\n\nUse the red, blue, and yellow as reference markers and the green as the target marker.", "1278808": "Yes, it means that you can classify all the proteins belonging to those six channels just by looking at the blue and green images. This is because all those six categories have some relationship with the nucleus, which is visible in the blue channel. So, you don't need to know about the microtubules or ER information. But, since this is a multi label problem, looking at all channels will be useful to make a prediction. An image might have label of nucleoplasm and ER. In that case, you would need the help of blue and yellow channels to classify it. " } }