Thanks for the video. I can't find a link to the workflow via QuPath documentation you mention in the beginning of the video. Could you please share this?
Hi! Thanks for this detailed and precise description! This is very helpful. I had a question: I am following the steps you described ( analyzing a 16-bit, 3-channel fluorescent IHC image), but my background shows high pixel values, too ( for example, I'm not seeing 0 for dapi where there's no dapi). Should I do a background correction?
Thank you very much for this great video! for me the single measurement classifier doesn´t detect positive cells even if I play with threshold. Do you have any ideas about this? that would be of a great help
A good trick is to move the threshold completely to the left to see if cell detections light up in the right color. Make sure that you have the detections visible and that the color of the cell class is different from the color of unclassified cell detections (red by default). If you still have problems please post the problem on forum.image.sc for a continued discussion.
Is it possible, or practical, to do multi-class learning and classification? I have tried this before, but I couldn't get it to work. This is what keeps me using Inform, which after years of no development, is a slow and unreliable piece of software.
QuPath can use ML for classification of classes - see here github.com/BIIFSweden/SpatialWorkshop/blob/main/part1_qupath/README.md#cell-classification or on the QuPath documentation.
This is a good video. Too bad there are so many commercials that I can barely watch it. Every time I pause to try to repeat it in QuPath I need to watch 2 more commercials. It makes it pretty hard to learn like this 😞
very good explanation you save my life much easier
Thanks for the video. I can't find a link to the workflow via QuPath documentation you mention in the beginning of the video. Could you please share this?
Hi! Thanks for this detailed and precise description! This is very helpful. I had a question: I am following the steps you described ( analyzing a 16-bit, 3-channel fluorescent IHC image), but my background shows high pixel values, too ( for example, I'm not seeing 0 for dapi where there's no dapi). Should I do a background correction?
Hello, thank you for your video. I wondered why my image's background is not 0, but up to 1000.
Thank you very much for this great video!
for me the single measurement classifier doesn´t detect positive cells even if I play with threshold. Do you have any ideas about this? that would be of a great help
A good trick is to move the threshold completely to the left to see if cell detections light up in the right color. Make sure that you have the detections visible and that the color of the cell class is different from the color of unclassified cell detections (red by default). If you still have problems please post the problem on forum.image.sc for a continued discussion.
Hi, what can I do if multiple nuclei are detected as one?
What image file types did you import here? Were they already spectrally unmixed with Inform software prior to analyzing with Qupath?
Yes, it is the spectrally unmixed data sets exported by the Inform software.
Is it possible, or practical, to do multi-class learning and classification? I have tried this before, but I couldn't get it to work. This is what keeps me using Inform, which after years of no development, is a slow and unreliable piece of software.
QuPath can use ML for classification of classes - see here github.com/BIIFSweden/SpatialWorkshop/blob/main/part1_qupath/README.md#cell-classification or on the QuPath documentation.
This is a good video. Too bad there are so many commercials that I can barely watch it. Every time I pause to try to repeat it in QuPath I need to watch 2 more commercials. It makes it pretty hard to learn like this 😞
Hello, if you still have trouble watching youtube with too many commercials, search for an applicable adblocker for your browser!