3 ms·

That is very exciting. Beyond the obvious applications in humans, it would be interesting to test whether love to see whether this could help reverse immune dam

That is very exciting. Beyond the obvious applications in humans, it would be interesting to test whether love to see whether this could help reverse immune damage from cross-species heart transplantation. Small comment, the Hubrecht Institute showing an "artistic representation" that looks exactly like actual histology data is very weird. They should make a cartoon graphic if they are trying to establish a model. How did they even make that? AI?

dekhn 2y ago
I don't think the image is an artistic representation- I think it's an actual micrograph of fluorescently labelled cells. They may have called it that because some level of postprocessing was applied. In short, when this type of micrograph is taken, it's generated using a sequence of laser pulses, each laser pulse has a different wavelength of light designed to match one of the fluorescent labels, which itself emits a different wavelength of light. The images themselves are collected in grayscale and then a false-color image is created by coloring each grayscale image (mapping gray intensity to luminance) and compositing them. The color mapping does not recapitulate the underlying colors of the images that were taken although usually when these images are rendered, a color very close to the emission wavelength of the fluorescent label is used. That said it's not impossible to conjure up prompts that would make a generative AI produce images that look a lot like a micrograph of cells. I skimmed the referenced paper and didn't see this image but I suspect the authors provided it in the PR package they must have distributed.
- pazimzadeh 2y ago
  Yes, fluorescence imaging is a standard technique that I have performed a lot. The colors are arbitrary, and there is always post-processing - adjusting the signal intensity so that any background signal from negative control slides is removed. However, that doesn't make it an artistic representation, which is what their caption says.
  - dekhn 2y ago
    I've seen that term used from time to time to describe false color imagery in astronomy.

dekhn 2y ago

I don't think the image is an artistic representation- I think it's an actual micrograph of fluorescently labelled cells. They may have called it that because some level of postprocessing was applied. In short, when this type of micrograph is taken, it's generated using a sequence of laser pulses, each laser pulse has a different wavelength of light designed to match one of the fluorescent labels, which itself emits a different wavelength of light. The images themselves are collected in grayscale and then a false-color image is created by coloring each grayscale image (mapping gray intensity to luminance) and compositing them. The color mapping does not recapitulate the underlying colors of the images that were taken although usually when these images are rendered, a color very close to the emission wavelength of the fluorescent label is used. That said it's not impossible to conjure up prompts that would make a generative AI produce images that look a lot like a micrograph of cells. I skimmed the referenced paper and didn't see this image but I suspect the authors provided it in the PR package they must have distributed.

pazimzadeh 2y ago
Yes, fluorescence imaging is a standard technique that I have performed a lot. The colors are arbitrary, and there is always post-processing - adjusting the signal intensity so that any background signal from negative control slides is removed. However, that doesn't make it an artistic representation, which is what their caption says.
- dekhn 2y ago
  I've seen that term used from time to time to describe false color imagery in astronomy.