AI can reveal new cell biological discoveries by simply looking at images
The human brain is good at comparing and analyzing images. You can sort dog photos by color, ear size, face shape and more. Could you compare the photos quantitatively? Could a machine, perhaps even more interestingly, extract information from images in a way that humans cannot?
A team of scientists from the Chan Zuckerberg Biohub at Stanford University has now developed a machine-learning method that allows them to compare and analyze images – in this case, microscopy images – without any prior knowledge. Nature Methods reported that their algorithm \”cytoself\” provides detailed information about the location and function of proteins within a cell. This could speed up the research process for cell biologists, and ultimately be used to speed up drug discovery and screening.
Loic Royer said, \”This is a very exciting development–we are applying AI to a brand new type of problem, and we’re still recovering all that humans know plus more.\” In the future, we may be able to do this with different types of images. This opens up many possibilities.
Source:
https://phys.org/news/2022-08-ai-reveal-cell-biology-images.html
Leave a Reply