The deep learning revolution in bioimage analysis over the past ~5 years has been nothing short of awe-inspiring; in just a few short years we’ve gone from having very few deep learning tools usable on light microscopy images to new approaches coming out weekly and routinely outperforming older classical approaches. We can now use deep learning to denoise our images, to segment objects in them, make measurements in them, and write code to analyze and graph those measurements.
While this is fantastic, it does create some other problems beyond the ones it solves: how do you figure out if a deep learning model is right for your data? Will you be able to install it on your computer, especially if you’re not computationally savvy? With new models coming out all the time, how do you even keep up with what’s out there and sort the real game-changers from the hype? Will your model be reproducible?
If these problems sound like something you’re trying to solve and you’ll be at the ASCB annual meeting at the at the Boston Convention & Exhibition Center December 2nd-6th, we’d love you to submit an abstract to our ASCB session on Deep Learning and Artificial Intelligence in Cell Imaging – abstracts are due August 1st! We’ll be featuring tools that cover a large number of ways deep learning can interact with images, with a special focus on tools that are user friendly to non-coders.
We look forward to seeing anyone interested in the answers to these questions on Saturday, December 2nd from 1:00 to 3:30 pm. We definitely also plan to see everyone at the “Beyond pretty pictures” subgroup on the morning of December 6th, which will focus on discovering cell biology from microscopy using quantitative methods. It should be a great conference, hope to see you there!