Congratulations!! 🥳🥳🥳

And a big thank you to Dr. Alix Le Marois and Dr. Erik Sahai for providing this amazing dataset❤️(see "Imaging of MAP kinase dynamics reveals endocytic regulation of pulsatile signalling and network re-wiring in response to targeted therapy in EGFR-mutant non-small cell lung cancer")

A huge thank you ❤️ to @lucapanconi.bsky.social , @sebjbauer.bsky.social , Emma Latron, Maxime Gestin, Erik Sahai, Alix Le Marois, @juliettegriffie.bsky.social and our funders @cancerresearchuk.org , @kawresearch.bsky.social , @erc.europa.eu
biorxiv.org/content/10.1... (5/5)

Our model is small and versatile – this allows us to train it on a common laptop on the whole dataset in under 20 minutes ⚡️😌and use limited manual data annotations✍️ for accurate predictions (4/n)

We were able to reliably predict cell fate from the very first frame ⚡️of 3 days long acquisition, and identify relevant biological processes🧬 involved in cancer persister cells, despite their known biological variability (3/n)

We propose a fully data-driven approach to early cell fate prediction from a microscopy time-lapse. We combine an adversarial autoencoder architecture with a multi-objective loss function, including contrastive learning, to optimise both cell fate prediction and interpretability (2/n)

Video 4: "How to Implement your Liquid Engine Method" youtu.be/gRGEjdT8opY (8/8)

Video 3: "How to Benchmark your Implementations with the Liquid Engine in 1 minute" (7/n) youtu.be/9hF7nLtFzoo

Video 2: "How to Build your Liquid Engine Class in 1 minute" youtu.be/QQsXrZ_jFa8 (6/n)

HOW CAN YOU USE #NanoPyx ? We recorded a series of 1-minute video tutorials that showcase how you can use #NanoPyx, and the capabilities of the #LiquidEngine on your own image analysis workflows! Video 1: youtu.be/Dx2lHoRB044 (5/n)

What methods does #NanoPyx offer? We have Image Registration, Denoising, Super-Resolution Image Reconstruction🔬 and Quality Control metrics! (4/n)

The #LiquidEngine has an AI-powered agent 🤖 that predicts the fastest way of running a specific analysis task, given the input data, the user hardware and task parameters. Incorporating the #LiquidEngine significantly improves the performance of bioimage analysis workflows!(3/n)