EPFL and Harvard scientists unveil a groundbreaking AI method for tracking neurons in moving animals. Using a convolutional neural network (CNN) with ‘targeted augmentation,’ the researchers address the challenge of computationally identifying and tracking neurons in animals with flexible bodies, such as worms. Published in Nature Methods and led by Sahand Jamal Rahi at EPFL’s School of Basic Sciences, the method significantly reduces manual effort by synthesizing reliable annotations for brain deformations. Tested on the roundworm Caenorhabditis elegans, the enhanced CNN measures interneuron activity, revealing complex behaviors. The researchers provide an accessible CNN with a user-friendly interface, promising to triple analysis throughput and accelerate brain imaging research.