Volumetric functional imaging is widely used for recording neuron activities in vivo , but there exist tradeoffs between the quality of the extracted calcium traces, imaging speed, and laser power. While deep-learning methods have recently been applied to denoise images, their applications to downstream analyses, such as recovering high-SNR calcium traces, have been limited. Further, these methods require temporally-linked pre-registered data with ultrafast rates. Here, we demonstrate supervised deep-denoising methods to circumvent these tradeoffs for several applications, including whole-brain imaging, large field-of-view imaging in freely moving animals, and recovering complex neurite structures in C. elegans . Our framework has 30x smaller memory footprint, and is fast in training and inference (50-70ms); it is highly accurate and generalizable, and further, only small, non-temporally-sequential, independently-acquired training datasets (∼500 images) are needed. We envision that the framework will enable faster and long-term imaging experiments necessary to study neuronal mechanisms of many behaviors.

Fast, Efficient, and Accurate Neuro-Imaging Denoising via Deep Learning