Effective and efficient neural networks for spike inference from in vivo calcium imaging

• Published in: Cell reports methods Vol. 3; no. 5; p. 100462
• Main Authors: Zhou, Zhanhong, Yip, Hei Matthew, Tsimring, Katya, Sur, Mriganka, Ip, Jacque Pak Kan, Tin, Chung
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 22.05.2023; Elsevier
• Subjects: calcium imaging; deep learning; neural networks; spike inference; deep learning; spike inference; CP: Neuroscience; calcium imaging; neural networks
• ISSN: 2667-2375; 2667-2375
• DOI: 10.1016/j.crmeth.2023.100462

Calcium imaging provides advantages in monitoring large populations of neuronal activities simultaneously. However, it lacks the signal quality provided by neural spike recording in traditional electrophysiology. To address this issue, we developed a supervised data-driven approach to extract spike information from calcium signals. We propose the ENS2 (effective and efficient neural networks for spike inference from calcium signals) system for spike-rate and spike-event predictions using ΔF/F0 calcium inputs based on a U-Net deep neural network. When testing on a large, ground-truth public database, it consistently outperformed state-of-the-art algorithms in both spike-rate and spike-event predictions with reduced computational load. We further demonstrated that ENS2 can be applied to analyses of orientation selectivity in primary visual cortex neurons. We conclude that it would be a versatile inference system that may benefit diverse neuroscience studies. [Display omitted] •ENS2 uses a U-Net architecture with MSE loss for effective spike inference•ENS2 is calibration free and handles calcium signals with varying signal-to-noise ratios•ENS2 enables characterization of orientation selectivity of V1 neurons•The calcium signal transient amplitude is a key determinant of inference performance Calcium imaging is a powerful tool for monitoring the activities of multiple neurons and understanding brain function, but its generally low signal-to-noise ratio (SNR) and slow dynamics limit its precision and temporal resolution compared with the traditional electrophysiological recording. To overcome this limitation, reliable models for spike inference from calcium signals will be beneficial. Here, we aim to develop efficient and calibration-free spike inference models that generalize well to a broad range of calcium data, including various calcium indicators, SNRs, brain regions, and so on. Zhou et al. develop a spike inference system from calcium signals based on a U-Net deep neural network. It is calibration free and computationally efficient. It performs consistently well with calcium signals of variable signal quality with a single model and facilitates analyses of orientation selectivity in V1 neurons.