Reference

A database and deep learning toolbox for noise-optimized, generalized spike inference from calcium imaging
Peter Rupprecht, Stefano Carta, Adrian Hoffmann, Mayumi Echizen, Antonin Blot, Alex C. Kwan, Yang Dan, Sonja B. Hofer, Kazuo Kitamura, Fritjof Helmchen, and Rainer W. Friedrich

Summary

The task of inferring neuronal action potentials (spikes) from calcium imaging data has historically been difficult due to the lack of “ground truth” datasets (simultaneous calcium and electrophysiological measurements). To address this, the authors constructed a massive, diverse database containing over 35 recording hours from 298 neurons across mice and zebrafish, covering a wide variety of calcium indicators, cell types, and signal-to-noise ratios.

Building upon this comprehensive dataset, the authors introduce CASCADE, a supervised deep learning-based algorithm for spike inference. Key features and contributions of CASCADE include:

  • Superior Performance: It directly infers absolute spike rates, outperforming traditional model-based inference algorithms.
  • Parameter-Free Generalization: To ensure high performance on novel, unseen data, CASCADE automatically retrains itself by resampling ground truth data to perfectly match the sampling rate and noise characteristics of the new input data. This eliminates the need for manual parameter tuning by the user.
  • Accessibility: Alongside the algorithm, the authors developed systematic performance assessments for unseen data and released the tool as an open-source resource, complete with a user-friendly cloud-based implementation.

simultaneous measurements of action potentials and calcium signals (‘ground truth’)

The relationship between spikes and the evoked calcium signals depends on multiple factors, including neuron type, calcium indicator type and concentration, optical resolution, sampling rate and noise level. Many of these parameters can vary substantially between experiments and even among neurons within the same experiment

we performed juxtacellular recordings and two-photon calcium imaging using different calcium indicators and in different brain regions of zebrafish and mice