A database and deep learning toolbox for noise-optimized, generalized spike inference from calcium imaging

Peter Rupprecht1,2, Stefano Carta3, Adrian Hoffmann3, Mayumi Echizen4,5, Antonin Blot6,7, Alex C Kwan8, Yang Dan9, Sonja B Hofer6,7, Kazuo Kitamura4,10, Fritjof Helmchen11, Rainer W Friedrich12,13

  1. Brain Research Institute, University of Zürich, Zurich, Switzerland. rupprecht@hifo.uzh.ch.
  2. Friedrich Miescher Institute for Biomedical Research, Basel, Switzerland. rupprecht@hifo.uzh.ch.
  3. Brain Research Institute, University of Zürich, Zurich, Switzerland.
  4. Department of Neurophysiology, University of Tokyo, Tokyo, Japan.
  5. Department of Anesthesiology, Graduate School of Medical and Dental Sciences, Tokyo Medical and Dental University, Tokyo, Japan.
  6. Sainsbury Wellcome Centre for Neural Circuits and Behaviour, University College London, London, United Kingdom.
  7. Biozentrum, University of Basel, Basel, Switzerland.
  8. Department of Psychiatry, Yale University School of Medicine, New Haven, CT, USA.
  9. Division of Neurobiology, Department of Molecular and Cell Biology, Helen Wills Neuroscience Institute, Howard Hughes Medical Institute, University of California, Berkeley, Berkeley CA, USA.
  10. Department of Neurophysiology, University of Yamanashi, Yamanashi, Japan.
  11. Brain Research Institute, University of Zürich, Zurich, Switzerland. helmchen@hifo.uzh.ch.
  12. Friedrich Miescher Institute for Biomedical Research, Basel, Switzerland. rainer.friedrich@fmi.ch.
  13. University of Basel, Basel, Switzerland. rainer.friedrich@fmi.ch.

Abstract

Inference of action potentials ('spikes') from neuronal calcium signals is complicated by the scarcity of simultaneous measurements of action potentials and calcium signals ('ground truth'). In this study, we compiled a large, diverse ground truth database from publicly available and newly performed recordings in zebrafish and mice covering a broad range of calcium indicators, cell types and signal-to-noise ratios, comprising a total of more than 35 recording hours from 298 neurons. We developed an algorithm for spike inference (termed CASCADE) that is based on supervised deep networks, takes advantage of the ground truth database, infers absolute spike rates and outperforms existing model-based algorithms. To optimize performance for unseen imaging data, CASCADE retrains itself by resampling ground truth data to match the respective sampling rate and noise level; therefore, no parameters need to be adjusted by the user. In addition, we developed systematic performance assessments for unseen data, openly released a resource toolbox and provide a user-friendly cloud-based implementation.

Presented By Peter Rupprecht | ORCID iD

NeuroscienceNicholas Liao2021-09 (Sep)