Differentiable simulations of optical systems can be combined with deep learning-based reconstruction networks to enable high performance computational imaging via end-to-end (E2E) optimization of both the optical encoder and the deep decoder. This has enabled imaging applications such as 3D localization microscopy, depth estimation, and lensless photography via the optimization of local optical encoders. More challenging computational imaging applications, such as 3D snapshot microscopy which compresses 3D volumes into single 2D images, require a highly non-local optical encoder. We show that existing deep network decoders have a locality bias which prevents the optimization of such highly non-local optical encoders. We address this with a decoder based on a shallow neural network architecture using global kernel Fourier convolutional neural networks (FourierNets). We show that FourierNets surpass existing deep network based decoders at reconstructing photographs captured by the highly non-local DiffuserCam optical encoder. Further, we show that FourierNets enable E2E optimization of highly non-local optical encoders for 3D snapshot microscopy. By combining FourierNets with a large-scale multi-GPU differentiable optical simulation, we are able to optimize non-local optical encoders 170

Motor systems must continuously adapt their output to maintain a desired trajectory. While the spinal circuits underlying rhythmic locomotion are well described, little is known about how the network modulates its output strength. A major challenge has been the difficulty of recording from spinal neurons during behavior. Here, we use voltage imaging to map the membrane potential of large populations of glutamatergic neurons throughout the spinal cord of the larval zebrafish during fictive swimming in a virtual environment. We characterized a previously undescribed subpopulation of tonic-spiking ventral V3 neurons whose spike rate correlated with swimming strength and bout length. Optogenetic activation of V3 neurons led to stronger swimming and longer bouts but did not affect tail beat frequency. Genetic ablation of V3 neurons led to reduced locomotor adaptation. The power of voltage imaging allowed us to identify V3 neurons as a critical driver of locomotor adaptation in zebrafish.

Calcium imaging has been widely adopted for its ability to record from large neuronal populations. To summarize the time course of neural activity, dimensionality reduction methods, which have been applied extensively to population spiking activity, may be particularly useful. However, it is unclear if the dimensionality reduction methods applied to spiking activity are appropriate for calcium imaging. We thus carried out a systematic study of design choices based on standard dimensionality reduction methods. We also developed a novel method to perform deconvolution and dimensionality reduction simultaneously (termed CILDS). CILDS most accurately recovered the single-trial, low-dimensional time courses from calcium imaging that would have been recovered from spiking activity. CILDS also outperformed the other methods on calcium imaging recordings from larval zebrafish and mice. More broadly, this study represents a foundation for summarizing calcium imaging recordings of large neuronal populations using dimensionality reduction in diverse experimental settings.

Motor systems must continuously adapt their output to maintain a desired trajectory. While the spinal circuits underlying rhythmic locomotion are well described, little is known about how the network modulates its output strength. A major challenge has been the difficulty of recording from spinal neurons during behavior. Here, we use voltage imaging to map the membrane potential of large populations of glutamatergic neurons throughout the spinal cord of the larval zebrafish during fictive swimming in a virtual environment. We characterized a previously undescribed subpopulation of tonic-spiking ventral V3 neurons whose spike rate correlated with swimming strength and bout length. Optogenetic activation of V3 neurons led to stronger swimming and longer bouts but did not affect tail beat frequency. Genetic ablation of V3 neurons led to reduced locomotor adaptation. The power of voltage imaging allowed us to identify V3 neurons as a critical driver of locomotor adaptation in zebrafish.

To accurately track self-location, animals need to integrate their movements through space. In amniotes, representations of self-location have been found in regions such as the hippocampus. It is unknown whether more ancient brain regions contain such representations and by which pathways they may drive locomotion. Fish displaced by water currents must prevent uncontrolled drift to potentially dangerous areas. We found that larval zebrafish track such movements and can later swim back to their earlier location. Whole-brain functional imaging revealed the circuit enabling this process of positional homeostasis. Position-encoding brainstem neurons integrate optic flow, then bias future swimming to correct for past displacements by modulating inferior olive and cerebellar activity. Manipulation of position-encoding or olivary neurons abolished positional homeostasis or evoked behavior as if animals had experienced positional shifts. These results reveal a multiregional hindbrain circuit in vertebrates for optic flow integration, memory of self-location, and its neural pathway to behavior.Competing Interest StatementThe authors have declared no competing interest.

The small size and translucency of larval zebrafish () have made it a unique experimental system to investigate whole-brain neural circuit structure and function. Still, the connectivity patterns between most neuronal types remain mostly unknown. This gap in knowledge underscores the critical need for effective neural circuit mapping tools, especially ones that can integrate structural and functional analyses. To address this, we previously developed a vesicular stomatitis virus (VSV) based approach called Tracer with Restricted Anterograde Spread (TRAS). TRAS utilizes lentivirus to complement replication-incompetent VSV (VSVΔG) to allow restricted (monosynaptic) anterograde labeling from projection neurons to their target cells in the brain. Here, we report the second generation of TRAS (TRAS-M51R), which utilizes a mutant variant of VSVΔG [VSV(M51R)ΔG] with reduced cytotoxicity. Within the primary visual pathway, we found that TRAS-M51R significantly improved long-term viability of transsynaptic labeling (compared to TRAS) while maintaining anterograde spread activity. By using Cre-expressing VSV(M51R)ΔG, TRAS-M51R could selectively label excitatory ( positive) and inhibitory ( positive) retinorecipient neurons. We further show that these labeled excitatory and inhibitory retinorecipient neurons retained neuronal excitability upon visual stimulation at 5-8 days post fertilization (2-5 days post-infection). Together, these findings show that TRAS-M51R is suitable for neural circuit studies that integrate structural connectivity, cell-type identity, and neurophysiology.

3D snapshot microscopy enables fast volumetric imaging by capturing a 3D volume in a single 2D camera image and performing computational reconstruction. Fast volumetric imaging has a variety of biological applications such as whole brain imaging of rapid neural activity in larval zebrafish. The optimal microscope design for this optical 3D-to-2D encoding is both sample- and task-dependent, with no general solution known. Deep learning based decoders can be combined with a differentiable simulation of an optical encoder for end-to-end optimization of both the deep learning decoder and optical encoder. This technique has been used to engineer local optical encoders for other problems such as depth estimation, 3D particle localization, and lensless photography. However, 3D snapshot microscopy is known to require a highly non-local optical encoder which existing UNet-based decoders are not able to engineer. We show that a neural network architecture based on global kernel Fourier convolutional neural networks can efficiently decode information from multiple depths in a volume, globally encoded across a 3D snapshot image. We show in simulation that our proposed networks succeed in engineering and reconstructing optical encoders for 3D snapshot microscopy where the existing state-of-the-art UNet architecture fails. We also show that our networks outperform the state-of-the-art learned reconstruction algorithms for a computational photography dataset collected on a prototype lensless camera which also uses a highly non-local optical encoding.

The brain is tasked with choosing actions that maximize an animal's chances of survival and reproduction. These choices must be flexible and informed by the current state of the environment, the needs of the body, and the outcomes of past actions. This information is physiologically encoded and processed across different brain regions on a wide range of spatial scales, from molecules in single synapses to networks of brain areas. Uncovering these spatially distributed neural interactions underlying behavior requires investigations that span a similar range of spatial scales. Larval zebrafish, given their small size, transparency, and ease of genetic access, are a good model organism for such investigations, allowing the use of modern microscopy, molecular biology, and computational techniques. These approaches are yielding new insights into the mechanistic basis of behavioral states, which we review here and compare to related studies in mammalian species.

Methods for one-photon fluorescent imaging of calcium dynamics can capture the activity of hundreds of neurons across large fields of view at a low equipment complexity and cost. In contrast to two-photon methods, however, one-photon methods suffer from higher levels of crosstalk from neuropil, resulting in a decreased signal-to-noise ratio and artifactual correlations of neural activity. We address this problem by engineering cell-body-targeted variants of the fluorescent calcium indicators GCaMP6f and GCaMP7f. We screened fusions of GCaMP to natural, as well as artificial, peptides and identified fusions that localized GCaMP to within 50 μm of the cell body of neurons in mice and larval zebrafish. One-photon imaging of soma-targeted GCaMP in dense neural circuits reported fewer artifactual spikes from neuropil, an increased signal-to-noise ratio, and decreased artifactual correlation across neurons. Thus, soma-targeting of fluorescent calcium indicators facilitates usage of simple, powerful, one-photon methods for imaging neural calcium dynamics.

Genetically encodable calcium ion (Ca) indicators (GECIs) based on green fluorescent proteins (GFP) are powerful tools for imaging of cell signaling and neural activity in model organisms. Following almost 2 decades of steady improvements in the GFP-based GCaMP series of GECIs, the performance of the most recent generation (i.e., jGCaMP7) may have reached its practical limit due to the inherent properties of GFP. In an effort to sustain the steady progression toward ever-improved GECIs, we undertook the development of a new GECI based on the bright monomeric GFP, mNeonGreen (mNG). The resulting indicator, mNG-GECO1, is 60% brighter than GCaMP6s in vitro and provides comparable performance as demonstrated by imaging Ca dynamics in cultured cells, primary neurons, and in vivo in larval zebrafish. These results suggest that mNG-GECO1 is a promising next-generation GECI that could inherit the mantle of GCaMP and allow the steady improvement of GECIs to continue for generations to come.