The application of microscopy in biomedical research has come a long way since Antonie van Leeuwenhoek discovered unicellular organisms. Countless innovations have positioned light microscopy as a cornerstone of modern biology and a method of choice for connecting omics datasets to their biological and clinical correlates. Still, regardless of how convincing published imaging data looks, it does not always convey meaningful information about the conditions in which it was acquired, processed, and analyzed. Adequate record-keeping, reporting, and quality control are therefore essential to ensure experimental rigor and data fidelity, allow experiments to be reproducibly repeated, and promote the proper evaluation, interpretation, comparison, and re-use. To this end, microscopy images should be accompanied by complete descriptions detailing experimental procedures, biological samples, microscope hardware specifications, image acquisition parameters, and image analysis procedures, as well as metrics accounting for instrument performance and calibration. However, universal, community-accepted Microscopy Metadata standards and reporting specifications that would result in Findable Accessible Interoperable and Reproducible (FAIR) microscopy data have not yet been established. To understand this shortcoming and to propose a way forward, here we provide an overview of the nature of microscopy metadata and its importance for fostering data quality, reproducibility, scientific rigor, and sharing value in light microscopy. The proposal for tiered Microscopy Metadata Specifications that extend the OME Data Model put forth by the 4D Nucleome Initiative and by Bioimaging North America [1-3] as well as a suite of three complementary and interoperable tools are being developed to facilitate the process of image data documentation and are presented in related manuscripts [4-6].

The microvasculature underlies the supply networks that support neuronal activity within heterogeneous brain regions. What are common versus heterogeneous aspects of the connectivity, density, and orientation of capillary networks? To address this, we imaged, reconstructed, and analyzed the microvasculature connectome in whole adult mice brains with sub-micrometer resolution. Graph analysis revealed common network topology across the brain that leads to a shared structural robustness against the rarefaction of vessels. Geometrical analysis, based on anatomically accurate reconstructions, uncovered a scaling law that links length density, i.e., the length of vessel per volume, with tissue-to-vessel distances. We then derive a formula that connects regional differences in metabolism to differences in length density and, further, predicts a common value of maximum tissue oxygen tension across the brain. Last, the orientation of capillaries is weakly anisotropic with the exception of a few strongly anisotropic regions; this variation can impact the interpretation of fMRI data.

Molecular profiles of neurons influence information processing, but bridging the gap between genes, circuits, and behavior has been very difficult. Furthermore, the behavioral state of an animal continuously changes across development and as a result of sensory experience. How behavioral state influences molecular cell state is poorly understood. Here we present a complete atlas of the Drosophila larval central nervous system composed of over 200,000 single cells across four developmental stages. We develop polyseq, a python package, to perform cell-type analyses. We use single-molecule RNA-FISH to validate our scRNAseq findings. To investigate how internal state affects cell state, we optogentically altered internal state with high-throughput behavior protocols designed to mimic wasp sting and over activation of the memory system. We found nervous system-wide and neuron-specific gene expression changes. This resource is valuable for developmental biology and neuroscience, and it advances our understanding of how genes, neurons, and circuits generate behavior.

Cellular versatility depends on accurate trafficking of diverse proteins to their organellar destinations. For the secretory pathway (followed by approximately 30% of all proteins), the physical nature of the vessel conducting the first portage (endoplasmic reticulum [ER] to Golgi apparatus) is unclear. We provide a dynamic 3D view of early secretory compartments in mammalian cells with isotropic resolution and precise protein localization using whole-cell, focused ion beam scanning electron microscopy with cryo-structured illumination microscopy and live-cell synchronized cargo release approaches. Rather than vesicles alone, the ER spawns an elaborate, interwoven tubular network of contiguous lipid bilayers (ER exit site) for protein export. This receptacle is capable of extending microns along microtubules while still connected to the ER by a thin neck. COPII localizes to this neck region and dynamically regulates cargo entry from the ER, while COPI acts more distally, escorting the detached, accelerating tubular entity on its way to joining the Golgi apparatus through microtubule-directed movement.

The ability to probe the membrane potential of multiple genetically defined neurons simultaneously would have a profound impact on neuroscience research. Genetically encoded voltage indicators are a promising tool for this purpose, and recent developments have achieved a high signal-to-noise ratio in vivo with 1-photon fluorescence imaging. However, these recordings exhibit several sources of noise and signal extraction remains a challenge. We present an improved signal extraction pipeline, spike-guided penalized matrix decomposition-nonnegative matrix factorization (SGPMD-NMF), which resolves supra- and subthreshold voltages in vivo. The method incorporates biophysical and optical constraints. We validate the pipeline with simultaneous patch-clamp and optical recordings from mouse layer 1 in vivo and with simulated and composite datasets with realistic noise. We demonstrate applications to mouse hippocampus expressing paQuasAr3-s or SomArchon1, mouse cortex expressing SomArchon1 or Voltron, and zebrafish spines expressing zArchon1.

The Importance Weighted Auto Encoder (IWAE) objective has been shown to improve the training of generative models over the standard Variational Auto Encoder (VAE) objective. Here, we derive importance weighted extensions to Adversarial Variational Bayes (AVB) and Adversarial Autoencoder (AAE). These latent variable models use implicitly defined inference networks whose approximate posterior density qφ(z|x) cannot be directly evaluated, an essential ingredient for importance weighting. We show improved training and inference in latent variable models with our adversarially trained importance weighting method, and derive new theoretical connections between adversarial generative model training criteria and marginal likelihood based methods. We apply these methods to the important problem of inferring spiking neural activity from calcium imaging data, a challenging posterior inference problem in neuroscience, and show that posterior samples from the adversarial methods outperform factorized posteriors used in VAEs.

Single-beam scanning electron microscopes (SEM) are widely used to acquire massive datasets for biomedical study, material analysis, and fabrication inspection. Datasets are typically acquired with uniform acquisition: applying the electron beam with the same power and duration to all image pixels, even if there is great variety in the pixels' importance for eventual use. Many SEMs are now able to move the beam to any pixel in the field of view without delay, enabling them, in principle, to invest their time budget more effectively with non-uniform imaging.

We present a simple, yet effective, auxiliary learning task for the problem of neuron segmentation in electron microscopy volumes. The auxiliary task consists of the prediction of Local Shape Descriptors (LSDs), which we combine with conventional voxel-wise direct neighbor affinities for neuron boundary detection. The shape descriptors are designed to capture local statistics about the neuron to be segmented, such as diameter, elongation, and direction. On a large study comparing several existing methods across various specimen, imaging techniques, and resolutions, we find that auxiliary learning of LSDs consistently increases segmentation accuracy of affinity-based methods over a range of metrics. Furthermore, the addition of LSDs promotes affinitybased segmentation methods to be on par with the current state of the art for neuron segmentation (Flood-Filling Networks, FFN), while being two orders of magnitudes more efficient—a critical requirement for the processing of future petabyte-sized datasets. Implementations of the new auxiliary learning task, network architectures, training, prediction, and evaluation code, as well as the datasets used in this study are publicly available as a benchmark for future method contributions.Competing Interest StatementThe authors have declared no competing interest.