Eliminating accidental deviations to minimize generalization error and maximize replicability: Applications in connectomics and genomics.

Replicability, the ability to replicate scientific findings, is a prerequisite for scientific discovery and clinical utility. Troublingly, we are in the midst of a replicability crisis. A key to replicability is that multiple measurements of the same item (e.g., experimental sample or clinical participant) under fixed experimental constraints are relatively similar to one another. Thus, statistics that quantify the relative contributions of accidental deviations-such as measurement error-as compared to systematic deviations-such as individual differences-are critical. We demonstrate that existing replicability statistics, such as intra-class correlation coefficient and fingerprinting, fail to adequately differentiate between accidental and systematic deviations in very simple settings. We therefore propose a novel statistic, discriminability, which quantifies the degree to which an individual's samples are relatively similar to one another, without restricting the data to be univariate, Gaussian, or even Euclidean. Using this statistic, we introduce the possibility of optimizing experimental design via increasing discriminability and prove that optimizing discriminability improves performance bounds in subsequent inference tasks. In extensive simulated and real datasets (focusing on brain imaging and demonstrating on genomics), only optimizing data discriminability improves performance on all subsequent inference tasks for each dataset. We therefore suggest that designing experiments and analyses to optimize discriminability may be a crucial step in solving the replicability crisis, and more generally, mitigating accidental measurement error.

Container-Based Clinical Solutions for Portable and Reproducible Image Analysis.

Medical imaging analysis depends on the reproducibility of complex computation. Linux containers enable the abstraction, installation, and configuration of environments so that software can be both distributed in self-contained images and used repeatably by tool consumers. While several initiatives in neuroimaging have adopted approaches for creating and sharing more reliable scientific methods and findings, Linux containers are not yet mainstream in clinical settings. We explore related technologies and their efficacy in this setting, highlight important shortcomings, demonstrate a simple use-case, and endorse the use of Linux containers for medical image analysis.

Using machine learning on clinical data to identify unexpected patterns in groups of COVID-19 patients.

As clinicians are faced with a deluge of clinical data, data science can play an important role in highlighting key features driving patient outcomes, aiding in the development of new clinical hypotheses. Insight derived from machine learning can serve as a clinical support tool by connecting care providers with reliable results from big data analysis that identify previously undetected clinical patterns. In this work, we show an example of collaboration between clinicians and data scientists during the COVID-19 pandemic, identifying sub-groups of COVID-19 patients with unanticipated outcomes or who are high-risk for severe disease or death. We apply a random forest classifier model to predict adverse patient outcomes early in the disease course, and we connect our classification results to unsupervised clustering of patient features that may underpin patient risk. The paradigm for using data science for hypothesis generation and clinical decision support, as well as our triaged classification approach and unsupervised clustering methods to determine patient cohorts, are applicable to driving rapid hypothesis generation and iteration in a variety of clinical challenges, including future public health crises.

Connectomics Annotation Metadata Standardization for Increased Accessibility and Queryability.

Neuroscientists can leverage technological advances to image neural tissue across a range of different scales, potentially forming the basis for the next generation of brain atlases and circuit reconstructions at submicron resolution, using Electron Microscopy and X-ray Microtomography modalities. However, there is variability in data collection, annotation, and storage approaches, which limits effective comparative and secondary analysis. There has been great progress in standardizing interfaces for large-scale spatial image data, but more work is needed to standardize annotations, especially metadata associated with neuroanatomical entities. Standardization will enable validation, sharing, and replication, greatly amplifying investment throughout the connectomics community. We share key design considerations and a usecase developed for metadata for a recent large-scale dataset.

The Brain Observatory Storage Service and Database (BossDB): A Cloud-Native Approach for Petascale Neuroscience Discovery.

Technological advances in imaging and data acquisition are leading to the development of petabyte-scale neuroscience image datasets. These large-scale volumetric datasets pose unique challenges since analyses often span the entire volume, requiring a unified platform to access it. In this paper, we describe the Brain Observatory Storage Service and Database (BossDB), a cloud-based solution for storing and accessing petascale image datasets. BossDB provides support for data ingest, storage, visualization, and sharing through a RESTful Application Programming Interface (API). A key feature is the scalable indexing of spatial data and automatic and manual annotations to facilitate data discovery. Our project is open source and can be easily and cost effectively used for a variety of modalities and applications, and has effectively worked with datasets over a petabyte in size.

Online learning for orientation estimation during translation in an insect ring attractor network.

Insect neural systems are a promising source of inspiration for new navigation algorithms, especially on low size, weight, and power platforms. There have been unprecedented recent neuroscience breakthroughs with Drosophila in behavioral and neural imaging experiments as well as the mapping of detailed connectivity of neural structures. General mechanisms for learning orientation in the central complex (CX) of Drosophila have been investigated previously; however, it is unclear how these underlying mechanisms extend to cases where there is translation through an environment (beyond only rotation), which is critical for navigation in robotic systems. Here, we develop a CX neural connectivity-constrained model that performs sensor fusion, as well as unsupervised learning of visual features for path integration; we demonstrate the viability of this circuit for use in robotic systems in simulated and physical environments. Furthermore, we propose a theoretical understanding of how distributed online unsupervised network weight modification can be leveraged for learning in a trajectory through an environment by minimizing orientation estimation error. Overall, our results may enable a new class of CX-derived low power robotic navigation algorithms and lead to testable predictions to inform future neuroscience experiments.

A framework for rigorous evaluation of human performance in human and machine learning comparison studies.

Rigorous comparisons of human and machine learning algorithm performance on the same task help to support accurate claims about algorithm success rates and advances understanding of their performance relative to that of human performers. In turn, these comparisons are critical for supporting advances in artificial intelligence. However, the machine learning community has lacked a standardized, consensus framework for performing the evaluations of human performance necessary for comparison. We demonstrate common pitfalls in a designing the human performance evaluation and propose a framework for the evaluation of human performance, illustrating guiding principles for a successful comparison. These principles are first, to design the human evaluation with an understanding of the differences between human and algorithm cognition; second, to match trials between human participants and the algorithm evaluation, and third, to employ best practices for psychology research studies, such as the collection and analysis of supplementary and subjective data and adhering to ethical review protocols. We demonstrate our framework's utility for designing a study to evaluate human performance on a one-shot learning task. Adoption of this common framework may provide a standard approach to evaluate algorithm performance and aid in the reproducibility of comparisons between human and machine learning algorithm performance.

MTNeuro: A Benchmark for Evaluating Representations of Brain Structure Across Multiple Levels of Abstraction.

There are multiple scales of abstraction from which we can describe the same image, depending on whether we are focusing on fine-grained details or a more global attribute of the image. In brain mapping, learning to automatically parse images to build representations of both small-scale features (e.g., the presence of cells or blood vessels) and global properties of an image (e.g., which brain region the image comes from) is a crucial and open challenge. However, most existing datasets and benchmarks for neuroanatomy consider only a single downstream task at a time. To bridge this gap, we introduce a new dataset, annotations, and multiple downstream tasks that provide diverse ways to readout information about brain structure and architecture from the same image. Our multi-task neuroimaging benchmark (MTNeuro) is built on volumetric, micrometer-resolution X-ray microtomography images spanning a large thalamocortical section of mouse brain, encompassing multiple cortical and subcortical regions. We generated a number of different prediction challenges and evaluated several supervised and self-supervised models for brain-region prediction and pixel-level semantic segmentation of microstructures. Our experiments not only highlight the rich heterogeneity of this dataset, but also provide insights into how self-supervised approaches can be used to learn representations that capture multiple attributes of a single image and perform well on a variety of downstream tasks. Datasets, code, and pre-trained baseline models are provided at: https://mtneuro.github.io/.

DotMotif: an open-source tool for connectome subgraph isomorphism search and graph queries.

Recent advances in neuroscience have enabled the exploration of brain structure at the level of individual synaptic connections. These connectomics datasets continue to grow in size and complexity; methods to search for and identify interesting graph patterns offer a promising approach to quickly reduce data dimensionality and enable discovery. These graphs are often too large to be analyzed manually, presenting significant barriers to searching for structure and testing hypotheses. We combine graph database and analysis libraries with an easy-to-use neuroscience grammar suitable for rapidly constructing queries and searching for subgraphs and patterns of interest. Our approach abstracts many of the computer science and graph theory challenges associated with nanoscale brain network analysis and allows scientists to quickly conduct research at scale. We demonstrate the utility of these tools by searching for motifs on simulated data and real public connectomics datasets, and we share simple and complex structures relevant to the neuroscience community. We contextualize our findings and provide case studies and software to motivate future neuroscience exploration.

An Integrated Toolkit for Extensible and Reproducible Neuroscience.

As neuroimagery datasets continue to grow in size, the complexity of data analyses can require a detailed understanding and implementation of systems computer science for storage, access, processing, and sharing. Currently, several general data standards (e.g., Zarr, HDF5, precomputed) and purpose-built ecosystems (e.g., BossDB, CloudVolume, DVID, and Knossos) exist. Each of these systems has advantages and limitations and is most appropriate for different use cases. Using datasets that don't fit into RAM in this heterogeneous environment is challenging, and significant barriers exist to leverage underlying research investments. In this manuscript, we outline our perspective for how to approach this challenge through the use of community provided, standardized interfaces that unify various computational backends and abstract computer science challenges from the scientist. We introduce desirable design patterns and share our reference implementation called intern.

CONFIRMS: A Toolkit for Scalable, Black Box Connectome Assessment and Investigation.

The nanoscale connectomics community has recently generated automated and semi-automated "wiring diagrams" of brain subregions from terabytes and petabytes of dense 3D neuroimagery. This process involves many challenging and imperfect technical steps, including dense 3D image segmentation, anisotropic nonrigid image alignment and coregistration, and pixel classification of each neuron and their individual synaptic connections. As data volumes continue to grow in size, and connectome generation becomes increasingly commonplace, it is important that the scientific community is able to rapidly assess the quality and accuracy of a connectome product to promote dataset analysis and reuse. In this work, we share our scalable toolkit for assessing the quality of a connectome reconstruction via targeted inquiry and large-scale graph analysis, and to provide insights into how such connectome proofreading processes may be improved and optimized in the future. We illustrate the applications and ecosystem on a recent reference dataset.Clinical relevance- Large-scale electron microscopy (EM) data offers a novel opportunity to characterize etiologies and neurological diseases and conditions at an unprecedented scale. EM is useful for low-level analyses such as biopsies; this increased scale offers new possibilities for research into areas such as neural networks if certain bottlenecks and problems are overcome.

Benchmarking Human Performance for Visual Search of Aerial Images.

Aerial images are frequently used in geospatial analysis to inform responses to crises and disasters but can pose unique challenges for visual search when they contain low resolution, degraded information about color, and small object sizes. Aerial image analysis is often performed by humans, but machine learning approaches are being developed to complement manual analysis. To date, however, relatively little work has explored how humans perform visual search on these tasks, and understanding this could ultimately help enable human-machine teaming. We designed a set of studies to understand what features of an aerial image make visual search difficult for humans and what strategies humans use when performing these tasks. Across two experiments, we tested human performance on a counting task with a series of aerial images and examined the influence of features such as target size, location, color, clarity, and number of targets on accuracy and search strategies. Both experiments presented trials consisting of an aerial satellite image; participants were asked to find all instances of a search template in the image. Target size was consistently a significant predictor of performance, influencing not only accuracy of selections but the order in which participants selected target instances in the trial. Experiment 2 demonstrated that the clarity of the target instance and the match between the color of the search template and the color of the target instance also predicted accuracy. Furthermore, color also predicted the order of selecting instances in the trial. These experiments establish not only a benchmark of typical human performance on visual search of aerial images but also identify several features that can influence the task difficulty level for humans. These results have implications for understanding human visual search on real-world tasks and when humans may benefit from automated approaches.

A substrate for modular, extensible data-visualization.

BACKGROUND: As the scope of scientific questions increase and datasets grow larger, the visualization of relevant information correspondingly becomes more difficult and complex. Sharing visualizations amongst collaborators and with the public can be especially onerous, as it is challenging to reconcile software dependencies, data formats, and specific user needs in an easily accessible package. RESULTS: We present substrate, a data-visualization framework designed to simplify communication and code reuse across diverse research teams. Our platform provides a simple, powerful, browser-based interface for scientists to rapidly build effective three-dimensional scenes and visualizations. We aim to reduce the limitations of existing systems, which commonly prescribe a limited set of high-level components, that are rarely optimized for arbitrarily large data visualization or for custom data types. CONCLUSIONS: To further engage the broader scientific community and enable seamless integration with existing scientific workflows, we also present pytri, a Python library that bridges the use of substrate with the ubiquitous scientific computing platform, Jupyter. Our intention is to lower the activation energy required to transition between exploratory data analysis, data visualization, and publication-quality interactive scenes.

A three-dimensional thalamocortical dataset for characterizing brain heterogeneity.

Neural microarchitecture is heterogeneous, varying both across and within brain regions. The consistent identification of regions of interest is one of the most critical aspects in examining neurocircuitry, as these structures serve as the vital landmarks with which to map brain pathways. Access to continuous, three-dimensional volumes that span multiple brain areas not only provides richer context for identifying such landmarks, but also enables a deeper probing of the microstructures within. Here, we describe a three-dimensional X-ray microtomography imaging dataset of a well-known and validated thalamocortical sample, encompassing a range of cortical and subcortical structures from the mouse brain . In doing so, we provide the field with access to a micron-scale anatomical imaging dataset ideal for studying heterogeneity of neural structure.

Toward a scalable framework for reproducible processing of volumetric, nanoscale neuroimaging datasets.

BACKGROUND: Emerging neuroimaging datasets (collected with imaging techniques such as electron microscopy, optical microscopy, or X-ray microtomography) describe the location and properties of neurons and their connections at unprecedented scale, promising new ways of understanding the brain. These modern imaging techniques used to interrogate the brain can quickly accumulate gigabytes to petabytes of structural brain imaging data. Unfortunately, many neuroscience laboratories lack the computational resources to work with datasets of this size: computer vision tools are often not portable or scalable, and there is considerable difficulty in reproducing results or extending methods. RESULTS: We developed an ecosystem of neuroimaging data analysis pipelines that use open-source algorithms to create standardized modules and end-to-end optimized approaches. As exemplars we apply our tools to estimate synapse-level connectomes from electron microscopy data and cell distributions from X-ray microtomography data. To facilitate scientific discovery, we propose a generalized processing framework, which connects and extends existing open-source projects to provide large-scale data storage, reproducible algorithms, and workflow execution engines. CONCLUSIONS: Our accessible methods and pipelines demonstrate that approaches across multiple neuroimaging experiments can be standardized and applied to diverse datasets. The techniques developed are demonstrated on neuroimaging datasets but may be applied to similar problems in other domains.

Neural Reconstruction Integrity: A Metric for Assessing the Connectivity Accuracy of Reconstructed Neural Networks.

Neuroscientists are actively pursuing high-precision maps, or graphs consisting of networks of neurons and connecting synapses in mammalian and non-mammalian brains. Such graphs, when coupled with physiological and behavioral data, are likely to facilitate greater understanding of how circuits in these networks give rise to complex information processing capabilities. Given that the automated or semi-automated methods required to achieve the acquisition of these graphs are still evolving, we developed a metric for measuring the performance of such methods by comparing their output with those generated by human annotators ("ground truth" data). Whereas classic metrics for comparing annotated neural tissue reconstructions generally do so at the voxel level, the metric proposed here measures the "integrity" of neurons based on the degree to which a collection of synaptic terminals belonging to a single neuron of the reconstruction can be matched to those of a single neuron in the ground truth data. The metric is largely insensitive to small errors in segmentation and more directly measures accuracy of the generated brain graph. It is our hope that use of the metric will facilitate the broader community's efforts to improve upon existing methods for acquiring brain graphs. Herein we describe the metric in detail, provide demonstrative examples of the intuitive scores it generates, and apply it to a synthesized neural network with simulated reconstruction errors. Demonstration code is available.

Quantifying Mesoscale Neuroanatomy Using X-Ray Microtomography.

Methods for resolving the three-dimensional (3D) microstructure of the brain typically start by thinly slicing and staining the brain, followed by imaging numerous individual sections with visible light photons or electrons. In contrast, X-rays can be used to image thick samples, providing a rapid approach for producing large 3D brain maps without sectioning. Here we demonstrate the use of synchrotron X-ray microtomography (µCT) for producing mesoscale (∼1 µm 3 resolution) brain maps from millimeter-scale volumes of mouse brain. We introduce a pipeline for µCT-based brain mapping that develops and integrates methods for sample preparation, imaging, and automated segmentation of cells, blood vessels, and myelinated axons, in addition to statistical analyses of these brain structures. Our results demonstrate that X-ray tomography achieves rapid quantification of large brain volumes, complementing other brain mapping and connectomics efforts.