Large scale similarity search across digital reconstructions of neural morphology

Abstract

Most functions of the nervous system depend on neuronal and glial morphology. Continuous advances in microscopic imaging and tracing software have provided an increasingly abundant availability of 3D reconstructions of arborizing dendrites, axons, and processes, allowing their detailed study. However, efficient, large-scale methods to rank neural morphologies by similarity to an archetype are still lacking. Using the NeuroMorpho.Org database, we present a similarity search software enabling fast morphological comparison of hundreds of thousands of neural reconstructions from any species, brain regions, cell types, and preparation protocols. We compared the performance of different morphological measurements: 1) summary morphometrics calculated by L-Measure, 2) persistence vectors, a vectorized descriptor of branching structure, 3) the combination of the two. In all cases, we also investigated the impact of applying dimensionality reduction using principal component analysis (PCA). We assessed qualitative performance by gauging the ability to rank neurons in order of visual similarity. Moreover, we quantified information content by examining explained variance and benchmarked the ability to identify occasional duplicate reconstructions of the same specimen. The results indicate that combining summary morphometrics and persistence vectors with applied PCA provides an information rich characterization that enables efficient and precise comparison of neural morphology. The execution time scaled linearly with data set size, allowing seamless live searching through the entire NeuroMorpho.Org content in fractions of a second. We have deployed the similarity search function as an open-source online software tool both through a user-friendly graphical interface and as an API for programmatic access.

Accelerating the continuous community sharing of digital neuromorphology data

“…We also augment all reconstructions with detailed morphometric measurements, such as total arbor length, number of branches, and fractal dimension 42,47 . The metadata and morphometrics enable both more powerful search and filtering capabilities, and stand-alone meta-analyses via conveniently retrievable summary reports 26,27 , adding functional value to the data.…”

“…Other microservices perform feature extraction in the form of neural descriptors, such as persistence vectors 30 and morphometric measurements 26 , from the standardized reconstructions. Those neural descriptors then serve as input to a similarity search microservice that flags potential duplicates upon data screening and allows users to find data morphologies of interest based on an archetype 27 . This decomposition into dedicated services not only enhances the scalability of the pipeline but also allows for parallel implementation of new functionalities, thus accelerating both data processing and continuous development.…”

“…This improvement was facilitated by the development of backend microservices, running as Docker containers 24 , and of routines for rapid and automated formatting, integrity control, ingestion, and incremental sharing of reconstructions. Additionally, we have developed a wide range of tools for further analysis and discovery 25 , including metadata annotation, search functionality, and data conversion [26][27][28] . These advancements have increased both the accessibility and usability of morphological tracings.…”

Automatic identification of scientific publications describing digital reconstructions of neural morphology

“…A potential direction to extend this line of research is to combine and integrate the described work with the recently introduced automated extraction, from the same articles, of rich metadata pertaining to the identified reconstructions, such as the animal species, sex, and age; the brain region and cell type; and the histological and imaging protocols (Bijari et al, 2022). At the same time, it may be worth exploring in future research the possibility of combining full text with figures (Jiang et al 2022;Ljungquist et al 2022), morphological metadata, and bibliographic information (such as journal and authors), to learn new models and improve the results further. (2012)(2013)(2014)(2015), automated search and manual evaluation (2016-2019), and automated search and evaluation (2020-2022).…”

Cell morphologies in the nervous system: Glia steal the limelight

“…Neurons have historically been divided into types by the shape of their arbors (Hamilton et al., 2012), but this approach has been relatively less explored for glia. Efficient metrics and high‐throughput tools for large‐scale comparisons (Bijari et al., 2021; Ljungquist et al., 2022) of heterogeneous glial populations may soon allow to discover the individual morphological signatures of microglia, astrocytes, oligodendrocytes, and other glial types.…”