scholarly journals MorphVAE: Generating Neural Morphologies from 3D-Walks using a Variational Autoencoder with Spherical Latent Space

2021 ◽  
Author(s):  
Sophie C Laturnus ◽  
Philipp Berens
Keyword(s):  
Ganglion Cells ◽  
Feature Space ◽  
Generative Models ◽  
Generative Model ◽  
Past Century ◽  
Retinal Ganglion ◽  
Cell Type ◽  
Latent Space ◽  
Variational Autoencoder ◽  
Degree Of Similarity

For the past century, the anatomy of a neuron has been considered one of its defining features: The shape of a neuron`s dendrites and axon fundamentally determines what other neurons it can connect to. These neurites have been described using mathematical tools e.g. in the context of cell type classification, but generative models of these structures have only rarely been proposed and are often computationally inefficient. Here we propose MORPHVAE, a sequence-to-sequence variational autoencoder with spherical latent space as a generative model for neural morphologies. The model operates on walks within the tree structure of a neuron and can incorporate expert annotations on a subset of the data using semi-supervised learning. We develop our model on artificially generated toy data and evaluate its performance on dendrites of excitatory cells and axons of inhibitory cells of mouse motor cortex (M1) and dendrites of retinal ganglion cells. We show that the learned latent feature space allows for better cell type discrimination than other commonly used features. By sampling new walks from the latent space we can easily construct new morphologies with a specified degree of similarity to their reference neuron, providing an efficient generative model for neural morphologies.

scholarly journals Collective dynamics of repeated inference in variational autoencoder rapidly find cluster structure

2020 ◽  
Vol 10 (1) ◽  
Author(s):  
Yoshihiro Nagano ◽  
Ryo Karakida ◽  
Masato Okada
Keyword(s):  
Activity Pattern ◽  
Latent Variables ◽  
Cluster Structure ◽  
Generative Models ◽  
Cluster Center ◽  
Specific Data ◽  
Global Cluster ◽  
Latent Space ◽  
Variational Autoencoder ◽  
Low Dimensional

Abstract Deep neural networks are good at extracting low-dimensional subspaces (latent spaces) that represent the essential features inside a high-dimensional dataset. Deep generative models represented by variational autoencoders (VAEs) can generate and infer high-quality datasets, such as images. In particular, VAEs can eliminate the noise contained in an image by repeating the mapping between latent and data space. To clarify the mechanism of such denoising, we numerically analyzed how the activity pattern of trained networks changes in the latent space during inference. We considered the time development of the activity pattern for specific data as one trajectory in the latent space and investigated the collective behavior of these inference trajectories for many data. Our study revealed that when a cluster structure exists in the dataset, the trajectory rapidly approaches the center of the cluster. This behavior was qualitatively consistent with the concept retrieval reported in associative memory models. Additionally, the larger the noise contained in the data, the closer the trajectory was to a more global cluster. It was demonstrated that by increasing the number of the latent variables, the trend of the approach a cluster center can be enhanced, and the generalization ability of the VAE can be improved.

scholarly journals Molecular codes for cell type specification in Brn3 retinal ganglion cells

2017 ◽  
Vol 114 (20) ◽  
pp. E3974-E3983 ◽  
Author(s):  
Szilard Sajgo ◽  
Miruna Georgiana Ghinia ◽  
Matthew Brooks ◽  
Friedrich Kretschmer ◽  
Katherine Chuang ◽  
...  
Keyword(s):  
Retinal Ganglion Cells ◽  
Visual Information ◽  
Ganglion Cells ◽  
Neuronal Cell ◽  
Neuronal Morphology ◽  
Retinal Ganglion ◽  
Cell Type ◽  
Type Specification ◽  
The Brain ◽  
Combinatorial Code

Visual information is conveyed from the eye to the brain by distinct types of retinal ganglion cells (RGCs). It is largely unknown how RGCs acquire their defining morphological and physiological features and connect to upstream and downstream synaptic partners. The three Brn3/Pou4f transcription factors (TFs) participate in a combinatorial code for RGC type specification, but their exact molecular roles are still unclear. We use deep sequencing to define (i) transcriptomes of Brn3a- and/or Brn3b-positive RGCs, (ii) Brn3a- and/or Brn3b-dependent RGC transcripts, and (iii) transcriptomes of retinorecipient areas of the brain at developmental stages relevant for axon guidance, dendrite formation, and synaptogenesis. We reveal a combinatorial code of TFs, cell surface molecules, and determinants of neuronal morphology that is differentially expressed in specific RGC populations and selectively regulated by Brn3a and/or Brn3b. This comprehensive molecular code provides a basis for understanding neuronal cell type specification in RGCs.

Supervised Determined Source Separation with Multichannel Variational Autoencoder

2019 ◽  
Vol 31 (9) ◽  
pp. 1891-1914 ◽  
Author(s):  
Hirokazu Kameoka ◽  
Li Li ◽  
Shota Inoue ◽  
Shoji Makino
Keyword(s):  
Source Separation ◽  
Generative Model ◽  
Separation Technique ◽  
Separation Performance ◽  
Unknown Parameters ◽  
Baseline Method ◽  
Latent Space ◽  
Variational Autoencoder ◽  
Training Examples ◽  
Class Labels

This letter proposes a multichannel source separation technique, the multichannel variational autoencoder (MVAE) method, which uses a conditional VAE (CVAE) to model and estimate the power spectrograms of the sources in a mixture. By training the CVAE using the spectrograms of training examples with source-class labels, we can use the trained decoder distribution as a universal generative model capable of generating spectrograms conditioned on a specified class index. By treating the latent space variables and the class index as the unknown parameters of this generative model, we can develop a convergence-guaranteed algorithm for supervised determined source separation that consists of iteratively estimating the power spectrograms of the underlying sources, as well as the separation matrices. In experimental evaluations, our MVAE produced better separation performance than a baseline method.

Mitochondrial Optic Neuropathies: How Two Genomes may Kill the Same Cell Type?

2007 ◽  
Vol 27 (1-3) ◽  
pp. 173-184 ◽  
Author(s):  
Valerio Carelli ◽  
Chiara La Morgia ◽  
Luisa Iommarini ◽  
Rosanna Carroccia ◽  
Marina Mattiazzi ◽  
...  
Keyword(s):  
Retinal Ganglion Cells ◽  
Ganglion Cells ◽  
Point Mutations ◽  
Mitochondrial Diseases ◽  
Ocular Involvement ◽  
Retinal Ganglion ◽  
Mitochondrial Network ◽  
Cell Type ◽  
Cellular Targets ◽  
Hereditary Optic Neuropathy

Ocular involvement is a prevalent feature in mitochondrial diseases. Leber's hereditary optic neuropathy (LHON) and dominant optic atrophy (DOA) are both non-syndromic optic neuropathies with a mitochondrial etiology. LHON is associated with point mutations in the mitochondrial DNA (mtDNA), which affect subunit genes of complex I. The majority of DOA patients harbor mutations in the nuclear-encoded protein OPA1, which is targeted to mitochondria and participates to cristae organization and mitochondrial network dynamics. In both disorders the retinal ganglion cells (RGCs) are specific cellular targets of the degenerative process. We here review the clinical features and the genetic bases, and delineate the possible common pathomechanism for both these disorders.

scholarly journals Generative models for capturing and exploiting the influence of process conditions on process curves

2021 ◽  
Author(s):  
Tarek Iraki ◽  
Norbert Link
Keyword(s):  
Latent Variable ◽  
Welding Process ◽  
Feature Space ◽  
Generative Models ◽  
Process Conditions ◽  
Process Data ◽  
Task Learning ◽  
Latent Space ◽  
Variable Space ◽  
Novel Method

AbstractVariations of dedicated process conditions (such as workpiece and tool properties) yield different process state evolutions, which are reflected by different time series of the observable quantities (process curves). A novel method is presented, which firstly allows to extract the statistical influence of these conditions on the process curves and its representation via generative models, and secondly represents their influence on the ensemble of curves by transformations of the representation space. A latent variable space is derived from sampled process data, which represents the curves with only few features. Generative models are formed based on conditional propability functions estimated in this space. Furthermore, the influence of conditions on the ensemble of process curves is represented by estimated transformations of the feature space, which map the process curve densities with different conditions on each other. The latent space is formed via Multi-Task-Learning of an auto-encoder and condition-detectors. The latter classifies the latent space representations of the process curves into the considered conditions. The Bayes framework and the Multi-task Learning models are used to obtain the process curve probabilty densities from the latent space densities. The methods are shown to reveal and represent the influence of combinations of workpiece and tool properties on resistance spot welding process curves.

Categorization of Physiologically and Morphologically Characterized non-α/non-β Cat Retinal Ganglion Cells Using Fractal Geometry

Fractals ◽  
1997 ◽  
Vol 05 (04) ◽  
pp. 673-684 ◽  
Author(s):  
H. F. Jelinek ◽  
I. Spence
Keyword(s):  
Fractal Dimension ◽  
Retinal Ganglion Cells ◽  
Ganglion Cells ◽  
Cell Types ◽  
Retinal Ganglion ◽  
Cell Type ◽  
Box Counting ◽  
Standard Values ◽  
Box Counting Method ◽  
Delta Cells

Non-α/non-β cat retinal ganglion cell images were obtained from the published literature, and a homogeneous group of cells was chosen as a standard for each currently accepted cell type (γ, δ and ε). The NIH box-counting method was chosen to determine the fractal dimension (Df) of all cells. The 'standard' values allowed comparisons with other morphologically and physiologically non-α/non-β classified cell types in the literature. We suggest, based on fractal analysis of the dendritic trees, that the morphologically defined γ, δ, and ε cells are distinct types. The W-tonic and W-phasic cell types were further divided into 2 subcategories (W-tonic1, W-tonic2, W-phasic1, W-phasic2). The fractal dimension, of the ε cells being equivalent to the W-tonic1 group and γ cell type equivalent to the W-phasic1 group. Delta cells may be equivalent to either the W-tonic2 or the W-phasic2 group. We discuss the value of the fractal dimension as an added morphological parameter for future morphophysiological classification schemes of vertebrate retinal ganglion cells.

Enriched populations of rat retinal ganglion cells: Studies using a cell-type specific surface marker

1983 ◽  
Vol 5 (6) ◽  
pp. 691-696 ◽  
Author(s):  
Richard Beale ◽  
David W. Beaton ◽  
Volker Neuhoff ◽  
Neville N. Osborne
Keyword(s):  
Specific Surface ◽  
Retinal Ganglion Cells ◽  
Ganglion Cells ◽  
Surface Marker ◽  
Retinal Ganglion ◽  
Cell Type ◽  
A Cell ◽  
Cell Type Specific

scholarly journals Melanopsin Ganglion Cells Are the Most Resistant Retinal Ganglion Cell Type to Axonal Injury in the Rat Retina

PLoS ONE ◽  
2014 ◽  
Vol 9 (3) ◽  
pp. e93274 ◽  
Author(s):  
Luis Pérez de Sevilla Müller ◽  
Allison Sargoy ◽  
Allen R. Rodriguez ◽  
Nicholas C. Brecha
Keyword(s):  
Ganglion Cell ◽  
Retinal Ganglion Cell ◽  
Ganglion Cells ◽  
Axonal Injury ◽  
Retinal Ganglion ◽  
Cell Type ◽  
Rat Retina

scholarly journals Lineage tracing analysis of cone photoreceptor-associated cis-regulatory elements in the developing chicken retina

2019 ◽  
Author(s):  
Estie Schick ◽  
Sean D. McCaffery ◽  
Erin E. Keblish ◽  
Cassandra Thakurdin ◽  
Mark M. Emerson
Keyword(s):  
Progenitor Cells ◽  
Cell Fate ◽  
Retinal Ganglion Cells ◽  
Ganglion Cells ◽  
Regulatory Elements ◽  
Lineage Tracing ◽  
Horizontal Cells ◽  
Retinal Ganglion ◽  
Cone Photoreceptors ◽  
Cell Type

During vertebrate retinal development, transient populations of retinal progenitor cells with restricted cell fate choices are formed. One of these progenitor populations expresses the Thrb gene and can be identified with the ThrbCRM1 cis-regulatory element. Short-term assays have concluded that these cells preferentially generate cone photoreceptors and horizontal cells, however developmental timing has precluded an extensive cell type characterization of their progeny. Here we describe the development and validation of a recombinase-based lineage tracing system for the chicken embryo to further characterize the lineage of these cells. The ThrbCRM1 element was found to preferentially form photoreceptors and horizontal cells, as well as a small number of retinal ganglion cells. The photoreceptor cell progeny are exclusively cone photoreceptors and not rod photoreceptors, confirming that ThrbCRM1-progenitor cells are restricted from the rod fate. In addition, specific subtypes of horizontal cells and retinal ganglion cells were overrepresented, suggesting that ThrbCRM1 progenitor cells are not only restricted for cell type, but for cell subtype as well.

Sign in / Sign up

Export Citation Format

Share Document