A Neuronal Morphology Classification Approach Based on Locally Cumulative Connected Deep Neural Networks

Zeng

et al. 2021

Sci Rep

The study of cellular complexity in the nervous system based on anatomy has shown more practical and objective advantages in morphology than other perspectives on molecular, physiological, and evolutionary aspects. However, morphology-based neuron type classification in the whole rat brain is challenging, given the significant number of neuron types, limited reconstructed neuron samples, and diverse data formats. Here, we report that different types of deep neural network modules may well process different kinds of features and that the integration of these submodules will show power on the representation and classification of neuron types. For SWC-format data, which are compressed but unstructured, we construct a tree-based recurrent neural network (Tree-RNN) module. For 2D or 3D slice-format data, which are structured but with large volumes of pixels, we construct a convolutional neural network (CNN) module. We also generate a virtually simulated dataset with two classes, reconstruct a CASIA rat-neuron dataset with 2.6 million neurons without labels, and select the NeuroMorpho-rat dataset with 35,000 neurons containing hierarchical labels. In the twelve-class classification task, the proposed model achieves state-of-the-art performance compared with other models, e.g., the CNN, RNN, and support vector machine based on hand-designed features.

Section: Discussionmentioning

confidence: 99%

Neuron type classification in rat brain based on integrative convolutional and tree-based recurrent neural networks

Zeng

et al. 2021

Sci Rep

“…Neuronal morphology and function vary across and within defined regions of the brain, and morphology is associated with function [22][23][24]. Therefore, well-defined cellular criteria are important and should be established prior to image acquisition to determine which neurons are suitable for analysis (Fig.…”

Section: Cellular Criteriamentioning

confidence: 99%

An Optimized and Detailed Step-by-Step Protocol for the Analysis of Neuronal Morphology in Golgi-Stained Fetal Sheep Brain

et al. 2022

Antenatal brain development during the final trimester of human pregnancy is a time when mature neurons become increasingly complex in morphology, through axonal and dendritic outgrowth, dendritic branching, and synaptogenesis, together with myelin production. Characterizing neuronal morphological development over time is of interest to developmental neuroscience and provides the framework to measure grey matter pathology in pregnancy compromise. Neuronal microstructure can be assessed with Golgi staining, which selectively stains a small percentage (1-3%) of neurons and their entire dendritic arbor. Advanced imaging processing and analysis tools can then be employed to quantitate neuronal cytoarchitecture. Traditional Golgi staining protocols have been optimized and commercial kits are readily available offering improved speed and sensitivity of Golgi staining to produce consistent results. Golgi stained tissue is then visualized under light microscopy and image analysis may be completed with several software programs for morphological analysis of neurons, including freeware and commercial products. Each program requires optimization, whether semi-automated or automated, requiring different levels of investigator intervention and interpretation, which is a critical consideration for unbiased analysis. Detailed protocols for fetal ovine brain tissue are lacking and therefore, we provide a step-by-step workflow of computer software analysis for morphometric quantification of Golgi-stained neurons. Here, we utilized the commonly applied FD Rapid GolgiStain kit (FD NeuroTechnologies) on ovine fetal brains collected at 127 days (0.85) gestational age for the analysis of CA1 pyramidal neurons in the hippocampus. We describe the step-by-step protocol to retrieve neuronal morphometrics using Imaris imaging software to provide quantification of apical and basal dendrites for measures of dendrite length (μm), branch number, branch order and Sholl analysis (intersections over radius). We also detail software add-ons for data retrieval of dendritic spines including the number of spines, spine density and spine classification, which are critical indicators of synaptic function. The assessment of neuronal morphology in the developing brain using Rapid-Golgi and Imaris software is labour-intensive, particularly during the optimization period. The methodology described in this step-by-step description is novel, detailed, and aims to provide a reproducible, working protocol to quantify neuronal cytoarchitecture with simple descriptions that will save time for the next users of these commonly used techniques.

“…Such morphological features can be used by a downstream machine learning pipeline. Segmentation, although not always perfect, can be often well performed, and although the annotation of neurons is still not yet fully automated, significant progress has been made on this topic with very promising results ( Li et al., 2019 ; Lin and Zheng, 2019 ; Schubert et al., 2019 ). Normalization of RNA quantities in different compartments can then be done by quantization methods, such as DypFISH ( Savulescu et al., 2021 ) and others.…”

Section: Limitations In the Data That Might Challenge Predictionmentioning

confidence: 99%

Prediction of RNA subcellular localization: Learning from heterogeneous data sources

et al. 2021

Summary RNA subcellular localization has recently emerged as a widespread phenomenon, which may apply to the majority of RNAs. The two main sources of data for characterization of RNA localization are sequence features and microscopy images, such as obtained from single-molecule fluorescent in situ hybridization-based techniques. Although such imaging data are ideal for characterization of RNA distribution, these techniques remain costly, time-consuming, and technically challenging. Given these limitations, imaging data exist only for a limited number of RNAs. We argue that the field of RNA localization would greatly benefit from complementary techniques able to characterize location of RNA. Here we discuss the importance of RNA localization and the current methodology in the field, followed by an introduction on prediction of location of molecules. We then suggest a machine learning approach based on the integration between imaging localization data and sequence-based data to assist in characterization of RNA localization on a transcriptome level.