Morphological Neuron Classification Using Machine Learning

Abstract

Classification and quantitative characterization of neuronal morphologies from histological neuronal reconstruction is challenging since it is still unclear how to delineate a neuronal cell class and which are the best features to define them by. The morphological neuron characterization represents a primary source to address anatomical comparisons, morphometric analysis of cells, or brain modeling. The objectives of this paper are (i) to develop and integrate a pipeline that goes from morphological feature extraction to classification and (ii) to assess and compare the accuracy of machine learning algorithms to classify neuron morphologies. The algorithms were trained on 430 digitally reconstructed neurons subjectively classified into layers and/or m-types using young and/or adult development state population of the somatosensory cortex in rats. For supervised algorithms, linear discriminant analysis provided better classification results in comparison with others. For unsupervised algorithms, the affinity propagation and the Ward algorithms provided slightly better results.

Connectivity characterization of the mouse basolateral amygdalar complex

“…The significant group differences are presented with whisker plots ( Figure 8F) and the FDR q-value significance level is visualized in a matrix ( Figure 8E). Furthermore, we used principal component analysis (PCA) to reduce the dimensionality of all measured morphological parameters and create a 3D scatterplot illustrating the segregation of BLAa domain specific neurons based on the measured features ( Figure 8B) 51,52 . Statistically significant differences in several morphological features were detected across all pairwise comparisons; however, neurons in BLA.am and BLA.al generally showed greater similarity to one another than they each did to BLA.ac.…”

“…Due to the anisotropic dimensions of the voxels and spatial undersampling relative to the curvature of the dendrite, we applied a local regression filter to address the aliasing artifact and to regularize dendritic tortuosity. Specifically, the NeuronFilter module in QIT was used to apply a locally weighted scatter-plot smoother (LOESS), which is a low bias approach that makes minimal assumptions (Supplementary Figure 1H 51,52 . The PCA shows the segregation of BLAa domain specific neurons based on the measured features ( Figure 8B).…”

Image-Based Profiling of Synaptic Connectivity in Primary Neuronal Cell Culture

“…To this end, one could consider building a library of reliable stretches and use machine learning to recognize analyzable dendrites during high-content screening. Although relatively new in the field of neuroscience, machine learning approaches have already been used for tracing of neurites ( Gala et al, 2014 ), and to classify cortical neurons in histological sections according to their morphology ( Vasques et al, 2016 ). Recently, a machine learning approach was also used for spine classification, which outperformed morphological feature-based methods ( Ghani et al, 2017 ).…”

Automatic Classification of Traced Neurons Using Morphological Features

“…In this study we focus the attention in the use of morphological features for supervised classification, which has been less treated and has shown better results than unsupervised techniques [14,23]. In this case, a priori information which is used in unsupervised algorithms only to validate the classification process, allowed us to build our models.…”