Neuronal Cell Type Classification Using Locally Sparse Networks

Abstract

The brain is likely the most complex organ, given the variety of functions it controls, the number of cells it comprises, and their corresponding connectivity and diversity. Identifying and studying neurons, the major building blocks of the brain, is a crucial milestone and is essential for understanding brain functionality in health and disease. Recent developments in machine learning have provided advanced abilities for classifying neurons, mainly according to their morphology. This paper aims to provide an explainable deep-learning framework to classify neurons based on their electrophysiological activity. Our analysis is performed on data provided by the Allen Cell Types database. The data contains a survey of biological features derived from single-cell recordings from mice. Neurons are classified into subtypes based on Cre mouse lines using an inherently interpretable locally sparse deep neural network model. We show state-of-the-art results in the neuron classification task while providing explainability to the decisions made by the model.