Classification of Stem Cells with Convolutional Neural Networks

Abstract

In this paper, the problem of image classification with convolutional neural networks is considered. In several fields of biology, automatized cell detection is a helpful tool for facilitating the process of cellular analysis. This paper answers the question whether a computer program can tell if an image contains muscle stem cells or not. Analogously to the neurons of the human brain, the creation of such a program involves training thousands of mathematically modeled artificial neurons to maximize the likelihood of producing correct classifications. This paper covers how such a network is implemented and shows how its performance depends on the network’s dimensions. It is revealed that a neural network indeed can replace and speed up the manual process of classifying images. With an image dataset of cells, the best performing networks manage to classify images with an accuracy of up to 90%.