Automatic Blood-Cell Classification via Convolutional Neural Networks and Transfer Learning

Abstract

The evaluation and diagnosis of cancer related diseases can be complex and lengthy. This is exacerbated due to manual analyses based on techniques that may take copious amount of time. In the last decade, different tools have been created to detect, analyze and classify different types of cancer in humans. However, there is still a lack of tools or models to automate the analysis of human cells to determine the presence of cancer. Such a model has the potential to improve early detection and prevention of said diseases, leading to more timely medical diagnoses. In this research, we present our current effort on the development of a Deep Learning Model capable of identifying blood cell anomalies. Our results show an accuracy that meets or exceeds the current state of the art, particularly achieving lower false negative rate in comparison to previous efforts reported.