Red Blood Cell Classification using Convolutional Networks Exploring Machine Learning Applications in Pathology

This project investigates the application of machine learning techniques to problems in pathology. Using a dataset compiled by pathologists at the Yale School of Medicine, I develop a model to classify red blood cells into different groups of pathological significance. A number of convolutional network architectures are explored and tested on the original dataset of 750 images. Using a deep convolutional neural network, I am able to achieve up to 97% accuracy, a significant improvement over previous approaches.