Abstract

Deep learning is a subdivision of machine learning that employs computation models made of several layers to learn features from data with different levels of abstraction. Implementation of these models has led to a startling improvement in areas such as visual object recognition, drug discovery, and genomics. This paper presents a deep learning algorithm, based on a convolution neural network to classify brain MRI into five classes. The designed model achieves a test accuracy of 97.5% demonstrating the potential of deep learning in automated disease diagnosis. A standalone application has also been developed to display the classifier output and activations of convolution and ReLu layers.

Authors

Edwin Ngera, Segera Davies
University of Nairobi, Kenya

Keywords

Deep Learning, Convolution Neural Network, Machine Learning, MRI