Theses and Dissertations
Analysis of artificial neural networks in the diagnosing of breast cancer using fine needle aspirates
Date of Award
Master of Science (MS)
Dr. Laura Grabowski
Dr. Emmett Tomai
Dr. Dongchul Kim
This thesis examines how Artificial Neural Networks can be used to classify a set of samples from a fine needle aspirate dataset. The dataset is composed of various different attributes, each of which are used to come to the conclusion as to whether a sample is benign or malignant. To automate the process of analyzing the various attributes and coming to a correct prediction, a neural network was implemented. First, a Feedforward Neural Network was trained with the dataset using a Backpropagation training method and an activation sigmoid function with one hidden layer in the architecture of the network. After training, the network performed a 10-fold cross validation to determine which model had the lower error scores and would perform the best on the data. The data was looped through the model and the trained network classified the samples as either benign or malignant. Once classified, the overall accuracy, specificity and sensitivity were analyzed to measure performance. Three other neural networks were compared to the Feedforward Network to see how they performed. These three neural networks included a NEAT Neural Network, a Support Vector Machine, and a Radial Basis Function Neural Network.
Vazquez, Janette, "Analysis of artificial neural networks in the diagnosing of breast cancer using fine needle aspirates" (2016). Theses and Dissertations. 163.
Copyright 2016 Janette Vazquez. All Rights Reserved.