Using Neural Networks in Diagnosing Breast Cancer

摘要

Computational methods can be used to provide a second opinion in medical settingsand may improve the sensitivity and specificity of diagnoses. In the current study, evolutionary programming is used to train neural networks and linear discriminant models to detect breast cancer in suspicious and microcalcifications using radiographic features and patient age. A cross validation protocol is used to train and atest the networks. ROC curves are used to assess the performance. Results indicate that a significant probability of detecting malignancies can be achieved at the risk of a small percentage of false positives. Typical areas under the ROC curves average 0.9 or better. The results compare well with others offered in the archive literature, while using an order-of magnitude fewer degrees of freedom in the neural classifiers. The research sets the stage for further investigation to automate the assessment of important indicators of breast cancer.