search railabs.duhs.duke.edu.

Papers Published

1. Floyd, C.E., Jr. and Yun, A.J. and Lo, J.Y. and Tourassi, G. and Sullivan, D.C. and Kornguth, P.J., Prediction of breast cancer malignancy for difficult cases using an artificial neural network, World Congress on Neural Networks-San Diego. 1994 International Neural Network Society Annual Meeting, vol. vol.1 (1994), pp. 127 - 32 .
   (last updated on 2007/04/15)

   Abstract:
   An artificial neural network was developed to predict breast cancer from mammographic findings. Radiologists read the mammograms and filled out a list of eight findings. These findings were encoded as features for an artificial neural network (ANN). Results from biopsy were taken as truth in the diagnosis of malignancy. The ANN was trained on a set of patient records and was tested on a set for which the radiologists' diagnosis was indeterminate. Performance for the network was evaluated in terms of sensitivity and specificity over a range of decision thresholds and was expressed as an ROC curve. The trained network was evaluated on a subset of patients for which the radiologists' diagnosis was indeterminate. With an optimal threshold, the neural network performed with a diagnostic accuracy of 0.84. This performance suggests that an artificial neural network may be used as a diagnostic aid for prediction of breast cancer

   Keywords:
   learning (artificial intelligence);medical diagnostic computing;neural nets;patient diagnosis;performance evaluation;