| |
Abstract:
We discuss the development of a Multi-Layer Perceptron neural
network classifier for use in preoperative differentiation between
benign and malignant ovarian tumors. As the Mean Squared
classification Error is not sufficient to make correct and
objective assessments about the performance of the neural
classifier, the concepts of sensitivity and specificity are
introduced and combined in Receiver Operating Characteristic
curves. Based on objective observations such as sonomorphologic
criteria, color Doppler imaging and results from serum tumor
markers, the neural network is able to make reliable predictions
with a discriminating performance comparable to that of experienced
gynecologists.
|
