Classification of signal-time curves from dynamic MR mammography by neuralnetworks

The aim of this study was to test the performance of artificial neural netw orks for the classification of signal-time curves obtained from breast mass es by dynamic MRI. Signal-time courses from 105 parenchyma, 162 malignant, and 102 benign tissue regions were examined. The latter two groups were his topathologically verified. Four neural networks corresponding to different temporal resolutions of the signal-time curves were tested. The resolution ranges from 28 measurements with a temporal spacing of 23s to just 3 measur ements taken 1.8, 3, and 10 minutes after contrast medium administration. D iscrimination between malignant and benign lesions is best if 28 measuremen t points are used (sensitivity: 84%, specificity: 81%). The use of three me asurement points results in 78% sensitivity and 76% specificity. These resu lts correspond to values obtained by human experts who visually evaluated s ignal-time curves without considering additional morphologic information. A LL examined networks yielded poor results for the subclassification of the benign lesions into fibroadenomas and benign proliferative changes. Neural networks can computationally fast distinguish between malignant and benign lesions even when only a few post-contrast measurements are made. More prec ise specification of the type of the benign lesion will require incorporati on of additional morphological or pharmacokinetic information. (C) 2001 Els evier Science Inc. All rights reserved.