A use of a neural network to evaluate contrast enhancement curves in breast magnetic resonance images

For the diagnosis of breast cancer using magnetic resonance imaging (MRI), one of the most important parameters is the analysis of contrast enhancemen t. A three-dimensional MR sequence is applied before and five times after b olus injection of paramagnetic contrast medium (Gd-DTPA). The dynamics of a bsorption are described by a time/intensity enhancement curve, which report s the mean intensity of the MR signal in a small region of interest (ROI) f or about 8 minutes after contrast injection. The aim of our study was to us e an artificial neural network to automatically classify the enhancement cu rves as "benign" or "malignant." We used a classic feed-forward back-propag ation neural network, with three layers: five input nodes, two hidden nodes , and one output node. The network has been trained with 26 pathologic curv es (10 invasive carcinoma [K], two carcinoma-in-situ [DCIS], and 14 benign lesion [BI). The trained network has been tested with 58 curves (36 K, one DCIS, 21 B). The network was able to correctly identify the test curves wit h a sensitivity of 76% and a specificity of 90%. For comparison, the same s et of curves was analyzed separately by two radiologists (a breast MR exper t and a resident radiologist). The first correctly interpreted the curves w ith a sensitivity of 76% and a specificity of 90%, while the second scored 59% for sensitivity and 90% for specificity. These results demonstrate that a trained neural network recognizes the pathologic curves at least as well as an expert radiologist. This algorithm can help the radiologist attain r apid and affordable screening of a large number of ROIs. A complete automat ic computer-aided diagnosis support system should find a number of potentia lly interesting ROIs and automatically analyze the enhancement curves for e ach ROI by neural networks, reporting to the radiologist only the potential ly pathologic ROIs for a more accurate, manual, repeated evaluation. Copyri ght (C) 2001 by W.B. Saunders Company.