Mammography is widely recognized as the most reliable technique for early detection of breast cancers. Automated or semi-automated computerized classification schemes can be very useful in assisting radiologists with a second opinion about the visual diagnosis of breast lesions, thus leading to a reduction in the number of unnecessary biopsies. We present a computer-aided diagnosis (CADi) system for the characterization of massive lesions in mammograms, whose aim is to distinguish malignant from benign masses. The CADi system we realized is based on a three-stage algorithm: (a) a segmentation technique extracts the contours of the massive lesion from the image; (b) 16 features based on size and shape of the lesion are computed; (c) a neural classifier merges the features into an estimated likelihood of malignancy. A data set of 226 massive lesions (109 malignant and 117 benign) has been used in this study. The system performances have been evaluated in terms of the receiver-operating characteristic (ROC) analysis, obtaining A(z) = 0.80 +/- 0.04 as the estimated area under the ROC curve. (c) 2006 Elsevier B.V. All rights reserved. RI Retico, Alessandra /I-6321-2012
An automatic system to discriminate malignant from benign massive lesions on mammograms
DELOGU, PASQUALE;FANTACCI, MARIA EVELINA;
2006-01-01
Abstract
Mammography is widely recognized as the most reliable technique for early detection of breast cancers. Automated or semi-automated computerized classification schemes can be very useful in assisting radiologists with a second opinion about the visual diagnosis of breast lesions, thus leading to a reduction in the number of unnecessary biopsies. We present a computer-aided diagnosis (CADi) system for the characterization of massive lesions in mammograms, whose aim is to distinguish malignant from benign masses. The CADi system we realized is based on a three-stage algorithm: (a) a segmentation technique extracts the contours of the massive lesion from the image; (b) 16 features based on size and shape of the lesion are computed; (c) a neural classifier merges the features into an estimated likelihood of malignancy. A data set of 226 massive lesions (109 malignant and 117 benign) has been used in this study. The system performances have been evaluated in terms of the receiver-operating characteristic (ROC) analysis, obtaining A(z) = 0.80 +/- 0.04 as the estimated area under the ROC curve. (c) 2006 Elsevier B.V. All rights reserved. RI Retico, Alessandra /I-6321-2012I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
