Extracting pathologic patterns from NIR breast images with digital image processing techniques

Abstract

With a method of digital image processing including a variety of units, e.g., Laplacian filters, rank value filters, edge detectors, gain units and summation units, some pathologic patterns can be extracted from near-infrared (MIR) breast images. These pathologic patterns express the structural and the vascular information of mammary tissue. The information is concealed in primary NIR breast images, and not appear in common condition. The analysis by these pathologic patterns combined with the primary images, the possibility of false positive and false negative diagnosis is reduced, as results, and the diagnosis on breast diseases is improved obviously. © Springer-Verlag 2004.