Computer-Aided Detection of Microcalcifications in Digital Mammograms to Support Early Diagnosis of Breast Cancer

Abstract

AbstractEarly detection of microcalcifications in mammograms is considered one of the best tools to prevent breast cancer. Although traditionally this task have been performed with analog mammograms, digital mammograms are currently an alternative for examination of breast to detect microcalcifications and any other kind of breast abnormalities. Digital mammography presents some advantages in comparison to its analog counterpart, such as lower radiation dosage for acquisition and possibility to storage for telemedicine purposes. Nevertheless, digitalization entails loss of resolution and difficulties to detect microcalcifications. Therefore, several methods based on digital image processing have been proposed to perform detection of microcalcifications in digital mammograms, to support the early detection and prognosis of breast cancer. However, sometimes computer-aided methods fail due to the characteristics of certain microcalcifications that are hard to detect either by visual examination and by computerized analysis. For this reason, this work presents a method based on contrast enhancement and wavelet reconstruction oriented to increase the rate of computer-aided detected microcalcifications. The images correspond to the mini-MIAS database, which provides mammograms of healthy women and with breast microcalcifications, including the respective coordinates of their locations. The work includes also the application of the method in resolution-enhanced mammograms via sparse representation, with the aim to determine the role of resolution enhancement for a possible improvement in the performance of the method.KeywordsContrast EnhancementSparse RepresentationTrue Positive RateDigital MammographyResolution EnhancementThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.