Speaker
Description
In breast cancer studies, to discriminate between malignant and benign lesions, computer-aided diagnosis (CAD) system has been used to diagnose breast cancer based on morphological characteristics. We propose a method to non-invasively distinguish malignant or benign microcalcifications using only mammographic image based on dual-energy method [1, 2]. In this study, a photon-counting spectral mammography system was simulated using Geant4 Application for Tomographic Emission (GATE) simulation tools. The dual-energy images were acquired using two energy bins. Microcalcifications were used as type I (calcium oxalate) and type II (calcium hydroxyapatite). For statistical analysis, the microcalcifications were classified as calcium hydroxyapatite or calcium oxalate based on a score calculation using the dual-energy images. The score values were calculated using the ratio values at high energy to low energy because there is less attenuation difference in the high energy region and a large attenuation difference in the low energy region. We confirmed that the contrast and noise were influenced because the classification method used in this study was based on the pixel values of the images. Therefore, we also calculated the score as the difference between the two types of microcalcifications using the dual-energy subtraction method [3]. Because of the improved contrast of microcalcifications, the classification performance was better in the dual-energy subtracted images. In addition, this study suggested the ability to automatically classify microcalcifications using segmentation methods and minimum and maximum threshold of score values. These results demonstrated the possibility of classifying microcalcifications based on spectral mammography to improve the diagnostic accuracy of breast lesions.
