| Summary: | 碩士 === 國立臺灣大學 === 資訊工程學研究所 === 100 === Breast cancer is currently the most common diagnosed cancer among women and a leading cause of death. Breast density is considered a significant risk factor and it is the strongest correlation with relative risk of developing breast cancer. Nowadays, the most common approaches to estimate density are using mammographic images so that mammography has been applied as a standard reference for breast abnormalities. However, the estimates of breast density may vary significantly, in part, due to the projective nature of mammography. Therefore, ongoing studies using three-dimensional 3-D computed tomography (CT) scans are evaluating quantitatively breast density for achieving potentially more consistent breast density analysis.
In this paper, we propose to assess the densities from images of two different modalities; CT and mammography, for the same patient. We use chest body model approach for template-based breast CT segmentation method based on non-rigid image registration and a fuzzy C-mean clustering for breast densities analysis. All the CT scans, in this study, are low-dose chest scans. For the breast density estimation of the two-dimensional (2-D) and 3-D mammography images are used different software programs. The assessment of the correlation between the mammographic and CT densities was performed by computing the Pearson''s correlation coefficient.
The analysis of breasts scans from 12 patients shows that the results from both modalities have high positive correlations. Although mammography has been applied as a standard reference, the results from this study demonstrate good potential for density assessment with CT x-ray and that low-dose chest CT, commonly used for lung cancer detection, could be also applied for breast density estimation.
|
|---|
