| Summary: | 碩士 === 國立中正大學 === 資訊工程研究所 === 92 === In the last decade, ultrasound imaging plays an important role in the field of breast cancer diagnosis because of its convenience and non-invasive. On the other hand, ultrasound is highly operator-dependent and some oversights would happen due to the tiredness of the physician or the poor image quality. Recently, the development of computer-aided diagnosis (CAD) provides a convenient way for doctors in detecting breast cancer using ultrasound images. However, the previous CAD systems have some limits with the requirements of human intervention. Hence, in this paper a novel automatic CAD system is proposed to find the suspicious frames among the whole breast ultrasound images. At first, the subsamping procedures for time-reducing are applied due to the huge amounts of whole breast US images. Then, the mammary zone where the tumors are usually located is identified through an automatic decision method. The poor image quality due to the speckle noise would be improved by using the anisotropic diffusion filter, and the image information would be enhanced by using the stick operator and the gray-level slicing technique. After applying the watershed segmentation, the suspicious segmented regions can be identified through several criteria defined according to the statistic and geometric features of a tumor. Also, a special feature, the coronal view of a suspicious region, will be used to improve the detection accuracy. By examining the US test database consisted of 13 cases, almost all the tumors and cysts could successfully be detected when an average of two false positive for each case is produced. The experimental results prove the accuracy of this purposed system.
|
|---|
