期刊名称：International Journal of Advances in Soft Computing and Its Applications
印刷版ISSN：2074-8523
出版年度：2019
卷号：11
期号：1
页码：64-80
出版社：International Center for Scientific Research and Studies
摘要：Doctor’s assessment on breast cancer by using ultrasound imagesis time consuming and may produce difference interpretation ofdiagnosis results depending on several factors such as experience,fatigue and human error. Therefore, a computer system is needed toresolving difference interpretation of doctor’s assessment. This studyaims to develop a method for classifying breast cancer based on edgecharacteristics using 84 breast ultrasound images. Geometric andstatistical features are used in this study to extract edgecharacteristics of breast cancer. The results show that the proposedmethod obtains good performance with accuracy of 90.48%,sensitivity of 92.50%, specificity of 88.64%, positive predictive valueof 88.10% and negative predictive value of 92.86%. These resultsindicate that geometric and statistical features have goodperformance for identifying edge characteristic of breast cancer.
关键词：ultrasound images; breast cancer; edge characteristics; MLP.
