Ultrasonography is one of the most useful diagnostic tools for human soft tissue and is one of the methods that are in routine use for distinguishing benign and malignant breast tumors. But its diagnosis is operator dependent. In previous researches texture analysis fo More
Ultrasonography is one of the most useful diagnostic tools for human soft tissue and is one of the methods that are in routine use for distinguishing benign and malignant breast tumors. But its diagnosis is operator dependent. In previous researches texture analysis for solid breast mass classification is used. In those works texture features of the tumor are used, but sonologists notice to the features of the surrounding area of the tumors for their diagnosis. In this research as well as the morphological features of the mass the features of the surrounding area of the mass are also considered. MLP neural network is used for classification. 36 breast sonography images are used that 18 of them proved to be benign and 18 of them proved to be malignant through biopsy. The features are used in different combinations and it is shown that using the texture features of behind the tumor area and the same depth near the tumor provide meaningful result and also compensate the different adjustments of the systems.
Manuscript profile
