Automated detection of breast tumors in ultrasonic images using fuzzy reasoning

This paper proposes a novel method for the automated detection of breast tumors in three dimensional ultrasonic images using fuzzy reasoning. The algorithm consists of two stages of processing. In the first stage a three dimensional LoG (Laplace of Gaussian) filtering is employed to generate a membership function for the fuzzy inference. In the second stage, each voxel in the three dimensional image is classified as “tumor”, “normal tissue” or “boundary” by fuzzy reasoning followed by the defuzzification using relaxation techniques. We apply the method to 10 cases of malignant and 10 cases of benign tumors. Both of the two types of extracted tumors agree well with the manually traced boundaries. Further we define a parameter for evaluation of the uneven complexity of the surface roughness for diagnosis of malignant and benign tumors. The results show a successful characterization