Breast tumor diagnosis system using three dimensional ultrasonic echography

Abstract

This paper describes a breast tumor diagnosis system using three dimensional ultrasonic echographic images. Since a malignant tumor is characterized by the morphological surface roughness, it is significant for the diagnosis to extract the exact boundaries of the tumor and show the three dimensional structure of the tumors. The proposed system consists of three dimensional image capturing system and a fuzzy reasoning based algorithm for the extraction of breast tumors. The method the authors proposed in this paper classifies all the voxels as one of "tumor", "normal tissue" or "boundary" by employing fuzzy reasoning and relaxation techniques. In order to evaluate the surface roughness, the authors define a parameter of a ratio of the surface area over the volume of the extracted tumor. The results for the clinical cases of three malignant and two benign tumors are successfully obtained by the system using an ultrasonic mechanical sector scanning probe of 10 MHz. It is found that the differences in the surface roughness between both types of tumors are clearly evident using a rendered surface image of the extracted tumor, and the average value of the defined parameter for the above three malignant tumors is 9.6, and 3.8 for the benign tumours.