Three-Dimensional Bioelectrical Impedance Spectroscopic Tomography for Visualization of Breast Tumor

Visualization of breast tumors with 3-D bioelectrical impedance spectroscopic tomography (3D-BIST) is proposed to identify the position, size, benignity, and malignancy of breast tumors. First, to visualize in 3-D, a conical sensor is developed based on the actual shape of the breast. Second, to confirm the feasibility of the proposed method, finite element simulation is employed. The influence of different impedance spectroscopy data acquisition methods (0-, 1-, and 3-skip) and different classifiers [Linear-SVM, radial basis function-support vector machines (RBF-SVM), and k-nearest neighbors (KNN)] on the classification results are explored in the simulation. The simulation results show that the 3D-BIST method allows visualization of breast tumor types; the image correlation coefficient (ICC) of imaging results at each position is larger than 0.7 when the breast tumor diameter
${d} \ge6$
mm; the best acquisition method for impedance spectroscopy data is 3-skip acquisition; and the best classifier is Linear-SVM classifier. The accuracy of the verification set calculated by fivefold cross-validation achieved 93.5%. The accuracy of identifying benign and malignant breast tumors with
${d} \ge8$
mm exceeded 89.0% on the test set. Finally, the proposed method was experimentally validated by using three kinds of biological tissues in place of normal breast tissue, benign breast tumor, and malignant breast tumor. In conclusion, the 3D-BIST method enables 3-D visualization of breast tumors and identification of breast tumor types. The proposed method has the advantages of noninvasive and non-radiation, which is expected to be applied to the detection of breast tumors.