Microwave detection of breast tumors: comparison of skin subtraction algorithms

Early detection of breast cancer is an important part of effective treatment. Microwave detection of breast cancer is of interest due to the contrast in dielectric properties of normal and malignant breast tissues. We are investigating a confocal microwave imaging system that adapts ideas from ground penetrating radar to breast cancer detection. In the proposed system, the patient lies prone with the breast extending through a hole in the examining table and encircled by an array of antennas. The breast is illuminated sequentially by each antenna with an ultrawideband signal, and the returns are recorded at the same antenna. Because the antennas are offset from the breast, the dominant component of the recorded returns is the reflection from the thin layer of breast skin. Two methods of reducing this reflection are compared, namely approximation of the signal with two time shifted, scaled and summed returns from a cylinder of skin, and subtraction of the mean of the set of aligned returns. Both approaches provide effective decrease of the skin signal, allowing for tumor detection.