Radar-based microwave imaging shows promise as a complementary technique to assist with diagnosing breast cancer. However, the recorded signals are dominated by a reflection from the skin, dwarfing reflections resulting from internal variations such as tumours. Therefore, suppression of the skin response is a key element in microwave image formation. This work combines neighbourhood selection and skin dominant window identification techniques with an adaptive filtering algorithm, resulting in a skin suppression method that is robust to variations in both geometry and dielectric properties. The algorithm was tested on four realistic 3D numerical models of varying complexity and was proven effective at suppressing the skin response while preserving the tumour response in all cases.
展开▼