ACO classification of thermogram symmetry features for breast cancer diagnosis

Abstract

Breast cancer is the most commonly diagnosed form of cancer in women. While mammography is the standard modality for detecting breast cancer, it has been shown that medical thermography, which uses cameras with sensitivities in the thermal infrared, is also well suited for the task, especially in dense tissue and for small tumors. In this paper, we present an approach of analysing breast thermograms that first extracts a series of image features describing bilateral (a)symmetry between left and right breast regions, and then uses these features in a subsequent classification step. For the classification, we employ an ant colony optimisation based pattern recognition algorithm that is shown to provide a concise rule base with good classification performance.