Breast tumor localization using skin surface temperatures from a 2D anatomic model without knowledge of the thermophysical properties

Breast cancer is the second most common type of cancer among women after nonmelanoma skin cancer. Use of mammography, the main method to diagnose the disease, has several limitations in parts of the population. The primary goal of this work was to detect and localize the geometric centers of mammary tumors using only superficial temperatures of the breast skin. The 2D anatomic geometry of the breast was simulated using the commercial software COMSOL to obtain the distribution of skin temperature in the three main types of breast cancer. Random errors of +/- 2% were added to the simulated temperatures. The temperature variation caused by each type of cancer on the healthy tissue was correlated with auxiliary temperature profiles. These auxiliary temperature profiles were obtained with no prior knowledge of the thermophysical properties of the tumor apart from the mean values for thermal conductivity and blood perfusion of the layers of healthy breast tissue. The results showed that the maximum error for geometric center estimation was 0.32 cm for invasive lobular carcinoma, with a diameter of 1 cm, positioned 5 cm from the skin surface. Thus, this work contributes to studies aiming to improve the use of infrared thermography for early breast cancer diagnosis, as the results showed that localization of tumors using only superficial temperature profiles does not require prior knowledge of the thermophysical properties of the tissues. (C) 2019 Elsevier B.V. All rights reserved.