Uso de correlações entre temperaturas superficiais e qualificação da técnica de imagens infravermelhas para a detecção e localização de tumores de mama

Abstract

Breast cancer is the most frequently diagnosed cancer type among women, excluding non-melanoma skin cancers. It is estimated that 59 700 women are likely to be diagnosed with breast cancer in the Brazil in 2018. Mammography is the main screening tool for the diagnosis of the disease. However, there are still several limitations to the examination, making it impossible to diagnose tumors for a large part of the population. The objective of this Thesis is to develop a procedure for the detection and location of a tumor present internally in a breast from infrared images. The proposed methodology is based on the use of superficial temperature correlations of breast skin to detect the heat source emitted by a tumor. The influence of thermophysical properties and parameters are analyzed, showing that the prior knowledge of the tumor characteristics is not necessary for the detection of its center. Initially, the estimation of cancer localization was performed using simulated thermography in the commercial software COMSOL from a three-dimensional hemispherical breast geometry for six different cases. For the simulated 3D cases where the tumor was positioned in the center of the breast and at different depths, the inclusions detected were accurately obtained. For the simulation where the tumor was inserted decentralized, the estimate was accurate, with a maximum error of 2 %. Experimental analysis was performed using an infrared camera to acquire thermographs from phantom tissues of the silicone. In two experimental cases where the tumor was positioned in the center and at different depths, the detection of the heat source center was obtained with only 3.8 % error. For the case where the tumor was decentralized, the error in the estimate was 4.5 %. Other numerical estimates were made using a real two-dimensional breast geometry and a three-dimensional geometry obtained by digital scanning of a mammary model, thus showing the precision of the technique in estimating tumors in irregular geometries. The technique shows great potential for in vivo applications and can become a great tool to increase the specificity of thermal images from infrared cameras.