This work was created by Lucas C. Mendes, Erick O. Rodrigues, Sandro C. Izidoro, Aura Conci and Panos Liatsis and published in the 27th International conference on Systems, Signals and Image Processing (IWSSIP)
This work proposes the use of Genetic Algorithms (GA) to identify the area of the breast from the background in thermographic breast images. The proposed method uses color information, a fitness function based on cardioids, and GA. This is the first work in the literature to propose a Region of Interest (ROI) extraction based on GA and cariods. ROI extraction can improve the accuracy of cancer detection and assist with the standardization of acquisition protocols. The method is able to successfully separate the breast region in 52 out of 58 images, while being fully automatic, and not requiring manual selection of seed points.
This work proposes a ROI extraction methodology for ther-mographic breast images based on genetic algorithms. Weexploit the shape of cardioids jointly with gray level data todefine a fitness function that is evaluated by a genetic algorithmover time. Our method does not require manual placement ofseed points and is therefore entirely automatic, in contrast toother works in the literature .
The proposed algorithm was able to provide perfect ROIextraction in 16 out of 58 images and satisfactory results ina further 36 cases. In the case of perfect ROI extraction, weconsider a tight and correct ROI extraction with no exclusionof critical breast parts. We consider instances of minor or verylittle exclusion of critical parts in the breasts as satisfactory results. The remaining 6 cases provided poor results, where thecardiod ended up being displaced over the arm of the patientor another unrelated area.
In terms of time evolution, the GA showed better resultsafter 50 generations, requiring approximately 60 seconds toidentify the ROI. Results tend to be better with healthy womenas gray level information is more homogeneous. In contrast,sick women (mainly those who have been sick for a long timeand with more severe instances of disease) present substantiallyless homogeneous gray level intensity patterns
