Peningkatan Citra Termogram untuk Klasifikasi Kanker Payudara berbasis Adaptive Neuro Fuzzy Inference System (ANFIS)

Abstract

The application of pattern recognition is related to bioinformatics of medicine, image pattern recognition of illness or analysis of disease. The aim of this study is to measure the accuracy of thermogram images using Adaptive Neuro Fuzzy Inference System (ANFIS) method with and without image processing. Image processing has several steps technique. First step is image pre-processing with wiener filter, histogram equalization, and region growing methods. The second step of image processing is statistical feature extraction. Several values extracted from thermograms. The last step is classification by ANFIS method. This study uses 60 breast thermogram samples with Fluke Ti20 Thermal Camera as acquisition. These samples divided into three classes that are normal thermogram, early breast cancer thermogram, and advanced breast cancer thermogram. From this research that has been done can be proved that ANFIS method without image processing giving an error value 0,6395 in the influence range 0.5 and decreased error value 0,4199 with image processing method in the same influence range.