Tumor size classification of breast thermal image using fuzzy C-Means algorithm

Breast cancer was a disease with the condition of the breast tissue became abnormal due to the development of cancer cells in the breast area. One method of breast cancer nondestructive detection was by through shooting the indicated breast cancer by using an infrared camera. The emission variations of infrared radiation on the image captured showed the level of cancer. The results of infrared camera imaging called as thermography image was processed in computing algorithm to classify the cancer in breast areas according to the characteristics of each image. The image feature extraction was obtained through the calculation of fractal dimension of the image by using the box counting algorithm. Image classification process was done by using the Fuzzy C-Means algorithm to determine the level of the breast cancer size based on the T component of the TNM system, namely T0, T1, T2 and T3 to the 22 image data to obtain the value of parameter cluster centers in Fuzzy C-Means. The test results showed that the feature extraction of breast thermography image using box counting fractal method gave the different value between normal breast and inflammatory cancer breast tissues. Normal breast tissue (T0) had a fractal dimension average less than T1, there was 1.161525 with deviation standard value was 0.593625. Breast with tumor T1 had a fractal dimension average less than T2, there was 1.45455 with deviation standard value was 0.4645. Breast with tumor T2 had a fractal dimension average less than T3, there was 1.6596 with deviation standard value was 0.2925, and breast with tumor T3 had a fractal dimension average 1.81294 with deviation standard value was 0.20199. The classification of tumor size using Fuzzy C-Means in 3 and 4 clusters with the use of 64×64 pixel box size in box counting process was more consistent than the use of 32×32 pixels box size.