Image Transformation Based Detection of Breast Cancer Using Thermograms

In today’s health sector scenario, it is availed that breast cancer is the major cause of death among urban women. Every third woman is suspected to be a victim of breast cancer. The detection of tumor at an early stage can prevail a better chance of survival. Among different modalities, thermography is observed to be an early detector of tumor. It records the temperature of the surface and thus it helps in identifying the tumor due to the indication of hot colors identified in thermography. In this study, we propose an approach for detection of breast cancer using thermograms based on image transformation methods. The input images are considered from the database available at DMR, Visual Labs, Brazil. The texture variations are better captured through the different transform methods. Thus, to capture the variation in the texture caused by the tumor, the images are transformed using the three different image transformations methods viz. Wavelet transform, Curvelet transform and Lifting Scheme based Wavelet transform. Then features are extracted from the transformed images and ranked using different techniques. The final set of features are classified using different classifiers viz. Support Vector Machine, K-Nearest Neighbor, Naïve Bayes, Random Forest, and Multi-Layer Perceptron. Among the different classifiers, random forest gives the best accuracy with lifting scheme-based wavelet transform images of 93.89%.