Texture analysis of breast thermogram for differentiation of malignant and benign breast

Abstract

In this paper, we developed a new local texture feature extraction technique, called block variance (BV), for texture analysis in the thermal breast image. Then, present a method based on the different features extracted from the texture image obtained using BV to differentiate the malignant breast thermograms from the benign breast thermograms. Variance is an established measure of contrast in the image. Block variance (BV) uses the local variation of intensities to identify the contrast-texture in the gray-scale thermal breast image. Asymmetric temperature distribution between right and left breast in thermal breast image is an indicator of the presence of abnormality. Thus, we investigate the potential of our proposed features in asymmetry measure. For our experiment purpose, we used a set of forty malignant and sixty benign thermal breast images of DMR database. A feed-forward neural network (FANN) with gradient decent training rule has been employed to evaluate the classification performance. The effectiveness of our proposed features is compared against a feature set derived by Acharya et al. [16] in terms of classification accuracy, sensitivity, and specificity. From the experimental results, it is shown that the proposed features perform better compared to Acharya et al. features in differentiating malignant breast thermograms from benign breast thermograms.