This document summarizes a study that uses wavelet transform based texture features and image classification to analyze breast thermograms. The study compares separable and non-separable discrete wavelet transforms. Texture features are extracted from wavelet decomposed images of the pectoral regions of two breasts. Principal component analysis and an Adaboost classifier are then applied to evaluate classification performance. The results show that complex non-separable 2D discrete wavelet transform features perform better than real separable counterparts for classifying malignant versus non-malignant breast lesions.