Higher order spectra analysis of breast thermograms for the automated identification of breast cancer

Breast cancer is a leading cancer affecting women worldwide. Mammography is a scanning procedure involvingX‐rays of the breast. It causes discomfort and may cause high incidence of false negatives. Breast thermography is a new screening method of breast that helps in the early detection of cancer. It is a non‐invasive imaging procedure that captures the infrared heat radiating off from the breast surface using an infrared camera. The main objective of this work is to evaluate the use of higher order spectral features extracted from thermograms in classifying normal and abnormal thermograms. For this purpose, we extracted five higher order spectral features and used them in a feed‐forward artificial neural network (ANN) classifier and a support vector machine (SVM). Fifty thermograms (25 each of normal and abnormal) were used for analysis.SVM presented a good sensitivity of 76% and specificity of 84%, and theANN classifier demonstrated higher values of sensitivity (92%) and specificity (88%). The proposed system, therefore, shows great promise in automatic classification of normal and abnormal breast thermograms without the need for subjective interpretation.