Interpretation Mueller matrix images based on polar decomposition and statistical discriminators to distinguish skin cancer

The Mueller matrix describes all the polarizing properties of a sample, and therefore the optical differences between cancerous and non-cancerous tissue should be present within the matrix elements. We present in this paper the Mueller matrices of three types of tissue; normal, benign mole, and malignant melanoma on a Sinclair swine model. Feature extraction is done on the Mueller matrix elements resulting An retardance images, diattenuation images, and depolarization images. These images are analyzed in an attempt to determine the important factors for the identification of cancerous lesions from their benign counterparts. In addition, the extracted features are analyzed using statistical processing to develop an accurate classification scheme and to identify the importance of each parameter in the determination of cancerous verses non-cancerous tissue.