Polarimetric imaging combining optical parameters for classification of mice non-melanoma skin cancer tissue using machine learning

Abstract

Polarimetric imaging systems combining machine learning is emerging as a promising tool for the support of diagnosis and intervention decision-making processes in cancer detection/staging. A present study proposes a novel method based on Mueller matrix imaging combining optical parameters and machine learning models for classifying the progression of skin cancer based on the identification of three different types of mice skin tissues: healthy, papilloma, and squamous cell carcinoma. Three different machine learning algorithms (K-Nearest Neighbors, Decision Tree, and Support Vector Machine (SVM)) are used to construct a classification model using a dataset consisting of Mueller matrix images and optical properties extracted from the tissue samples. The experimental results show that the SVM model is robust to discriminate among three classes in the training stage and achieves an accuracy of 94 % on the testing dataset. Overall, it is provided that polarimetric imaging systems and machine learning algorithms can dynamically combine for the reliable diagnosis of skin cancer.