Classification of skin cancer based on fluorescence lifetime imaging and machine learning

To evaluate the development stage of skin cancer accurately is very important for prompt treatment and clinical prognosis. In this paper, we used the FLIM system based on time-correlated single-photon counting (TCSPC) to acquire fluorescence lifetime images of skin tissues. In the cases of full sample data, three kinds of sample set partitioning methods, including bootstrapping method, hold-out method and K-fold cross-validation method, were used to divide the samples into calibration set and prediction set, respectively. Then the binary classification models for skin cancer were established based on random forest (RF), K-nearest neighbor (KNN), support vector machine (SVM) and linear discriminant analysis (LDA) respectively. The results showed that FLIM combining with appropriate machine learning algorithms can achieve early and advanced canceration classification of skin cancer, which could provide reference for the multi-classification, clinical staging and diagnosis of skin cancer.