A signature-based model for learning lung cancer stage from multiplex immunofluorescence image data with spatial summary functions

This study investigates the use of signatures as features within a classification model designed to determine lung cancer stages from multiplex immunofluorescence image data with their spatial summary functions. Precisely, we evaluated the performance of six different classifiers: Linear Discriminant Analysis, AdaBoost, Logistic Regression, Decision Tree, Bagging, and K-Nearest Neighbors. We used different features, including the raw functional data and signatures. Our findings indicate that signatures have potential as valuable features in medical and cancer diagnosis classification tasks.