Curvature statistics: An underused measure for epithelial to mesenchymal transition classification in pancreatic cancer

Epithelial to mesenchymal transition (EMT) and mesenchymal to epithelial transition (MET) plays an essential role in the local progression and metastasis of pancreatic cancer, whereby malignant cells undergo a physical transformation associated with increased mobility and resettle in a new site. Discovering EMT and MET’s underlying molecular processes have opened new therapeutic agents, however knowledge of this complex network is not complete and design consistencies in literature limit inferences and prevent meta-analysis of data. Instead of focusing on chemical pathways, we revisit ways to characterize cellular shape through cellular curvature distribution. We demonstrate this measure can improve classification models of epithelial, mesenchymal, and transitioning cells, with pancreatic cancer cells by computational analysis of optical microscopy images, which can improve risk stratification and treatment decisions.