Shape factor analysis of progressive rat hepatoma images.

We explore the hypothesis that shape factors can be used as tools to probe disease progression in addition to being used for simple classification. We define and apply a new shape factor to digital images of tumor cross sections of progressive rat hepatoma. Using a number of standard pathological measures, each image is also associated with a "disease time," a continuous variable between 0 and 1 with 0 being disease initiation and 1 being a near-fatal condition. The images are converted to data files that represent the 2-D shapes of the tumors. Their shape factors are then determined and plotted versus disease time. The results show that the shape factor can indicate and localize the tumor structural phase transition that occurs between uniform growth and infiltration.