Contra Cancrum: Segmentation of Different Tissue Types in Lung Tumors from PET/CT Follow-Up Scans

The European research project ContraCancrum (Clinically Oriented Translational Cancer MultilevelModelling) aims at developing a multi-level platform for simulating malignant tumor development and tumor as well as normal tissue response to therapy. Lung cancer is one of the applications, which is targeted first. In the complex multi-level processing chain of ContraCancrum image processing (registration as well as segmentation) is a key step. Image processing is necessary to adapt the general tumor model of ContraCancrum to an individual patient with an individual tumor. Image processing thus provides the input necessary for a spatially resolved and patient specific simulation. In this paper we describe the methods used for registering morphological (CT) with functional (PET) images of the tumor as well as registering a time series of multimodal images before, during, and after therapy. Furthermore, we describe how the functional information can be used to segment tumors into regions with different metabolic properties.