Visual analytics for the exploration of multiparametric cancer imaging

Tumor tissue characterization can play an important role in the diagnosis and design of effective treatment strategies. In order to gather and combine the necessary tissue information, multi-modal imaging is used to derive a number of parameters indicative of tissue properties. The exploration and analysis of relationships between parameters and, especially, of differences among distinct intra-tumor regions is particularly interesting for clinical researchers to individualize tumor treatment. However, due to high data dimensionality and complexity, the current clinical workflow is time demanding and does not provide the necessary intra-tumor insight. We implemented a new application for the exploration of the relationships between parameters and heterogeneity within tumors. In our approach, we employ a well-known dimensionality reduction technique [5] to map the high-dimensional space of tissue properties into a 2D information space that can be interactively explored with integrated information visualization techniques. We conducted several usage scenarios with real-patient data, of which we present a case of advanced cervical cancer. First indications show that our application introduces new features and functionalities that are not available within the current clinical approach.