Semiautomatic Transfer Function Initialization for Abdominal Visualization Using Self-Generating Hierarchical Radial Basis Function Networks

As being a tool that assigns optical parameters, used in interactive visualization, Transfer Functions (TF) have very important effects on the quality of volume-rendered medical images. Unfortunately, finding accurate TFs is a tedious and time-consuming task because of the tradeoff between using extensive search spaces and fulfilling the physician's expectations with interactive data exploration tools and interfaces. Therefore, it is necessary to integrate different features into the TF without losing user interaction. By addressing this problem, we introduce a semiautomatic method for initial generation of TFs. The proposed method uses a fully Self-Generating Hierarchical Radial Basis Function Network (SEG-HRBFN) to determine the lobes of a Volume Histogram Stack (VHS) which is introduced as a new domain by aligning the histograms of the image slices of a CT/MR series. The new self-generating hierarchical design strategy applied on RBFN allows for recognizing suppressed lobes corresponding to suppressed tissues in VHS and also for representing the overlapping regions which are parts of the lobes but cannot be represented by the Gaussian bases associated to the lobes due to the overlapping. Approximation with a minimum set of basis functions using SEG-HRBFN provides the possibility of selecting and adjusting suitable units to optimize the TF. The proposed method allows the integration of spatial knowledge, local distribution of the tissues, and their intensity information into the TF while preserving the user control. Its applications on different CT and MR data sets show enhanced rendering quality in abdominal studies.