The combination of volume data acquired by multiple modalities has been recognized as an important but challenging task. Modalities often differ in the structures they can delineate and their joint information can be used to extend the classification space. However, they frequently exhibit differing types of artifacts which makes the process of exploiting the additional information non-trivial. In this paper, we present a framework based on an information-theoretic measure of isosurface similarity between different modalities to overcome these problems. The resulting similarity space provides a concise overview of the differences between the two modalities, and also serves as the basis for an improved selection of features. Multimodal classification is expressed in terms of similarities and dissimilarities between the isosurfaces of individual modalities, instead of data value combinations. We demonstrate that our approach can be used to robustly extract features in applications such as dual energy computed tomography of parts in industrial manufacturing.

It is a difficult task to design transfer functions for noisy data. In traditional transfer-function spaces, data values of different materials overlap. In this paper we introduce a novel statistical transfer-function space which in the presence of noise, separates different materials in volume data sets. Our method adaptively estimates statistical properties, i.e. the mean value and the standard deviation, of the data values in the neighborhood of each sample point. These properties are used to define a transfer-function space which enables the distinction of different materials. Additionally, we present a novel approach for interacting with our new transfer-function space which enables the design of transfer functions based on statistical properties. Furthermore, we demonstrate that statistical information can be applied to enhance visual appearance in the rendering process. We compare the new method with 1D, 2D, and LH transfer functions to demonstrate its usefulness.

We define a transfer function based on the first and second statistical moments. We consider the evolution of the mean and variance with respect to a growing neighborhood around a voxel. This evolution defines a curve in 3D for which we identify important trends and project it back to 2D. The resulting 2D projection can be brushed for easy and robust classification of materials and material borders. The transfer function is applied to both CT and MR data.

Dense clinical data like 3D Computed Tomography (CT) scans can be visualized together with real-time imaging for a number of medical intervention applications. However, it is difficult to provide a fused visualization that allows sufficient spatial perception of the anatomy of interest, as derived from the rich pre-operative scan, while not occluding the real-time image displayed embedded within the volume. We propose an importance-driven approach that presents the embedded data such that it is clearly visible along with its spatial relation to the surrounding volumetric material. To support this, we present and integrate novel techniques for importance specification, feature emphasis, and contextual cutaway generation. We show results in a clinical context where a pre-operative CT scan is visualized alongside a tracked ultrasound image, such that the important vasculature is depicted between the viewpoint and the ultrasound image, while a more opaque representation of the anatomy is exposed in the surrounding area.