Comparing Cross-Sections and 3D Renderings for Surface Matching Tasks Using Physical Ground Truths

Within the visualization community there are some well-known techniques for visualizing 3D spatial data and some general assumptions about how perception affects the performance of these techniques in practice. However, there is a lack of empirical research backing up the possible performance differences among the basic techniques for general tasks. One such assumption is that 3D renderings are better for obtaining an overview, whereas cross sectional visualizations such as the commonly used Multi-Planar Reformation (MPR) are better for supporting detailed analysis tasks. In the present study we investigated this common assumption by examining the difference in performance between MPR and 3D rendering for correctly identifying a known surface. We also examined whether prior experience working with image data affects the participant's performance, and whether there was any difference between interactive or static versions of the visualizations. Answering this question is important because it can be used as part of a scientific and empirical basis for determining when to use which of the two techniques. An advantage of the present study compared to other studies is that several factors were taken into account to compare the two techniques. The problem was examined through an experiment with 45 participants, where physical objects were used as the known surface (ground truth). Our findings showed that: 1. The 3D renderings largely outperformed the cross sections; 2. Interactive visualizations were partially more effective than static visualizations; and 3. The high experience group did not generally outperform the low experience group.