A Fiducial-Based Tangible User Interface for White Matter Tractography Steven R. Gomez, Radu Jianu, and David H. Laidlaw Department of Computer Science Brown University {steveg,jr,dhl}@cs.brown.edu Abstract. We present a method for interacting with 3D brain tract visualizations using a webcam and a fiducial marker that can be con- structed cheaply in any home or office. Our contributions are a fiducial- based tracking architecture in the context of white matter tractography, and a preliminary evaluation with domain scientists providing usability and design insights. Expert feedback indicates that model positioning in our system is easier than in previous methods using traditional input devices or two-dimensional input interfaces, and that tract selection may be faster to execute using our tool, given training and practice. 1 Introduction Scientists can now explore the shape and connectivity of fibrous tissues, such as muscle and brain white matter, through visualizations of Diffusion Tensor Magnetic Resonance Imaging (DTI) data that commonly render variations of streamlines, such as streamtubes and hyperstreamlines, in 3D. In the case of the brain, these streamline models are visually dense as a consequence of the brain’s complex circuitry. As a result, typical interactions with white matter tracts, such as bundle selection or inspection of the model, may be difficult for a user to perform with conventional interface tools. In this paper, we present a new method for interacting with neural fiber tracts using a computer vision-based interface that allows for intuitive manipulation of the DTI model. We use fiducial tracking to position the brain and perform 3D selection of fiber tracts. In lieu of a typical interface, e.g. keyboard and mouse, or specialized input devices, the user holds and moves a homemade marker object in front of a webcam to manipulate the model. This marker can be constructed inexpensively from a pattern and common household materials. In our experiments, we con- structed a cardboard cube and decahedron whose faces are covered with paper Augmented Reality (AR) patterns, which can be printed on any printer. Figure 1 shows a user interacting with a DTI brain model using our system. We have obtained feedback from experts in an anecdotal study for an initial prototype. Results suggest that this type of lightweight 3D interaction has the potential to enable faster interaction with dense fiber tract collections. Expert feedback indicates that our new representation is more intuitive – and may be easier to use and learn – than conventional DTI model interaction methods.