Interactive visualization of multiscale biomedical data: an integrated approach Debora Testi 1 , Gordon Clapworthy 2 , Stephen Aylward 4 , Alejandro Frangi 3 , Richard Christie 5 1 BioComputing Competence Centre, SCS srl, Italy 2 University of Bedfordshire, UK 3 Universitat Pompeu Fabra, Spain 4 Kitware Inc, USA 5 University of Auckland, New Zealand ABSTRACT Visualization of multiscale data is becoming increasingly important in all biomedical research projects. Huge amounts of data are being collected worldwide at all scale levels (from genes to body level). However, there is a clear lack of software tools to support the interactive inspection of these large data collections. A research consortium has been established to identify the challenges associated with real use case scenarios and to produce, in the next year, an open source library (MSVTK) to be used by any research project to effectively inspect and visualize multimodal, multidimensional and multiscale data. We will review here the challenges associated with the development of this visualization software library and the preliminary approaches under investigation. KEYWORDS: multiscale, biomedical data, visualization. INDEX TERMS: [Biomedical and Medical Visualization], [Data Fusion and Integration], [Multi-field, Multi-modal and Multi- variate Data], [Multidimensional Data] 1 INTRODUCTION Recent years have seen an increasing trend in biomedical research towards the consideration of systemic processes. These are phenomena which are commonly observed in living organisms and which reflect the outcome of interactions between multiple sub-systems. In the past, the complexity of these problems had forced researchers to concentrate on individual sub-systems, and the most common boundary separating these has been spatiotemporal scale. However, the biomedical community has recently started to produce the first results of the so-called integrative approach, and a number of multiscale datasets have become available. Their analysis has made evident that there is a shortage of appropriate tools for visualizing and exploring data that are defined across a broad range of spatial and/or temporal scales. At the same time, the number of biomedical problems that will demand multiscale visualization in the coming years suggests that this area should start to receive urgent attention but, surprisingly, it received almost no mention in the Visualization Research Challenges [1] document produced jointly by the USA National Institute for Health and National Science Foundation in 2006. Multiscale visualization is not a new visualization issue and it has been investigated in other scientific contexts, such as information visualization with, perhaps, the most relevant work being undertaken in geographical data visualization, as exemplified by well-known solutions such as Google Earth. While these approaches are extremely effective within the context of their specific target problem, not all of these solutions can be generalized to other domains. In particular, many of the approaches do not translate well to the biomedical area, where datasets are usually of higher dimension and contain a greater variety of field data. 2 THE MSV PROJECT Within this context, an international consortium has been established involving leading groups in visualization from three continents with the aim of providing solutions to these problems. The MSV 1 project, funded by the European Commission (EC) but also involving partners from the USA and New Zealand, will implement an open source software library for the multiscale visualization of, and interaction with, biomedical data, which will provide a suitable resource for the biomedical community to use in this rapidly evolving area. This library with be compatible with, and distributed alongside, the Visualization Toolkit (VTK) 2 . This objective will be achieved by: • clearly defining the problem and its challenges; • identifying exemplary problems and data to be used for the MSVTK testing; • defining the best practices to address this type of problems; • implementing the first open source version of the MSVTK library. 3 THE CHALLENGES The identification of the challenges began with an international consensus meeting involving representatives from both research and industry. From the meeting outcomes, a review of the state of the art, and the internal discussion, it was confirmed that there is a need for a specialized software library to interact and visualize multiscale biomedical data. To better identify real users’ needs in terms of multiscale data visualization, partners have contacted EC-funded projects and NA-MIC initiatives. From the collected inputs, it has become evident that the MSV project has to address a number of challenges that have been classified and that have to be taken into account in the design and development phases (an example is given in Figure 1). In detail, the MSV problem is defined by: • information which will be: - at very different spatial and temporal scales, ranging from the molecule (nanometers and nanoseconds) up to the body level (meters and years); - in different forms (medical images, computer models, signals, etc.); - of heterogeneous dimensionality (1D, 2D, 3D, 3D+t); 1 http://www.msv-project.eu 2 http://www.vtk.org