 Antoine Rosset, MD, (left) and Osman Ratib, MD, PhD, using freeware. |
The rapid evolution of medical imaging techniques and the increasing number of multidimensional and multimodality studies constitute a challenge for PACS workstations and image display programs. With the improvement of speed and resolution of multi-detector CT scanners and the emergence of multi-modality examinations like PET/CT, traditional two-dimensional image display programs are just not suitable for the interpretation of these large sets of images. The conventional way of reviewing these images slice-by-slice is too cumbersome for interpreting between 800 and 1,000 slices that can be routinely acquired today with multidetector CT scanners. These large sets of images require additional image processing and reformatting to be more suitable for efficient and rapid image navigation and image interpretation. In most cases this can be achieved only on high-end, dedicated 3D workstations that can provide thick-slab maximum intensity projections (MIP), orthogonal and oblique multiplanar reformatting (MPR), and three-dimensional volume and surface rendering. Multimodality examinations like PET/CT add another dimension to the data, requiring the viewers to be able to navigate interactively in four or five dimensions through the superposition and fusion of the volume of two different image modalities. These recent changes in the acquisition modalities require radiologists to use expensive dedicated 3D processing workstations to properly interpret these examinations. Furthermore, access to these visualization and rendering tools is usually limited to high-end users in a radiology department, preventing referring physicians, surgeons, and other care providers from benefiting by the extraordinary value of the multidimensional imaging techniques for decision-making and patient management.
Radiologists at UCLA have elected to develop a completely new software platform that will allow users to efficiently and conveniently navigate through large sets of multidimensional data without the need for high-end, expensive hardware or software. We designed and implemented a complete multidimensional image navigation program for the UNIX-based Macintosh Os X system. The program, called OsiriX, was designed based on a completely new architecture and user interface design. It is tailored for large sets of multidimensional and multimodality images, such as combined PET/CT studies, that require three-dimensional image fusion and volume rendering. The software is developed on a Macintosh platform taking advantage of the underlying UNIX kernel of the new Os X operating system. It also benefits from the extremely fast and optimized 3D graphic capabilities of Open GL graphic standard that is widely used for computer games and animations and is highly optimized on the Macintosh platform. The software was developed in Objective C using the well-known OpenStep development environment under the Cocoa/Xcode development suite provided by Apple. Because OpenGL and OpenStep are also available on other UNIX platforms, the OsiriX software could easily be ported to other UNIX platforms such as Linux, for example.
 The OsiriX viewer allows one to open any kind of DICOM images, including angiographies, PET/CT with image fusion, MRI, CT. |
CUSTOMIZED USER INTERFACE
In the design of the software, special attention was given to adapt the user interface to the specific complex tasks of navigating through large sets of image data. An innovative approach to interactive image navigation employing a standard multipurpose jog-wheel that is widely used in the video and movie industry was implemented. This device allows the users to navigate in the different dimensions of an image set much faster than with a traditional mouse or with on-screen cursors and sliders. The graphic user interface also uses the very intuitive and user-friendly Macintosh interface. Users can easily change and customize the software by adding and removing tools and items from the program toolbar and menu bars. This allows for adapting the software for very specific tasks that require only a limited number of functions, preventing the users from becoming overwhelmed by an excessive number of unnecessary tools and functions that they do not need. OsiriX provides a large number of image processing and image rendering tools based on the open source libraries ITK and VTK of image processing that are becoming widely used in the medical imaging community. The adoption of these libraries ensures that all new developments in image processing that could emerge from other academic institutions using these libraries could be directly ported to the OsiriX program.
An important and challenging aspect of the development of this project was integration of all these technologies together. OpenGL, VTK, ITK, DICOM Offis, Papyrus, and Quicktime are C/C++ cross-platform toolkits. The missing link is the framework that keeps all these components together within a simple and user-friendly graphic user interface. This is the reason why OsiriX was developed on a Macintosh platform to benefit from its well-known user-friendly interface features and convenience and ease of use.
The OsiriX software is also designed to benefit from multi-threading and parallel processing capabilities of the newest generation of Apple computers. In particular, all graphics manipulations and rendering functions are tilled in separate threads and distributed to each available processor when possible. The support of multi-processor architecture is seamless and performs particularly well on the new generation of dual processor-G5 computers.
Another potential improvement in performance expected for very computational-intensive image processing and rendering techniques would be achieved by implementing a grid computing technology federating several networked computers. Grid computing technology can significantly improve the performance of 3D imaging rendering techniques. The VTK library already supports grid computing for 3D rendering through the open-source and cross-platform standard Message Passing Interface (MPI) (ref: www.lam-mpi.org/ ). Apple Computer recently announced a new grid technology dubbed X-Grid (ref www.apple.com ) that will facilitate the management and configuration of a large cluster of computers. With the X-Grid technology, it will be easy to connect all computers of a radiology department to perform parallel rendering during processor idle time. In a large academic radiology department, it is common to have a relatively large number of computers that are only partially used and remain idle for significant portions of time. The grid computing technology allows taking advantage of the capacity of these idle computers for performing computational intensive tasks needed for 3D rendering applications across the network. By adopting the X-Grid technology, OsiriX, being already multi-thread compliant, could benefit from significant improvement in performance if used in a network of multiple computers.
OsiriX is provided free of charge and the source code is distributed under the GNUopen source licensing agreement: homepage.mac.com/rossetantoine/osirix .
Osman Ratib, MD, PhD, is vice chair of information systems, UCLA Radiological Sciences, David Geffen School of Medicine at UCLA.
Antoine Rosset, MD, is a radiologist and software developer from the University Hospital of Geneva and research fellow at UCLA.
