Dipl.-Phys. Thomas Grunert, Dipl.-Phys. Frank Weisser, Dipl.-Math. Jürgen Wassermann, Dipl.-Phys. Edelhard Becker
We developed a PACS workstation with an integrated CASE-tool that consists of four major parts: A ready-to-use workstation provides user interface, base functionality and applications for a broad community of clinical users. Together with the development interface , it serves as a common framework for application development. A development toolbox and a sophisticated object class library allow for efficient implementation of research ideas and consistent integration of new applications into the PACS workstation. They also support the development of collaborative applications. A visual programming environment is just under development.
Images play an important role in medical diagnosis and therapy. The rapid development of digital acquisition technology has opened up new horizons in medical image processing (IP) and computer graphics (CG). New application areas have evolved during recent years, such as radiation treatment planning and the simulation of all kinds of surgery. Therefore, clinical researchers from a variety of disciplines are increasingly confronted with the problem of making efficient use of digital equipment. But progress is generally hampered by the lack of adequate software tools, that allow research ideas to be rapidly realized.
Clinical researchers have put a tremendous amount of effort into the development of new applications from the scratch, which led to a large diversity of applications. Through the usage of IP and CG libraries such costs could be reduced, while still remaining on a high level due to permanent re-development. Pure medical workstations like Analyze, on the other hand, offer a number of IP and CG methods under a more or less comfortable user interface. But they cannot easily be extended by new algorithms and applications since they support neither the consistent integration of new modules, nor their rapid development. A few clinically oriented research groups have proposed application systems with a focus on user interface design and functionality. Some of these groups try to achieve system extensibility by providing source code and/or function libraries for application development.
On the other hand, most PACS-frontends currently commercially available, focus on image management functionality. Since PACS customers are usually research-oriented institutions, there is an increasing demand for more complex applications in the field of IP and CG. Unfortunately, commercially available medical imaging workstations do not possess application development capabilities at all, or only very restrictive extension interfaces.
Figure 20:
Principle of integrating a new application into the PACS workstation.
Figure 19:
Architectural system overview of the MEDStation.
To overcome the shortcomings of the approaches mentioned above, we propose a research platform based on a turnkey application system (PACS workstation, the so-called MEDStation). This platform provides a clinically oriented user interface, base functionality and applications, addressing a broad community of clinical clients. The development interface is the second key element of our development system, that allows new applications to be consistently integrated into the MEDStation (see Fig.19 We are aware of the fact, that, especially for legal reasons, a clear distinction between routine and research equipment is desirable. The following observations, however, justify our concept of a research system with integrated routine features:
Many solution ideas expressed by clinicians focus on extensions to available methods.
Providing an environment similar to a routine system makes it easier to become familiar with the research platform.
Integrating novel applications under a common user interface increases consistency and avoids the confusion of using a variety of different interfaces.
The cooperation between different functions and applications within a common environment increases software reusability.
The application development is speeded up by a development toolbox that allows for rapid prototyping of new applications as well as their integration into the turnkey system. One of the most promising characteristics of the object-orientation paradigm seems to be its know-how transfer capability. Thus, the basis for both, the application system and the development toolbox of the proposed platform is a sophisticated object class library that consists of:
Objects interfacing the turnkey application system (management objects).
Objects for algorithmic research and
application development
(IP and CG).
The latter allow an easy and quick application development by providing structure and functionality of most of the real-world objects from the area of medical imaging and graphics. A consistent integration of new applications into the turnkey system may be performed by the use of management objects. These provide methods for user interface access as well as study handling and I/O control. In this manner, a new application may be realized by simply combining already existing objects, ideally by visual programming.
One of the primary goals of the system is to address a broad community of clinical clients. Therefore, an analysis of the requirements for a medical workstation has to cover various disciplines. We performed such an analysis at the University Hospitals of Tübingen. From the results we derived a functionality specification for a universal medical graphics workstation. We distinguished between six functionality categories:
Access:
Image storage/retrieval, data compression, interpretation of file
formats and communication (esp. ACR-NEMA, DICOM).
Presentation:
Study handling, multiple image display, layout
manipulation, handling of multiple layouts
(like panels of an alternator).
Manipulation:
Image processing operations (e.g. zoom,
pan, mirror, contrast/brightness adjustment,
negate, filter, arithmetics).
Evaluation:
Local/global greyvalue statistics and
geometric properties (2D/3D distance, angle, profile).
Documentation:
Image annotation and hardcopy.
Analysis:
Advanced image processing and graphics methods (e.g.
segmentation, registration, classification,
multiplanar reconstruction, volume rendering).
Especially in medical informatics the success of a new system largely depends on the adequacy of its man-machine interface. We used the lightbox characteristics as a guideline for the user interface design of our system. We extended the lightbox into the third dimension by attaching a whole study to a user defined place on the digital lightbox
Figure 21:
User interface presentation layer of the turnkey system
The whole workstation screen is used for image display. Only a small part is reserved for a menu field. In analogy to the observed characteristics of conventional lightboxes, the following look and feel is supported (see Fig. 21
The display area of the lightbox can be arbitrarily devided into sections.
An image or a patient study can easily be selected from the database panel and fixed to an arbitrary lightbox segment via a drag-and-drop mechanism.
Unrestricted manoeuvering on the lightbox panel is performed by realtime zoom and pan, thus controlling the panel section actually displayed on the screen and its spatial resolution.
The logical link between images and lightbox segments can be graphically reorganized on the lightbox editor.
Several Panels, each containing a lightbox with configurable layout, can be handled (alternator).
Although we based the design of our user interface on the lightbox metaphor we did not restrict ourselves to a pure 2D medium. Many additional features enhance the interpretation process, such as fast skimming through image series, arbitrary slicing through volume datasets, or volume rendering.
A medical imaging and graphics workstation suitable for clinical research must contain mechanisms for application development support and integration of new algorithms. In addition to the presented PACS workstation, we propose an elaborate extension interface. Figure 20 illustrates the mechanism of integrating a new application into the turnkey system.
From the programmers point of view, the class study plays A central role in the system (see Fig. 20 A patient study may contain textual elements (e.g. patient name, medical report, anamnesis), graphical objects, images, audio and video. The study provides methods for accessing, manipulating and displaying these elements and their hierarchical structure. Display management is performed by the turnkey system. The application programmer does not have to deal with this highly complex task. Furthermore studies provide mechanisms for event handling in order to maintain the dataflow through the application. To allow the communication between applications, a message mechanism in form of a mailbox system is integrated. The management of user input in the screen area is also done by the study, such as painting, picking and movements in 2D and 3D.
The application builder toolbox is another key element of our development system. It is based on an object class library which covers a broad spectrum of medical imaging and graphics functionality. The architecture of the class library is based on the client/server model and the data flow approach. Its structure represents a large amount of software development expertise in this area and, therefore, allows inexperienced application programmers to quickly achieve a professional program design: The CG and IP objects are wrapped into a communication shell, which allows the application programmer to simply connect the object instances to an object pipeline. The communication shell manages the processing of events in the pipeline (e.g. starting new data processing - all along the pipeline - after changing some parameters). Thus, it relieves the application programmer from the necessity to program all the event-handling by himself. The client/server architecture supports distributed computing. Finally, the server is reconfigurable at runtime, which allows it to be used as an interpreter for visual programming. A visual programming environment is under development.
Today, most teleradiology applications focus on image management functionality, like their conventional counterparts - the PACS-workstations. On the other hand, since PACS customers are usually research-oriented institutions, there is an increasing demand for more complex teleradiology applications.
In fact there are a great number of systems for computer-assisted radiology, but the experts still have to meet or work in a time-consuming asynchronous way. Unfortunately there is no system available for the development of collaborative applications.
The MEDStation is also a development system for collaborative applications in the area of computer assisted radiology. Key features are: a dynamic and self-defining communication protocol allowing dynamic applications without the need of the error-prone defining of a (fixed) protocol in header-files, an arbitrary number of participants in a session and an easy-to-use object class library for the building of applications.
As illustrated in fig. 22 the development system consists of five parts: (1) The previously described fully equiped PACS-workstation as the basis for the development system. (2) A sophisticated object-class library with objects for visualization and image-processing as well as management objects for the consistent integration of new applications, study handling, I/O a.s.o. The visualization and image-processing objects are wrapped into template based communication shells, so that they can be connected to form "living" pipelines. Intermediate complementary interface-objects form a layer between the "processing" objects and the user interface. These intermediate objects are designed to register their interface at the moment of instantiation, which forms a dynamic communication protocol in the dispatcher (3). The intermediate objects also manage external input or user input depending on whether he has the right of speech. Furthermore they insure the integrity of the user interface in case of the absence of the right of speech. The dispatcher (3) handles the registration and deregistration of the interface, forming the dynamic communication protocol and the I/O of the actions needed for the synchronisation of the cooperating systems. (4) The admission of - an arbitrary number of - participants to a session, the distribution of the actions to all participants of the session and the management of the right of speech is done by a central session manager. (5) Finally, an elaborate extension interface - also based on the communication shell architecture - offers the option to extend the functionality layer of our development platform by the consistent integration of new application modules.
Figure 22:
Architectural system overview of the development system.
Our evaluation results suggest that clinical research related to digital imaging and graphics benefits from the availability of the toolkit system in three ways: First, the path from a research idea to its implementation and evaluation is shortened. Second, novel applications are motivated by the system, and the realization of complex applications becomes possible. Third, the communication between clinical researchers and computer scientists is promoted by the platform. However, the high level of programming knowledge necessary to efficiently use our toolbox, has been an obstacle for addressing a broader community of clinical researchers. Excellent programming skills are not widely found in a clinical environment. A solution is a graphical programming interface, which is just under development.
The MEDStation is available for Unix (e.g. SGI, SUN, Linux, ...) and WindowsNT.
Miguel Encarnação