Objective
A Picture Archiving and Communication System, supported by a multi-media medical image data-base, will be integrated with all other components of a Hospital Information System (HIS), resulting in a second generation distributed PACS architecture test bed.
A picture archiving and communication system (PACS) supported by a multimedia medical image database, is being integrated with all other components of a hospital information system (HIS), resulting in a second generation distributed PACS architecture test bed.
The research is in the stage of preprototyping. The first systems are being tested in the laboratories and prepared for integration into the clinical environment.
Initial efforts have been focussed on addressing integration issues. Integration of PACS into a hospital environment is a prerequisite for clinically useful systems, and involves the integration of all departmental systems, as well as the adoption of standards and coordination with CEN TC251.
These key objectives are in a stage of pre-prototyping. The first systems are being tested in the labs and prepared for the move to the clinical environment.
"Integration" has been focused during the first year, on the following items:
- Integration in a hospital environment stressing the need for having clinical useful systems.
- Integration with the other information systems; the departmental systems.
- Integration by means of standards and coordination with CEN TC 251.
These independent modules which are developed can be seen as an advanced integrated PACS environment, or can be used as single components contributing towards a multi-vendor PACS installation.
It is impossible giving a total overview of the activities, only some of the development were reflected given an idea of the status.
First version prototypes are working for:
- Integration of PACS-RIS-HIS
- Data Compression
- Network-integration and management
- Adaptive workstation
- Query optimalization and indexing by content tools
- Decision support tools
Different clinical sites are selected where prototype evaluation in clinical routine will be done.
The sites include clinical sites in: Aachen, Brussels, Graz, Pisa, Marburg, Rennen, Luxembourg.
For the interworking with the departmental systems a generic interface is being developed. The functional specifications were composed based on:
- The user requirements for this interface, including those for image migration,
- The clinical procedures around the PACS systems in Brussels, (Emergency/Orthopedics) and Neurosurgery), and Marburg (Neuroradiology)
- The functions of the interface,
- The triggering events and needed data to support the PACS image file migration process (e.g. perfecting),
- The types of messages across the interface.
Since users are mobile, a way of leaving messages for them is required.
Current teleradiology systems allow communication to a remote point but are focused on real-time (synchronous) communication about the images, requiring that the communicating parties arrange a communication session in advance.
IMagePHONE will extend the conferring and consulting facilities to allow multi-media messages to be easily composed and sent to a recipient, for his later review and reply.
The IMagePHONE communications work-station is intended to enhance communications about images within a hospital. It is focusing on the aspects of the interaction between radiologists and their customers in the hospital.
The layered schema approach of multi-database systems allow harmonisation and integration between a heterogeneous collection of text and image database. This can be applied to the issues of integrating clinical image modalities such as MRI and CT with textual information from HIS/RIS systems. Also the ability of multimedia database systems to provide a single query interface gives users transparent access to multimedia data through the decomposition of "global" queries into a set of sub queries to the required text or image sites. This allows the possibility of query optimisation to provide efficient query response times, including image indexing by content access for brain CT images. Network management and reliability control are key functions where a dedicated control system is developed, which can function in the EuriPACS architecture are in others.
The amount of data to handle has always created difficulties in the technical realisations of PACS. mainly related towards performance issues. A major focus is given to the implementation of compression and decompression. The objective is to select the well matched algorithm to medical image compression subjected to PACS constraints and diagnostic quality assessment of decompressed images.
The selected fields are:
- digitised: chest, hand radiographs, mammograms
- digital: MRI of the head, CT of liver
Compression factor 10:1 - No visible degradation of the decompressed image in JPEG, adaptive DCT coding, full-frame sub-band coding (scalar quantization). In the scalar sub-band coding scheme the PPSNR is lower than in the other techniques.
Compression factor 16.1 - No visible degradation of the decompressed image in JPEG, adaptive DCT coding, full-frame subband coding (vector quantization, JPEG scheme results in a lower PPSNR. Visible degradation is observed in decompressed image for the direct vector quantization scheme.
Compression factor 40:1 - Significant degradation of the image.
A report with an analysis of the "General trends and methods for expert systems in medicine", an analysis of radiological diagnosis and the possibilities to support the diagnostic process, cost (not in a financial sense) - benefit considerations and a bibliography with 66 items are developed.
"Guidelines for Collecting Images" contains a short introduction to the DECIS-philosophy and aims, and some considerations about image selection and indexing. Appendices give preliminary examples of radiological signs for the classification and diagnosis of bone tumours and the traumatic knee. "Computer Aided Diagnosis of Bone Tumours" describes the prototype. The prototype system runs on DEC-stations under OSF/Mothif. The database contains 34 bone tumours (or tumour like lesions) and a frequency matrix for 159 signs (i.e a 34 x 159 matrix). The dialogue has 52 questions (with up to 23 different answers per question). During the dialogue, additional windows can be opened.
Ensuring the overall integration of the different modules and systems a modelling of the hospital environment is done. Aiming at modelling biomedical management in the hospitals and between hospitals.
This modelling activity will be performed with effective involvement from both technical and clinical people, starting with the review of general requirements of a Medical Imaging Management System and progressing with the development of a three layers model, leading to a step-wise approach of the design of the model:
- The conceptual layer is an abstract specification of the requirements, describing the functions which must be provided as well as the information needed to perform these functions.
- The organisational layer describes an operational model, still abstract and independent of the products that will be used.
- The layer of implementation specifications represents the set of technical solutions which will be used for the implementation and involves the products.
The modelling activity, based on the NIAM method and supported by a specific CASE tool managing the consistency of the model, has been split into two broad domains: the image context, modelling the concepts of patient, clinician, exam request/report on the one hand, and the image kernel, dealing with the notions of image variables, indexation, sampling/quantification, and access methods on the other hand.
The functional model development is moving accordingly, describing data flows within the model and their associated processing.
The Care Unit Image Server Toolkit project is focusing on providing cost effective access to clinical images in the care unit. A user study is being carried out to determine the imaging the needs of physicians and other staff present in the care unit. The emphasis is on the needs of Care Unit staff which may differ from the needs of radiologists and radiological department staff. An intensive care unit at the Brussels Hospital has been selected as the Care Unit for which the prototype system will be designed. Widespread installation of image capable terminal in care units requires keeping costs low which require using standard commonly available hardware platforms. The performance available today in a high performance workstation is indicative of the performance that will be available in two years from a run of the mill personal computer.
- An evaluation protocol has been established that focuses on availability of information, communication, efficiency and effects on reporting.
- The before measurements in the three sites are almost completed. For each site, a description of the use of alphanumerically and image data in the radio-diagnostic process has been described. Data on performance of the film-based system (availability, retrieval time, etc.) have been collected (nearly completed).
- The functional specifications of the software to support cost-analysis have been finalised, and the preliminary version of the software package is reaching completion. It will be tested in the three sites in the beginning of 1993.
The EurIPACS is being in a curtail phase; clinical validation is necessary, first proto-types will be the first test cases for the needed objective. Lessons are learned that a very concentrated approach is needed in this hybrid field of medical imaging.
EurIPACS is also playing a strategically role in the validation of standards and gives major inputs to the work of CEN TC251, EWOS and ISO/IPI.
Fields of science
- engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringcontrol systems
- natural sciencescomputer and information sciencesdatabases
- medical and health sciencesclinical medicinecritical care medicine
- natural sciencescomputer and information sciencesartificial intelligenceexpert systems
- engineering and technologymedical engineeringdiagnostic imaging
Topic(s)
Data not availableCall for proposal
Data not availableFunding Scheme
Data not availableCoordinator
1090 Brussels
Belgium