Gastos fijos y variables

The different tasks involved in diagnostic radiology have different information requirements and are amenable to different forms of computer support. A radiology department will normally exist within a hospital which will have its own information system providing for the storage, capture and transmission of patient-related and other information (a hospital information system or HIS). There will often be a separate computer system providing these functions within the radiology department (a radiology information system or RIS). There may be, in addition to these systems, some form of computerised patient record or electronic healthcare record (EHCR).

A number of the imaging modalities will involve digital capture, storage and display of images. There may be another system for the archiving and communication of images independent of the individual imaging systems (a picture archiving and communication or PACS system). There are clearly considerable benefits to be obtained from the integration of these different systems. Such integration or ‘inter­ operability’ is now a major topic of research in medical informatics and many groups are working on the development of standards and formal models that will facilitate the development of systems which can share information. Two directly relevant projects are the development of the DICOM standard and the MIMOSA approach to modelling PACS systems.

The American College of Radiology and the National Electrical Manufacturers Association jointly developed a standard for Digital Imaging and Communication (DICOM) which was published in 1993 [DICOM 1993]. Continuing work on this standard has involved working groups from European and Japanese standardisation bodies and has been adopted by vendors of imaging devices and clinical workstations. The standard covers such topics as the encoding of image data and file transfer protocols as well as notions like the definition of Information Objects which allow the

representation of concepts such as ‘patient’ or ‘visit’ and the specification of operations performed on Information Objects as well as the expected behaviour of the user and provider of the operation.

The MIMOSA^ project [Garfagnil994] developed an approach to modelling medical image management systems which tackles the issue of inter-operability between PACS and HIS/RIS. The project, which was supported by the EU as a research rather than a standardisation initiative, attempts to reconcile two different views of medical imaging: the HIS view which considers medical images as a subset of medical information and the PACS view which is concerned with the acquisition, processing and storage of images without being concerned with why they are produced or how the information they contain is used. The MIMOSA model consists of three loosely coupled models: the data model describes the structure and relationships of the data represented in the system, the functional model describes what the system does, the dynamic model describes the behaviour of the system over time.

It is worth reflecting on the ultimate goal of this work. It is theoretically possible to have a filmless, paperless hospital in which all information, image-based and otherwise, exists only on computer. This is not necessarily desirable. Indeed, CT images which can only be created by computer and which must be viewed on a computer screen in order to establish the appropriate display parameters, are routinely printed on film to be seen by the radiologist. Film offers many advantages to the radio­ logist: they can pick it up, view it next to other films, scan large numbers o f images and carry them from office to office.

The development of networks of communicating computer systems supporting the different tasks performed in radiology will require balancing the advantages to be

gained by standardisation and computerisation with the flexibility and familiarity of traditional ways of working. Consider the four tasks described in the previous section.

• Establishing the need for a radiological examination requires communi­ cation between a referring clinician and a radiologist. This might be supported by a computer, if both clinicians regularly use email or video­ conferencing, but it may be more conveniently done some other way. It may involve reference to guidelines, it could involve reference to research material. Again, this kind of information can be conveniently accessed by computer, but it doesn’t have to be. It could involve calculations or infer­ ences made by a computer on the basis of some form of stored knowledge. • Once the need for an examination is established the examination must be

performed. The efficient use of imaging devices and the attendant personnel requires the use of computerised scheduling systems. Such systems are generally provided with radiology information systems and the basic concept, that of the worklist, is being defined in standards to allow inter­ operability of PACS and RIS systems. Many image generation techniques are now digital and require computers. Others may become computerised. • The interpretation of images, if at some point they exist in digital form, may

be facilitated by computer systems which either enhance, analyse or interpret the image or which provide information that the radiologist can use in making a decision.

• The generation of radiological reports can be assisted by systems of menus which provide access to standard lexicons, or by systems which provide access to reporting standards. Voice recognition and word-processing software can ease the creation of a written report and information systems can be used to manage the sending out of reports and appropriately timed follow up letters.

The research described in this thesis focuses on the kind of computer system which would be used by a radiologist to help make a decision about an image. The work, however, is also guided by a desire to consider this as part of a process of care and to bear in mind that a computer system designed to support part of a process must sit within a network of inter-operable systems supporting the whole process.

1.4

Summary

The opening section of this chapter set out four aims for this thesis:

• to consider the range of systems which could potentially improve medical decision making from a unifying perspective

• to consider how different kinds of system could be integrated into a single decision aid

• to propose a generic model for decision support tools which allows users to access relevant information

• to develop prototype systems which demonstrate the practicality o f these ideas

The following section described four different activities involved in carrying out a radiological investigation: establishing the need for the investigation, performing the examination, interpreting the results and providing a report. Each of these tasks can be supported by the different computer systems which are currently in place in most radiology departments. A number of projects are now working on the integration and inter-operability of these different kinds of system. That work provides the context within which the research described in this thesis should be considered.