Focus in this paper is architecture of the computer system, i.e.
how that system is designed in terms of interactions between large
building blocks. This is a technology perspective. System architecture
is not that visible to the radiological user, in contrast to the
user interface. Moreover, whereas evolutions in the user interface
clearly are of immediate concern to any radiologist [1], the same
can hardly be claimed for system architecture. However, architecture
of the overall system has an important impact on functionality
and cost, especially for large, complex systems. Radiologists, and in
particular decision makers, should have at least a basic understanding
of PACS architecture, and in particular of the consequences of
architectural choices.
There is no inherently good or bad architecture for a computer
system. Design choices are determined by requirements as well as
technological possibilities. Both have changed since the introduction
of large-scale PACS.
Technological progress has been steady over the last decades.
In many areas of computer industry, capacity or power doubles
roughly every 18 months. This observation is informally referred to
as Moore’s law. This is literally an exponential increase in power,
or about a factor of 100 after a decade. General belief is that this
law will remain valid for many years to come – in fact, this expectation
itself and the resulting drive from marketing and product
development already help sustain the trend.
An area in which technological progress has enabled fundamentally
new radiological applications at a large scale, is teleradiology.
That topic is covered elsewhere in this issue. For in-hospital PACS,
advances such as faster and more ubiquitous computers, faster inhouse
networks or cheaper storage were no absolute requirement.
It may therefore seem that these advances will “only” enable us to
build faster, more functional and less expensive systems. However,
even for the more traditional aspects of PACS the consequences are
more profound.
Firstly, whereas PACS used to require state of the art technology,
it can now do with the same hardware used by the other information
systems in the hospital. A PACS increasingly becomes a
software application. There is no convincing reason anymore for
that application not to use the standard hardware the hospital has
chosen within its overall IT strategy. And that IT strategy will only to
a limited extent be influenced by PACS.Wecome back to this observation
throughout the different sections in this paper. Secondly, this
technological progress sets the scene fornewexpectations andnew
requirements in PACS, which become driving factors for the design.
In this paper we discuss a few of those.
A first influencing factor is the shift of storage needs to applications
outside radiology. PACS no longer is the single application
that requires huge amounts of storage. This favors an approach in
which storage is provided centrally and the PACS has to adapt to
the overall storage policy.
A second factor we discuss is that a growing number of departments
start to routinely acquire and use digital images.