Connecting a clinical application to a Picture Archiving and Communication System, known as PACS, was one of the more specialised pieces of integration work we have taken on. The imaging world runs on its own decades old standard called DICOM, with its own way of moving data and its own expectations about reliability that are different from a typical web based integration. We underestimated how different those assumptions were the first time we built this, and here is what we learned.
It is tempting to treat this kind of integration as "we just need to read and show image files," but most actual PACS integrations also involve a way of asking for and fetching studies, not only showing files that happen to arrive. We learned to build our integration around that asking and fetching step from the very start, a thin layer whose only job is "fetch studies for this patient and date range," rather than assuming files would simply appear somewhere for our application to pick up.
A single imaging study can be tens or hundreds of megabytes across many separate images. We treat fetching and working with a study as background work with its own status, the same way we would treat any other job that takes live, rather than something expected to finish inside a normal web request. Showing clinical staff a clear message such as "fetching study, please wait," turned out to matter a great deal more than we expected, compared to a request that simply seems to hang with no explanation.
The patient identifier in our own clinical system and the identifier used by the imaging system are not guaranteed to be the same value, especially when working with an outside radiology provider rather than an imaging system we run ourselves. We built a clear matching step instead of assuming the two ids would simply line up, and we treat any study that cannot be matched as something visible that a staff member needs to resolve, never as a failure that quietly disappears.
Whether to show a full image viewer inside our own application, or simply link out to a separate viewing tool, changes the clinical workflow a great deal. We learned to decide this on purpose. Does a radiologist or a referring doctor need full, diagnostic quality viewing inside our own screens, or is a lighter preview with a link to a dedicated viewer enough. These two choices carry very different engineering and licensing costs, and we now make this decision early, with actual input from clinical staff, not as an afterthought.
A PACS connection that is down for an hour is not simply inconvenient. It can delay a diagnosis. We build retry attempts, a clear message when the imaging system cannot be reached, and a written fallback plan for clinical staff, with the same seriousness we would give to any piece of the system that touches patient safety directly.
Maybeach Tech has connected clinical applications to imaging and diagnostic systems. Get in touch and let us talk about your radiology or PACS integration.
