Product
What to demand from a web DICOM viewer before you buy it
“Web viewer” shows up on the spec sheet of practically every RIS/PACS on the market. The problem is that the term covers two very different things: anything from a complete diagnostic workstation to a basic image display that shows the image and lets you zoom in, but will not carry a reading session.
The distinction matters because a radiologist notices it within the first ten minutes of use — and by then the contract is signed. These are the points worth verifying beforehand, ideally with a real study from your own center.
1. Is it genuinely zero-footprint?
The concrete question: does anything have to be installed on the workstation? A client, a plugin, an extension, a desktop component?
If the answer is yes, it is not a web viewer: it is a desktop viewer with a web layer on top. You pay for that difference every time the operating system updates and breaks the plugin, and every time you have to roll it out station by station.
2. Does it have MPR and 3D, or only slice by slice?
Multiplanar reconstruction — being able to view a CT in axial, coronal and sagittal — is not a luxury. It is the minimum for many studies, and it is exactly the first capability that falls away in lightweight web viewers.
GPU-rendered 3D volume is the next step up. The useful technical question is where the rendering happens: if it is on the server, performance degrades when several radiologists read at once; if it uses the GPU on each radiologist’s machine, the load is distributed and the system scales without additional infrastructure.
3. How does it load a study with thousands of slices?
Explicitly ask for the test with a heavy CT, not with a chest X-ray.
A well-built viewer uses progressive loading: it brings in what you need to see first and the rest arrives in the background, so you start reading within seconds. A poorly built one downloads the entire study before showing the first slice, and that wait gets multiplied by every study in the day.
4. Does it compare with prior studies, with series synchronized?
A good part of reading is comparative. A viewer that opens the current study but makes you hunt for the prior one in another tab, without synchronizing scrolling between series, adds friction at exactly the moment the radiologist is reasoning.
5. Do measurements stay linked to the report?
Here is the difference between a viewer that is an isolated tool and one that is part of the workflow.
If the radiologist’s measurements and markers are available as data, the system can compare them against what was written in the report and flag when something does not match — a measurement that differs, a laterality that was switched. If the viewer is disconnected from the report, that check is simply impossible.
6. What is it built on?
It is worth asking about the rendering engine. Foundations like Cornerstone3D and OHIF underpin clinical and research viewers worldwide: they are well known, auditable and have a community behind them.
A proprietary engine that nobody outside the vendor knows anything about is not disqualifying on its own, but it is one more dependency you take on.
7. Does it connect to what you already have?
If the viewer speaks DICOMweb — WADO, QIDO, STOW — it can connect to an existing PACS without forcing you to replace it on day one. That lets you evaluate it against your own archive before committing the whole operation, which is the best possible test.
How to test it
A practical recommendation: do not evaluate the viewer in a guided demo with cases the vendor picked. Ask for access, hand over a heavy study from your own archive, and let a radiologist on your team use it unaccompanied for half an hour. The gaps make themselves obvious.
The NOVA Imaging viewer is fully web-based, with windowing, measurements, MPR, GPU 3D volume and comparison with priors, built on Cornerstone3D and OHIF. You will find the details on the RIS/PACS platform page, and we can give you access with a demonstration case so a radiologist on your team can evaluate it.