PACS stands for Picture Archiving and Communication System. In practice it is three things bolted together: somewhere to keep images, a way for imaging machines to send them there, and a viewer that lets a radiologist read them and attach a report.
If your centre currently burns CDs, hands patients a film envelope, or keeps studies on the ultrasound machine's own hard disk until it fills, this article is about what changes when you stop doing that.
The vocabulary, briefly
DICOM
DICOM is the standard that medical imaging equipment uses to store and exchange images. A DICOM file is not just a picture — it carries the patient identifiers, the machine, the settings, and the study and series structure alongside the pixels. This matters because it is why a study can be reliably matched to the right patient without anyone typing a name twice.
Modality
The machine producing the images: X-ray, CT, MRI, ultrasound, mammography. Each behaves slightly differently in how much data it produces and how it prefers to send it.
Modality worklist
This is the feature that quietly saves the most time and prevents the most errors. Instead of a technologist typing the patient's name into the machine at the console, the machine queries a worklist and pulls the scheduled patient's details directly. Typed names produce misspellings, and misspellings produce studies that cannot be found later. If you buy nothing else, ask whether worklist is supported by both your PACS and your machines.
Viewer
The software used to read the images. A modern PACS offers a browser-based viewer, so a radiologist can read from home without installing anything. For diagnostic reading — as opposed to review — the display quality and the reading environment matter, and requirements vary by jurisdiction and modality.
What actually goes wrong without a PACS
- Studies get lost with the hardware. Images on a modality's local disk disappear when the disk fails or is cleared to make room.
- Prior studies are unavailable. Comparison against a previous scan is often the whole clinical point, and it only works if the prior is findable in seconds.
- The radiologist has to be physically present. This constrains hours far more than most centres realise.
- Reports and images live apart. A report in a Word file and images on a CD is two things to lose instead of one.
- Patients lose the CD — and then need a repeat scan, which means an avoidable dose and an avoidable cost.
Storage: the part people underestimate
Image volume varies enormously by modality. A plain radiograph is small. A multi-slice CT or a long MRI sequence is not, and a busy centre accumulates data faster than expected.
Rather than trusting a vendor's estimate, measure your own: take a typical week, sum the size of the studies your machines produced, and multiply. That number, plus your legal retention period, is your actual storage requirement. Retention periods for medical images are set by local regulation and are often measured in years, so ask a local authority or your medical association rather than assuming.
| Question | Why it matters |
|---|---|
| How much did we generate last week? | The only reliable basis for a storage estimate is your own mix of modalities. |
| How long must we retain studies? | Set by local regulation; drives total capacity far more than daily volume does. |
| Where is the second copy? | A single copy is not a backup. Ask where the second one lives and who can restore it. |
| What is the cost per additional TB? | Storage growth is continuous. A low entry price with expensive expansion is a trap. |
On-premise or cloud?
Neither is universally correct. The honest trade-off:
| On-premise | Cloud / hosted | |
|---|---|---|
| Works during an internet outage | Yes, within the building | No — this is the key risk |
| Remote reading | Needs VPN or exposure, done carefully | Built in |
| Upfront cost | Higher — servers and storage | Lower |
| Who handles backups and patching | You | The provider — verify what they commit to |
| Upload bandwidth needed | Minimal | Substantial, especially for CT and MRI |
Where connectivity is unreliable, a common and sensible pattern is a local store that holds recent studies and serves the reading room, with a copy pushed off-site for backup and remote access. It is more moving parts, but it fails gracefully.
Test upload speed before committing to cloud PACS. A CT study that takes twenty minutes to reach the server is a workflow problem, not a technical detail.
Questions to ask any PACS vendor
- Will it talk to my specific machines? Ask them to confirm against your actual make and model list, not "DICOM compatible" in general.
- Do you support modality worklist? And will they configure it on the machines, or is that your problem?
- Can a radiologist read from a browser? If a desktop install is required, ask what happens on a machine they do not control.
- How do I export a study? Both for a patient and in bulk if I ever leave you. Ask for DICOM export specifically.
- How are prior studies surfaced? Automatically alongside the current one, or by manual search?
- Where is the data stored, and under whose jurisdiction?
- What does year three cost once storage has grown?
Reporting is half the product
A PACS that stores images beautifully but makes reporting awkward will be worked around within a month. Look for report templates for your common studies, so a normal chest radiograph is a few clicks rather than a paragraph retyped daily, with the radiologist editing only what differs. Then check that the finished report can reach the referring doctor and the patient without anyone walking it to a counter.
See our PACS and the rest of the suite
LabSmart PACS provides secure storage for X-ray, CT, MRI and ultrasound studies, a browser-based viewer, radiology report templates and instant sharing. Like the rest of our software it is free to use — you pay only for the hosting that runs your private instance.
See all live demos → See our PACS → PathLab for pathologyA reasonable first step
If you are not ready to replace anything, start by fixing identity and retention: get every study off local machine disks and into one archive with a second copy, and turn on modality worklist so patient details stop being typed at the console. Those two changes eliminate most of the ways studies get lost, and they are independent of whichever viewer or reporting workflow you eventually choose.