Platform & software · Cloud based PACS
Cloud based PACS: cloud PACS systems, vendors and pricing compared
The short answer
A cloud based PACS stores and serves imaging studies from a vendor-operated data center rather than from servers in your building, so radiologists reach the archive over the internet instead of over the hospital LAN. Cloud is a hosting decision, not a regulatory category: the FDA clears the software function under product code LLZ (21 CFR 892.2050), and of the 2,285 clearances ever granted under that code, only 7 carry the word cloud in the device name. That means you cannot tell whether a cloud PACS has been through FDA review by reading its marketing, and several widely promoted cloud PACS products hold no 510(k) under their own name at all. Look the vendor up by applicant name in the 510(k) database before you shortlist.
The phrase cloud PACS gets attached to at least four different products. One is a traditional PACS someone installed on a rented virtual machine. One is a vendor-managed instance of conventional PACS software. One is genuinely cloud-native, written to run multi-tenant from the start. And one is not a PACS at all: it is a cloud archive sitting behind the on-premise PACS you already own. All four are sold with the same word, and they do not cost the same, migrate the same, or fail the same way.
The question that separates them fastest is not technical. It is regulatory. Image management and processing software is a Class II device in the United States, cleared under product code LLZ, and the clearance attaches to the software function rather than to where the servers sit. So the useful test is simple: does this vendor appear as the applicant on a 510(k) under LLZ, and for which product? Some cloud PACS vendors have a long clearance history. Some have none. The table further down this page lists real records, with K numbers you can verify yourself.
Radiological.ai is not a PACS and does not replace one. It works alongside whatever archive you run, cloud or on-premise, over standard DICOM and HL7 interfaces. It flags suspected findings for a second look, pushes urgent studies up the worklist and drafts the structured report into your template. The radiologist reviews, edits and signs every study.
Last updated September 2026
Worklist
Structured report
DraftRun the assistant to draft this report for review.
Illustrative sample · not a real patient study, not a diagnosis
Decision support for qualified clinicians. Radiological.ai does not provide a diagnosis and is not a substitute for professional judgment.
Flag · prioritize · draft · you review and sign
Decision support not a diagnosis
You review & sign
Cloud is one axis of the buying decision and it is rarely the one that decides the project. If the open question is which PACS to buy at your study volume rather than where it runs, that is PACS software and PACS system vendors compared by study volume. If the real problem is where studies live long term and how they survive a vendor change, that is enterprise imaging and vendor neutral archive software, where the archive is treated as its own layer. The operational half, meaning scheduling, worklists and charge capture, sits in radiology information system software.
On the display side, whether a browser-based viewer is cleared for primary interpretation is a separate question from whether the archive is in the cloud, and it is covered in DICOM viewer software and which viewers hold an FDA clearance. For how an assistant like this one connects to whatever you run, see radiology PACS integration. For the wider picture of what imaging software costs, drawn from public federal contract awards, see our radiology software pricing research, and for a shortlist aimed specifically at outpatient sites, the best cloud PACS software for imaging centers.
Side by side
The four things vendors mean by cloud PACS
These are four different purchases with different cost curves, different exit costs and different failure modes. Establish which column a proposal actually sits in before you compare it to anything else, because a hosted instance of conventional PACS software and a multi-tenant cloud-native platform are not the same product with different branding.
| What you are comparing | Self-hosted in your own cloud tenancy | Vendor-hosted single tenant | Cloud-native multi-tenant | Cloud archive behind an on-premise PACS |
|---|---|---|---|---|
| What it actually is | Conventional PACS software you install on rented infrastructure such as AWS or Azure | The same conventional software, run for you as a dedicated instance by the vendor | Software written from the start to serve many organizations from shared infrastructure | Your existing PACS keeps reading; long-term storage moves off-site |
| Who patches and updates it | You do, including the operating system and the database | The vendor, usually on their release schedule | The vendor, continuously, with everyone on one version | The archive vendor for storage; you still own the PACS |
| Typical commercial shape | Perpetual or subscription license plus your own infrastructure bill | Annual subscription per site or per study volume band | Per-study or per-user subscription, lowest entry cost | Storage priced per study or per terabyte, added to existing PACS costs |
| Where the exit cost hides | Low software lock-in, high operational burden | Migration is a vendor project and priced as one | Data export terms matter more than anything else in the contract | Lowest disruption, because the reading system does not change |
| Best fit | Health systems with a real cloud engineering team | Groups replacing an aging on-premise PACS without adding staff | Imaging centers and new sites starting without legacy infrastructure | Anyone whose PACS works but whose storage is full |
| Question to ask first | Who is on call at 2am when the VM stops responding? | What is the contractual restore time after an outage? | What format is my data in when I leave, and what does export cost? | Does retrieval of a 10-year-old prior add delay to the read? |
Compiled September 2026. Clearance statements are from the openFDA 510(k) API, product code LLZ, queried September 2, 2026. Contract figures are from USAspending.gov prime award records and describe what federal buyers paid, not vendor list prices.
Primary source
Cloud and web based PACS vendors with a verified FDA 510(k)
Every row is a real record in the FDA 510(k) database, pulled through the openFDA API on September 2, 2026 under product code LLZ, which is image management and processing software under 21 CFR 892.2050. The K number is the most recent clearance held by that applicant; the count is every LLZ clearance under that applicant name. Look any of them up at accessdata.fda.gov. This is not a ranking and not a shortlist. It answers one narrow question most buyers never ask, which is whether the company selling you imaging software has been through review under its own name.
| Applicant on the submission | Most recent LLZ product | K number | Decision date | LLZ clearances held |
|---|---|---|---|---|
| Fujifilm Medical Systems U.S.A. | ASPIRE Bellus II | K171463 | 2017-09-29 | 13 |
| Sectra Imtec AB | Sectra Workstation IDS7 | K081469 | 2008-06-12 | 12 |
| Novarad Corporation | OpenSight | K172418 | 2018-09-21 | 6 |
| Merge Healthcare Incorporated | Merge Universal Viewer (MUV) | K250301 | 2025-04-14 | 6 |
| Intelerad Medical Systems | IntelePACS | K192176 | 2020-04-02 | 5 |
| eRAD, Inc. | eRAD PACS | K241223 | 2024-10-31 | 5 |
| RamSoft, Inc. | OmegaAI Image Viewer | K222476 | 2022-11-17 | 5 |
| Change Healthcare | Change Healthcare imaging software | See accessdata.fda.gov | Multiple | 4 |
| DICOM Grid dba Ambra Health | Ambra PACS | K231360 | 2023-06-07 | 2 |
| Visage Imaging GmbH | Visage Ease Pro | K142196 | 2015-04-28 | 2 |
| Konica Minolta Healthcare Americas | EXA | K203743 | 2021-12-10 | 1 |
Why it works
What your group gets with cloud based PACS
Cloud is a hosting choice, and the FDA does not clear hosting
Product code LLZ covers image management and processing software regardless of where it runs. Across all 2,285 LLZ clearances ever granted, only 7 name the word cloud in the device, and those include products as different as a dental imaging package and a 3D segmentation service. So a vendor page that says cloud-based and FDA-cleared in the same paragraph is describing two unrelated properties. Verify the clearance by applicant name and product, not by the hosting description, and be aware that some well-marketed cloud PACS products return no LLZ record under their own company name at all.
The exit terms matter more than the monthly price
On-premise PACS locks you in with hardware and a migration project. Multi-tenant cloud locks you in with data gravity and export terms, which are cheaper to ignore at signature and far more expensive later. Ask three things in writing before you sign: what format studies are returned in, whether the export includes presentation state and report objects or only the pixel data, and what the export costs per study. Migration is commonly quoted at roughly $0.05 to $0.50 per study, so an archive of 200,000 studies is a five-figure line item that belongs in the business case rather than in a surprise a few years out.
Read latency is the clinical risk, not uptime
Vendor uptime numbers describe whether the service answered, not whether a radiologist could work. The measure that matters is time to first image on a large study over your actual connection, and time to retrieve a decade-old prior that has been moved to cheaper storage tiers. Ask for both, test them during the evaluation on your worst-connected site rather than at headquarters, and put the retrieval time for archived priors into the contract. A cloud PACS that serves new studies quickly and takes two minutes to return a comparison is a workflow problem that no service level agreement about uptime will surface.
What it handles
Flagged, prioritized and drafted for your review
The assistant pre-reads each study, surfaces a region of interest for review, re-prioritizes the worklist, and drafts the structured report in your template. You confirm, edit and sign.
- Works with cloud and on-premise PACS alike
- Connects over standard DICOM and HL7 interfaces
- No change to your archive or storage contract
- Flags suspected findings for a second look
- Reorders the worklist so urgent studies surface first
- Drafts the structured report into your template
- Radiologist reviews, edits and signs every study
Region of interest flagged for review
A focal region is surfaced on the sample study for the radiologist to review. The assistant does not characterize it as a diagnosis.
Suspected finding flagged for radiologist review. Correlate clinically and confirm. Draft for review and sign-off.
Why Radiological.ai
One assistant across the whole read
Not three vendors stitched together. Flag, prioritize and draft in one calm pane, on X-ray, CT and MRI, with the radiologist signing every study.
Flags suspected findings
A second set of eyes surfaces regions of interest for review on every study, so a suspected finding is less likely to slip past late in a shift.
Prioritizes the worklist
Suspected-critical studies move to the top, so urgent reads surface ahead of routine follow-ups across your sites and shifts.
Drafts the report
A structured draft arrives in your template, ready to edit and sign. The draft saves the typing and the measuring, never the judgment.
Good questions
Questions about cloud based PACS
Explore more
More ways teams read with Radiological.ai
Radiology AI platform
One assistant across the whole read, not three point tools stitched together.
Learn moreAI medical imaging
A workflow assistant for the imaging your team reads every day.
Learn moreAI radiology assistant
An assistant that reads alongside you and drafts before you sit down.
Learn moreFrom the blog: best AI radiology software in 2026, how much radiology AI costs, how to implement AI in a radiology workflow, and AI radiology companies.
Read more studies, with the assistant alongside you
Flag suspected findings, prioritize the worklist, and draft the structured report. You review and sign every study.
Radiological.ai is a workflow and decision-support tool for qualified clinicians. It does not provide a diagnosis and is not a substitute for professional medical judgment.