The wedge
Why on-premise matters for radiology AI
Many cloud-based AI workflows depend on moving sensitive imaging data away from the clinical site. OxRad takes a different path: inference runs on-site, so patient imaging data remains under the clinic's control.
Why does on-premise matter for radiology AI?
On-premise radiology AI runs inference on hardware inside the clinic, so a patient’s scan is never uploaded to a third party. The data stays in the building where it was created. That removes the largest privacy risk in cloud reporting, the transfer itself, and makes sovereignty the architecture rather than a policy promise.
On-device vs cloud radiology AI
The same report, two architectures. The difference is where your patients’ data goes.
The data stays inside the buildingIs cloud radiology AI a privacy risk?
The risk is not abstract. To draft a report in the cloud, the study has to leave the site and sit, however briefly, on infrastructure you do not control. That transfer and storage widen the attack surface, create a chain of custody you cannot fully see, and raise hard questions about data residency and patient consent across jurisdictions.
On-premise inference removes the transfer. The scan is read where it is created, the draft is produced locally, and a radiologist signs. There is no off-site copy to secure, subpoena, or breach. You are not asked to trust a vendor’s cloud governance, because the data never reaches it.
What about cost?
Cloud reporting carries a recurring, per-study inference cost that scales with your volume. Edge inference on hardware you control removes that line item: once the OxRad on-site node is in place, drafting a report does not incur a cloud charge.
You trade a managed appliance and its setup for data sovereignty and predictable economics. For a busy scan centre, the arithmetic and the privacy posture point the same way.
Common questions
- Does radiology AI send patient data to the cloud?
- Most cloud reporting tools do: the study leaves the site for inference on the vendor's servers. OxRad does not. Inference runs on the OxRad on-site node in your clinic, so the scan is never uploaded.
- Is cloud radiology AI a privacy risk?
- The transfer itself is the largest risk. Sending a study off-site for inference, and storing it with a third party, widens the attack surface and complicates data residency and consent. Keeping inference on-premise removes the transfer entirely.
- How do I run AI radiology reporting without sending data to the cloud?
- Deploy the OxRad on-site node in your clinic. A study arrives in standard DICOM, OxRad Intelligence drafts the report locally, and a radiologist reviews and signs. Nothing transits the cloud at any step.
- Is OxRad GDPR or HIPAA compliant?
- Compliance is a property of your deployment, not a badge OxRad can grant. The on-premise architecture supports it by keeping patient data on-site and minimising what is processed, backed by a tamper-evident audit. Specific regulatory certifications are targets in progress, listed on the Security page, and are not presented as held today.
Evaluate on-site AI reporting that keeps patient data on your premises.
We are taking a small number of pilot partners.