Standard DICOM in. Any scanner, any PACS.
GE, Philips, Siemens, Canon, and handheld POCUS. No new hardware. No new views. No acquisition change. InVision analyzes the studies your lab already produces — in your data center or in our validated cloud — and returns the result into the study your cardiologist is already reading.
Cloud or on-premise. Your choice, not ours.
Both models run the same FDA-cleared algorithms and produce the same output. The difference is where the compute sits and what crosses your network boundary.
In your data center
All inference runs locally on a virtual machine inside your network, behind your firewall and under your access controls.
- PHI never leaves your network.
- No image data is retained after processing completes.
- Your existing firewall and security infrastructure do not need to change.
- Shortest turnaround, because there is no study upload step.
In our environment
Studies route to InVision's validated cloud; results return to your reporting workflow.
- No local infrastructure to provision.
- Hosted in the United States, in your geographic region.
- Covered by your executed BAA.
- Scales with volume without a hardware purchase.
What an on-premise deployment needs
One virtual machine, sized comparably to a typical PACS application server. Your infrastructure team provisions it; our engineer installs and configures everything on it.
| Requirement | Baseline | Notes |
|---|---|---|
| Virtual machine | 32 vCPU · 128 GB RAM | Comparable to a typical PACS application server. Sized with your team during scoping. |
| GPU | NVIDIA — recommended | Not mandatory. Recommended to minimize turnaround time. |
| Container runtime | Docker | Full dependency list provided during scoping. |
| Vendor access | Remote access to the VM | For installation, configuration, testing, and support. |
Vendor-neutral by design
InVision reads standard DICOM. If your scanner writes it and your PACS stores it, InVision can analyze it — there is no acquisition protocol to change and no proprietary export to configure.
Because intake is standard DICOM rather than a vendor integration, a mixed fleet is the normal case, not an exception. Labs running several scanner generations across multiple sites do not need to normalize hardware before deploying, and adding a scanner later does not require new integration work.
Standing up intake means one change on your side: a DICOM routing rule that also routes completed studies to the InVision virtual machine. InVision is post-processing software — it analyzes completed studies and does not sit in the acquisition path, so it cannot delay or interrupt a scan.
How results come back
Into the study your cardiologist is already reading — not into a separate viewer, and not behind a second login.
DICOM secondary capture
The result is written back into the originating study as a secondary capture image, available to the reader as the last frame. This is how most deployments are configured.
Structured report field
In parallel, the result can populate a field in your structured report template, so it lands in the report the cardiologist signs.
Positives worklist
Results can also be delivered as a list of flagged studies, for services that want a single queue to work from.
InVision integrates with Syngo Dynamics and Epic Cupid, with results flowing into the study, into the PACS reporting environment, and into the EHR. Billing capture is part of the same build: a checkbox in the structured report template records that AI was used, and that selection is what carries CPT 0932T through to the claim.
Turnaround and uptime
Because Precision Cardiac Amyloid needs only the PLAX and A4C views — normally acquired in the first half of a study — the result is frequently ready before the study itself is finished.
| Measure | On-premise | Measured from |
|---|---|---|
| Median turnaround | 0 min | Study completion — the result is typically already available. |
| 75th percentile | 2 min | Study completion. |
| Maximum | 10 min | Study completion. |
| Result into PACS | < ~5 min | InVision receipt of the required images. |
| Uptime | 99.9% | On-premise deployment, assuming your PACS is online. |
When comparing turnaround figures across vendors, check whether the quoted number includes the full study upload time or only the algorithm runtime measured from receipt of the last frame. Those are very different numbers in a cloud deployment.
About four weeks, kick-off to go-live
The schedule below is our standard implementation. Actual duration depends far more on your change-control and security review calendars than on anything InVision installs.
Contracting and access
Security and compliance review, contracting, and provisioning of the virtual machine with vendor access. We review your current echo and billing workflows as they are today, and agree the DICOM routing rules and output selection criteria.
Kick-off and scoping
Roles and responsibilities agreed, success metrics and timelines confirmed, DICOM routing rules finalized and pointed at the VM, AI billing codes configured, and support procedures reviewed.
Integration and testing
Result delivery into the PACS verified, test studies run end to end, and the algorithm run in shadow mode against live routing so you can confirm the rules behave correctly before anything reaches a reader. Turnaround and billing capture are both tested.
Clinical go-live
Results begin flowing into the clinical workflow, alongside training for cardiologists, sonographers, and the billing team.
Hypercare and reporting
24×7 support for the first 30 days, then monthly reporting covering usage, results, follow-ups, and invoicing detail.
What your team actually does
Roughly 60 person-hours in total across the four weeks, spread across the roles below. Our forward deployed team carries the integration and validation load.
| Role | Effort | Responsibility |
|---|---|---|
| IT project manager | 4–6 hrs/wk4 weeks | Coordination across all phases. |
| Enterprise architecture | ~2 hrs | Provision the virtual machine and grant vendor access. |
| PACS application specialist | ~16–20 hrs | Configure routing rules and result return, interface testing, AI billing code. |
| EHR application specialist | ~4–8 hrs | AI billing code and result field setup. |
| Billing / RCM analyst | ~8 hrs | Charge capture workflow alignment and training. |
| Reading cardiologists | 1 hr each | Program overview, workflow, and billing training. |
| Lead sonographer | 1 hr each | Program overview and workflow alignment. |
Updates and change control
No downtime to your clinical systems, and nothing changes in your environment without your sign-off.
Infrequent by design
Our FDA-cleared algorithms are static — they cannot be modified without a new clearance. Updates occur only when a newly cleared version is released, and are adopted at your election.
Scheduled with you
Periodic interface and field-mapping maintenance, scheduled in advance. Processing is asynchronous and per-study, so your PACS, EHR, and echo workflow stay online throughout.
Updates are applied in a mutually agreed maintenance window, targeted off-hours and under 30 minutes. Studies arriving during that window queue and process automatically when the service returns. Every update follows a documented change-control process: advance written notice and change summary, deployment first to your test instance, regression testing of interfaces and field mapping plus an end-to-end sample study, your sign-off, then production deployment with a defined rollback plan.
We also ask for advance notice of your own planned PACS or EHR upgrades, so we can confirm no interface changes are needed and pre-test against them.
Support, from the people who built the models
Support is delivered directly by our clinical and engineering team — the same people who built, cleared, and validated the algorithms.
Implementation
A named Forward Deployed Engineer owns integration, testing and validation, and clinical enablement, with a project plan and escalation path agreed at kick-off.
Hypercare
24×7 phone and email support for the first 30 days after go-live, with a one-hour call-back target.
Ongoing
Email support Monday–Friday, 8am–5pm local, with same-day response or under four hours into the next business day. A named account executive runs monthly and quarterly business reviews.
As an FDA-cleared Software as a Medical Device, support also carries the post-market obligations of a device manufacturer: formal complaint intake and handling, adverse event evaluation and reporting, model performance monitoring, and version control for any approved algorithm update — all governed under our ISO 13485 quality management system.
Start a scoping call
Thirty minutes with our clinical and technical teams to map InVision onto your PACS, CVIS, and EHR — and to decide whether cloud or on-premise is the right fit for your security posture.