mSI™, a relative risk score at the point of nodule.
The lung nodule on the scan already contains the answer. Two nodules that look identical differ in their radiomic features — sub-visual properties of the tissue, present from the moment the scan was acquired. RevealAI-Lung measures them at the point of nodule and returns that nodule’s Malignancy Similarity Index (mSI™). The score appears inside the viewer the radiologist is already working in, without a second scan or a waiting interval.
Installed where your imaging already lives.
RevealAI-Lung runs as Docker images inside your own environment — on-premise in your virtualization or container platform, or in your own cloud, securely connected to the facility network. Either way it is your infrastructure and your controls, with no specialty GPU hardware and no vendor-operated cloud.
What your IT team is actually agreeing to.
| Item | Detail |
|---|---|
| Deployment | Docker images in your existing container or virtualization platform — on-premise, or in your own cloud environment connected to the facility network |
| Hardware | 16 CPU cores, 16 GB memory. No specialty GPU server required. |
| Storage | 1 TB minimum; additional storage extends how long clinicians can access results per case |
| Ports | 80/443 (HTTP/HTTPS) and 11112 (DICOM storage). All others closed by firewall rule. |
| PHI | Encrypted at rest with a site-specific private key. HIPAA and GDPR conformant. |
| Logging | System-level and application-level access both logged |
| Integration | Standard DICOM routing. Typically 5–10 business days from agreement. |
Two integration patterns.
Inside the PACS viewer
The server receives CT series by DICOM C-STORE. The radiologist launches RevealAI-Lung from within the existing viewer and results display in an integrated window.
From a detection system
RevealAI-Lung consumes the DICOM structured report from a third-party CADe system and scores only the nodules already flagged. Results are pushed back to PACS.