RayStation

Evidence summary

Manufacturer nonclinical verification and validation supported the 510(k). An independent hospital study found generally close agreement between measured data and RayStation's VMC++ electron Monte Carlo calculations, but did not report the software version and did not assess clinical outcomes.

• Public source
• Product family
• Checked 2026-09-02

• Source
• Source

Postmarket safety review

FDA recall Z-2489-2025 includes RayStation 9B and 9B SP1. DICOM objects exported through the RayGateway interface to Accuray iDMS were not guaranteed to receive unique SOP Instance and Series Instance UIDs. FDA lists the recall as open and classified. FDA recall Z-1187-2025 includes RayStation 9B and 9B SP1. It concerns inconsistent interpretation of density uncertainty in robust optimization, robust evaluation, and perturbed-dose calculations for proton and light-ion plans using an HU-to-mass-density calibration curve. FDA lists the recall as open and classified. FDA recall Z-1745-2024 affects RayStation 9B builds 9.1.0.933 and 9.2.0.483 because the reported source-to-surface distance could be too high. FDA lists the recall as open and classified. FDA recall Z-1482-2024 includes RayStation 9B. It addresses a use error in which elemental silicon could be selected instead of silicone gel polymer as an implant material override, with potentially greater consequence for ion-beam planning. FDA lists the recall as open and classified. FDA recall Z-1403-2022 affected RayStation 9B SP1 because treatment-course information might not propagate from RayStation with RayTreat to RayCare in some workflows. FDA lists the recall as terminated. FDA recall Z-2138-2021 includes RayStation 9B and 9B SP1 used with RayCare because setup beams at the localization point could produce incorrect initial delivery positions. FDA lists the recall as terminated. FDA recall Z-1838-2021 includes RayStation 9B and addresses potentially misleading evaluation of edited clinical goals for voxelwise worst-dose distributions, plus a contouring issue under specified slice-spacing conditions. FDA lists the recall as terminated. FDA recall Z-1598-2021 includes RayStation 9B because the Total dose display in Dose tracking could be incorrect when a treatment course contained more than one plan. FDA lists the recall as terminated. FDA recall Z-1427-2021 includes RayStation 9B and 9B SP1 because flags edited in RayCare after patient check-in might not update in RayTreat. FDA lists the recall as open and classified. FDA recall Z-1597-2021 includes RayStation 9B because merging clinical beams with beams carrying approximate dose could cause mixed dose to be labeled as clinical dose for some LINAC workflows. FDA lists the recall as terminated. FDA recall Z-2039-2023 includes RayStation 9B and 9B SP1 because applying specified geometry functions directly to deep-learning-segmented ROIs could create incorrect structures and affect plan evaluation. FDA lists the recall as terminated. FDA recall Z-0589-2021 includes RayStation 9B and concerns treatment-plan reports that could show an incorrect percentage of an ROI outside the dose grid and, for 9B, incorrect six-degree-of-freedom couch angles. FDA lists the recall as terminated. FDA recall Z-0478-2021 includes RayStation 9.1 and 9.2 because a missing referenced ROI or POI in an imported RT Structure Set could become linked to the wrong structure or point. FDA lists the recall as terminated.

• FDA source
• Exact product version
• Checked 2026-09-04

Each published event was manually matched to the exact FDA submission and affected product or version; a recall does not establish an AI-algorithm defect or current site exposure unless the event states that scope.

• Source
• Source
• Source
• Source
• Source
• Source
• Source
• Source
• Source
• Source
• Source
• Source
• Source

• Verify security and privacy controls: No exact RayStation 9B cybersecurity architecture or supported-platform security guide was established from the reviewed public sources. Obtain version-specific security, patching, identity, encryption, audit, remote-access, backup, and vulnerability-management documentation.
• Verify local validation and lifecycle monitoring: The reviewed public sources do not define an exact 9B operational monitoring program. Require local dose-calculation commissioning, end-to-end testing, DICOM transfer checks, machine and beam-model change control, backup testing, service-pack regression testing, incident review, and recall surveillance.

Recovered from the prior exact-submission extraction and normalized under the current provenance rules. The exact FDA summary, current vendor product context, peer-reviewed product-family dosimetry evidence, public pricing observation, CMS payment context, and every exact-identifier FDA recall candidate were manually reviewed. Security and operational-monitoring details remain explicitly unresolved for the exact 9B releases rather than being inferred from later documentation. Automated exact-name discovery found 20 PubMed and 5 ClinicalTrials.gov candidate records. Candidates require human product and version matching; zero candidates is not evidence that no studies exist. 14 FDA recall candidates were manually reviewed: 13 published and 1 rejected as a neighboring or otherwise inapplicable version. Native FDA recall identifiers produced 14 postmarket candidate records; 14 have been reviewed (13 published, 1 rejected) and 0 remain unreviewed.

Candidate leads remain unpublished until a human confirms the exact product and tested version.

Source classes: fda ai list, fda decision summary, peer reviewed publication, vendor product page, vendor pricing observation, reimbursement policy, fda device recall, pubmed, clinical trials, openfda device recall