When the OEM Said "Pass", VERIFLUX Said Otherwise

A major NHS teaching hospital in the north of England runs functional MRI on a 3T system, with scanner stability monitored through the VERIFLUX platform from Gold Standard Phantoms, using the fBIRN-compatible Veriflux phantom. In early 2026, two consecutive scans failed.

The team also ran their scanner manufacturer's own inbuilt fMRI stability QA, which passed. So: the OEM says the scanner is fine. VERIFLUX says it isn't. Who do you believe?

When the hospital raised the discrepancy, Prof Xavier Golay, Ph.D., CEO of Gold Standard Phantoms, reviewed the stability reports himself and wrote back with a technical analysis.

He explained why the OEM result was unreliable: the manufacturer's phantom is a smaller-diameter liquid sphere with unknown fill. SNR measured on it cannot be compared directly to VERIFLUX results. He also flagged that the RDC figure indicated something was physically wrong, regardless of how SNR was being calculated. An RDC of 4.69 means correlated noise across roughly 4.5 voxels, leading to a patch of over 360 mm³ of tissue within which independence of activation will remain questionable.

Two channels in the RF coil were faulty. One had a blown fuse. The root cause: a known bug in an earlier firmware version allowed B1 amplitude to be set too high for the coil on another sequence parameter card, damaging the hardware over time.

A replacement coil was ordered. The team confirmed they would run a full VERIFLUX QA before accepting it into service.

Scanner manufacturers build QA tools for routine servicing. They use their own phantoms with their own metrics. There is no longitudinal baseline, no independent reference, and no incentive to surface failures that generate expensive call-outs.

VERIFLUX tracks each site against its own history, using the internationally established fBIRN protocol. A single low SNR reading might be noise. The same reading plotted against 18 months of stable data is a red flag. That context is exactly what the OEM tool lacks.

And when the flag goes up, GSP's response is not automated. It is a scientist who knows the physics, reviewing your specific data and telling you what it means.