Radiation-safety features in interventional imaging are almost always pitched at the patient. Dose to the patient is what gets recorded, reported, benchmarked and regulated, and it is the number a hospital can point at. US20260221372A1, published 30 July 2026 and assigned to Canon Kabushiki Kaisha, is directed somewhere else. Its claims are about the dose reaching the person operating the equipment.

The recited system feeds captured images to an image recognition model configured to determine whether a subject of a user is present in the path of the X-ray beam, and on detection reduces the radiation provided to that subject by adjusting collimation, inserting a filter, or changing imaging parameters. Claim 2 gives the parameter route as reducing tube current or reducing frame rate. Once a predefined condition is met, the system raises the radiation reaching the detector again.

Why the target matters commercially

Occupational exposure in interventional cardiology and radiology is a long-running workforce issue rather than a per-procedure metric. It accumulates across a career, it attaches to the individual rather than the case, and the existing mitigations are physical and procedural — leaded aprons and gloves, shields, table drapes, and training about hand placement. Those are consumables and habits, not capabilities of the imaging system, and they are typically the hospital's problem rather than the vendor's.

A filing that moves that mitigation into the imaging chain changes who owns the problem. It also changes what a purchaser is evaluating: not just image quality per unit of patient dose, but whether the machine actively manages the exposure of the people standing next to it. The record makes no market claim of any kind, and none should be read into it — but the choice of what to protect is itself the signal.

A data processing system as claimed in claim 12, wherein the action to control the imaging conditions comprises preventing an automatic exposure control system from increasing an intensity of the X-rays produced by the X-ray imaging apparatus.— System and Method for Adjusting Fluoroscopy Imaging Conditions, US20260221372A1

Claim 13, quoted above, is the one worth dwelling on because it describes a problem the existing safety equipment creates. Automatic exposure control increases X-ray output when the signal reaching the detector falls. A dense object in the beam makes that signal fall. A leaded glove is a dense object. Under claim 12, an ungloved hand and a gloved hand trigger different actions, and claim 14 assigns a separate recognition model to each. For the gloved case the recited action is not to dim the beam but to stop the exposure-control loop from compensating — the protective equipment, left unmanaged, provokes the machine into producing more radiation.

There is a second-order effect worth naming. A machine that logs when it detected an operator in the beam, and what it did about it, is generating occupational-exposure evidence that did not previously exist in the imaging chain. The claims recite no logging, no reporting and no record-keeping, so this is not something the document covers — but detection systems tend to produce data as a by-product, and dose to staff has historically been measured by personal dosimeter badges rather than by the equipment. Any shift in where that measurement originates would be consequential for how occupational exposure is monitored, and it is worth watching whether later filings in the family claim it.

What the record does and does not establish

Several limits belong in the same breath. This is a published application, not a granted patent: unenforceable in this form, unexamined on the public record, and subject to narrowing. It discloses no clinical data, no dose-reduction figure, no test result, and no regulatory status. It names no product, no programme and no timeline. Nothing here indicates a cleared or marketed device, and a filing establishes only that a problem was considered worth protecting a solution to.

It is also worth being precise that the mechanism recited is detection and control, not shielding. The claims never recite a performance level. Claim 5 adjusts collimation based on the reported location of the detected subject to prevent at least part of that location from being irradiated, and claim 6 conditions that adjustment on the region of the object under examination continuing to be irradiated — the system is constrained not to protect the operator by depriving the clinician of the image. Whether that balance holds in practice is an empirical question this document does not answer.

Two structural features suggest the filing anticipates being judged on image quality anyway. Claims 9 and 10 apply a denoising filter or a noise-reduction machine learning model to frames captured while the dose is reduced, and gate that post-processing on an image quality parameter falling below a threshold within an identified region of interest. Claim 15 lists what restores normal operation: failure to detect the operator's part in subsequent images, a change in capture angle, user input, or expiry of a time limit. The design assumption is that the interruption is brief and the picture has to survive it.

One reading caution. The abstract describes "an unintended part of an operator (e.g. a hand)" while every claim recites "a subject of a user"; only claim 12 ties that subject to a hand specifically. The two vocabularies should not be merged when describing what is covered. The record also carries published-text slips reproduced rather than repaired, including "adjust imaging conditions including one or more:" without an "of" in all three independent claims.

The application arrives in a drop with a cluster of adjacent imaging and device-workflow records, including boundary correction in blood-vessel tomographic images, demagnetisation of tools in an electromagnetically tracked endoscope and an interface moving heart-pump controller data into hospital records. Different problems, one direction of travel: the instrumentation around the procedure is being asked to notice more of what is happening in the room.