Rising repeat rate and contradictory exposure indices on new DR rooms

Read the numbers first. The deviation index is defined as 10 × log10(EI / target EI). A DI of +3 means the detector received about twice the target exposure (10^0.3 ≈ 2.0); +2.4 means about 1.7 times. So the system believes these images are overexposed relative to the departmental target.

Then set it against the appearance. If the detector really were receiving twice the intended air-kerma, a correctly calibrated chain would produce a cleaner, not a grainier, image. The two observations contradict each other. That contradiction is the key finding: it points away from the exposure technique and towards the imaging chain itself.

Most likely explanations, in order.

1. Detector gain or offset calibration drift after the software upgrade - the EI is computed from a signal that no longer corresponds to dose at the detector.
2. EI target or EI calibration changed or misapplied in the upgrade - the target value, the EI algorithm, or the ROI used to compute EI may differ from what the department assumes.
3. Processing or look-up table mismatch - a chest LUT applied to a lateral lumbar spine, or an incorrect window/level preset, produces 'burnt out' soft tissue while leaving noise visible.
4. AEC chamber selection or field alignment error - a displaced chamber gives the wrong exposure and a misleading EI.

What is not supported. A genuine underexposure would give a negative DI, typically -2 to -3. Simply telling staff to increase mAs would double an already doubled dose and hide the real fault; it must be explicitly ruled out as an interim measure.

Order matters - rule out the cheap explanations before opening the detector.

1. Documentation review (radiographer/QC lead, same day). Pull the software release notes, the EI target history, the processing preset list, and the commissioning report. Confirm which EI definition and target the room is actually using and whether the upgrade reset it. Confirm whether the reject analysis codes are consistent between the two rooms - if only one room is affected, the fault is local.
2. Compare rooms. A single chest phantom image on both rooms with identical kV, mAs, SID and grid tells you in 20 minutes whether this is a system fault or a technique/education problem.
3. AEC and output consistency (medical physicist). Chamber-by-chamber reproducibility, chamber selection versus field position, kVp accuracy, mAs linearity, timer accuracy and half-value layer. This confirms whether dose delivered matches dose intended.
4. Detector dark noise, offset and flat-field uniformity (medical physicist). Uniform exposure across the detector; look for gain drift, dead pixels or a non-uniform field post-upgrade. This is the decisive test for calibration drift.
5. EI calibration with a reference dosimeter (medical physicist). Measure entrance surface dose at the detector and compare with the EI the system reports - the only way to prove EI is lying, rather than merely odd.
6. Monitor and viewing conditions (QC lead). Verify DICOM grey-scale display calibration and ambient lighting, because 'grainy' is a perceptual judgement made on a monitor.

Ownership. Tests 1, 2 and 6 are within the radiographer's and QC lead's scope; tests 3-5 must be done or supervised by a qualified medical physicist, with results filed in the QC log and copied to the vendor under the service contract.

Immediate measures (0-48 hours).

• Escalate in writing to the head of radiology, the RSO and the vendor; treat it as an equipment incident, not a staff performance issue.
• Post a technique chart for chest and lateral lumbar spine with fixed kV and mAs, and require the DI to be recorded on a simple log for every image. This gives you data with the variable removed.
• Tell radiologists explicitly that the DI on these rooms is under investigation and that a repeated image should not be assumed to mean a poor technique.
• Suspend automatic repeat requests based solely on DI; repeats should be confirmed against anatomy, positioning and motion, not against a number.
• Monitor dose-area product per room weekly and compare with the pre-upgrade baseline.

Avoiding the trap. The instinct under 'noisy image' pressure is to raise mAs. Here that would roughly double patient dose to chase a fault that is in the detector or processing chain - the definition of a wrong optimisation. State clearly in the incident record that mAs must not be increased until EI calibration is verified.

Short term. If the fault cannot be corrected within days, divert complex chest and lateral spine work to a third room or the second hospital site, and revert the software if the vendor supports a rollback.

Then fix the system. Restore the QC schedule: flat field and uniformity at least monthly and after any software change, AEC annually and after service, and a written acceptance protocol for the next upgrade. Repeat the reject analysis after three months to prove the change worked.