Couple of double negatives in there, but I assume you meant that you can't imagine machines that do not fail safes? The answer is they do, but are easily overridden with the click of a mouse or with a "research" key. Usually the operator gets a benign message such as:
"Warning, you are setting this machine outputs above the recommended manufacturer presets. Proceed?"
Not:
"Continuing with these inputs will result in a twofold increase in the lifetime risk for cancer in this patient, and exceeds the ICRP limits. Proceed?"
Obviously if I, or some with my background, were at the controls we would abort or reset the machine even with the first message. A dental tech? A dentist? , a radiologist? Odds are much lower. Remote actually, since the image must be obtained for care to proceed right?
Additionally, the scans can be repeated.......ad-infinitum. This is where the danger really multiplies. Very few people outside of radiation protection experts really understand or appreciate the risks involved with CT or x-ray, much less with repeating those scans. I'd be willing to be that most if not all of those cases were due to "re-scanning" multiple times.
Detection of miscalibrations is not something that's done for each scan. It's part of daily QA that physicists perform in radiology and radiation oncology departments, usually in the morning before the machines are used clinically. Machine outputs are tested using ionization chambers and the outputs are verified/compared with those on the treatment monitors.
Keeping a dosimeter in the FOV or treatment field perturbs it (think of Schrodingers cat here). This leads to dose inaccuracies and in geneities in the case of a linear accelerator or gamma knife, etc. It degrades image quality in the case of a CT or x-ray.
We'd need a non-ionizing imaging modality that offers the same contrast as ionizing modalities between materials with higher and lower Z. Do that, and the case is closed. You're also set for life.
