About the addition of an automated quality control feature

[LuisOlivaresJ]

Hi @jfcabana and @petruong,

I enjoyed reading an article since, as mentioned in the conclusions, it would be helpful to have an automated quality control system during each reading.
I was thinking about incorporating optical filters in each reading (for example on the center of the scanner and on the sides). OMG should automatically identify the filters and compare the optical density of the filters with reference values (for example to the values during film calibration).

What do you think?

[jfcabana]

Hi Luis,
I haven't had the time to read the paper, but the idea is interesting. I'm not sure how easily feasible it would be though. Another approach that might be simpler would be the two-point calibration correction described in this paper. @petruong recently talk to me about this, I don't know if he worked on something yet? It is similar to what we are already doing when applying a correction factor from a reference film, but it also introduces a 0 dose reference to better correct the slope of the calibration curve. That should be simple to implement.

[petruong]

Hello there @LuisOlivaresJ,

I wanted to take a bit more time in order to do a better reading of the paper you referenced before responding (understanding of the applicability/advantages for utilizing optical filters in the film scan/analysis process), but as @jfcabana mentioned, we briefly discussed the incorporation of the optical density (dose) reading for the proposed zero dose reference film (from the same film calibration lot) in order to account for the aging/drift in the calibration LUT. Nothing has been implemented yet, but the brief idea would be to take the normalization factor (a = normalization film median dose / expected film dose) applied to the film dose points subtracted by the reference zero median dose (b): Corrected dose = (a * dose_film) - b
For reference, the current algorithm is as such: Corrected dose = a * dose_film

In your case for optical filter included in the scanning process (I imagine encompassing the zero and irradiated film region), the reference zero median dose (b) factor would take this into account (i.e., zero dose being the optical filter applied). Though, the biggest challenge would be if the optical density is way outside the range of the calibration LUT, in which this "b" dose would be heavily extrapolated outside calibration dose points.

I mention all this, because I plan on going on paternity leave in the coming weeks, so no progress on this reference zero dose factor on my part for the time being.

Cheers,
Peter

[LuisOlivaresJ]

Hi,

The idea of using optical filters is to have a quality control mechanism to identify potential errors during the scanning process (for example if auto correction of color or brightness has been applied unconsciously). Also, as a way to measure inter-scan reproducibility of the lamp/detector system. Optical density measurements of the filters should not be used to create corrections factors.
By the other hand, as you said, to account for aging/drift and also for inter-scan variations, the use of the optical density of the zero dose film is a better approach.

[jfcabana]

Ah I see what you're talking about now. Yes using optical filters just as a reference for detecting potential errors might be a good idea. A downside would be that it would require to always scan a larger area of the scanner to encompass this filter, but it is probably not a big deal.