Automatically determine the qualified slice factor?

[smiths]

I think there should be a way to determine if the qualified slice factor is definitely wrong. I haven't experimented much, but when the value was too low, only a few slices were segmented. If we can find the volume of the volume rendering, then we could check to see if the volume is in a reasonable range. If not we could try a different qualified slice factor.

[JovieL25]

This should be easily detected by getting the Dice similarity coefficient between the cropped volume and the segmented volume. The coefficient should be at least higher than the threshold.

An even faster way to detect it is that the first few slices to be segmented should at least be qualified. I think that the qualification is not really to "qualify" a segmented slice but to eliminate the slice that includes parts that are not aorta. It is more likely an upper bound.

[smiths]

This sounds good @JovieL25. This could be another example where you improve the algorithm and show that the CI tests continue to pass. We can postpone doing this depending on how much time we have, but it would be a nice little addition. 😄