Do I have to perform Image normalization before division ?

Hi all

I am trying to estimate left ventricle ejection fraction 'LVEF' from two images derived from a seriese of images (maximum and minimum image). The LVEF is calculated by:

LVEF = (max image - minimum image)/ maximum image. *

*Please note that maximum and minimum iamges have different grey values and different pixel counts.

Do I have to do image normalisation before doing this calculation?. I there any consideration that i should take into account before dividing images to extract a parameter ?.

Best,

Ahmad

[Walter Roberson]

It depends what kind of image normalization you are referring to, but Probably Yes, image normalization would be needed if you are dealing with optical images.

If you are dealing with readings for non-optical sensors, then whether you need to normalize or not depends on the technology involved.

[Ahmad Alenezi]

Hi Walter

Thanks for your reply

The images were captured using nuclear medicine scintilation detectors (Gamma Camera). They are a cameras that work by detecting gamma rays emitted from the subject of interest. The produced images are black and white intesity images.

Do you think normalisation is required for this kinf of images ?

Best,

Answer 1

[Walter Roberson]

The implication of doing that kind of calculation is that when the ventricle is in one state, the density of the muscle blocks the radiation source more than when ventricle is in the other state when the ventricle is open with blood.

And that in turn implies that different patients might have different densities or different amounts of (whatever) to go through, and that therefore the numbers being seen are not absolute numbers but rather relative numbers. Or perhaps the numbers are absolute numbers (e.g., a particular brightness might represent a particular absolute radiation count), but that the significance of the numbers are relative.

When you have relative numbers or numbers of relative significance, then you should normalize.

When you have numbers of absolute significance you probably should not normalize.