ImageJ change pixels values when opening 16 bit images

I have 16 bit images of Integer unsigned pixels (maximum value : 4750).
I would like to open the image to analyse and quantifie values with imageJ.
When I open thoses images with ImageJ I have a 16 bit images , but the pixels values change… 4750 becomes something like 37500… (half of a 65536).
I tried to open this files with Bio-formats importer plugin without autoscale but it doesn’t help… same result…
Why it happens? How to solve this trouble?
Thank you

Could you provide us with an example image?

The auto scale function just adjusts brightness and contrast so that you immediately see something but does not change the actual intensity values.

yes here is an example image

Could you tell us also from which system this image is coming from? The content looks quite strange. Is this the original?

I made the image by myself : I’ve done a labview program and I fill the 886X886 matrix with unsigned interger values (0 to 4570) . The image is saved as 16 bit png file
This values comes from a coordinate file ( intensity value for X,Y coordinate).

Here is also Histograms
One from NiVision Software (with the good values)
One from ImageJ (something happen during the image opening )

If I open the image in imageMagick the max intensity is also 37756.
I guess that is what the .png contains then…

Maybe the saving rescales the values, the .png is saved incorrectly (endianness??) or the histogram in the NiVision is not showing you the entire story?
But I have not experience with labview…

thank you for the answer
Probably there is a problem during the saving but i don’t understand why Ni Vision can read correctly the file (values<4700) and the other software like ImageJ can’t or take in account something else…
I will trie to save the image in an other format to see what happen .

I saved the image in Tiff format and now ImageJ reads the good values
Anyone knows what happen when saving in PNG format ?

When saving an image in Fiji as .png it uses the brightness contrast settings for minimum and maximum displayed value and rescales the grey values within this range to the entire bit range of the new image.

I guess that is what happens here as well.