Lookup tables and subsequent measurements

[Christopher Lloyd-2]

Good afternoon,

In the course of my research I need to change the greyscale values of 8-bit grey images that come
from an ultrasound machine to a new lookup table.  I've made the custom lookup table, and when I
apply this table to an image using Image -> Color -> Edit LUT... -> Open (and selecting the saved
LUT) you can immediately see that the displayed image changes to conform to the new greyscale
LUT.  However, the research involves looking at the mean greyscale value of a region of interest in
the image; application of the custom lookup table has no effect on the measurement values that
come out of ImageJ, either for the Analyze -> Measure route or for a line profile that is drawn in the
image.

So, my question is: can a lookup table be applied to an image that actually has some effect on the
measurements done on the image, or is the LUT a superficial way of viewing the images without
altering the underlying data whatsoever?  I couldn't find any matching inquiries in the archive, except
for one suggestion about going from grey to RGB and back (which didn't work, sadly).

[Manuel Jose Muñoz Martinez]

Goog aftrnoon Christopher
I'm working on a similar project, I apply a spectrum LUT over grayscale
ultrasound images. I obtain values analizyng the histogram of a ROI, pehaps
this is what yu´re looking for.

2008/4/23 Christopher Lloyd <[hidden email]>:

> Good afternoon,
>
> In the course of my research I need to change the greyscale values of
> 8-bit grey images that come
> from an ultrasound machine to a new lookup table.  I've made the custom
> lookup table, and when I
> apply this table to an image using Image -> Color -> Edit LUT... -> Open
> (and selecting the saved
> LUT) you can immediately see that the displayed image changes to conform
> to the new greyscale
> LUT.  However, the research involves looking at the mean greyscale value
> of a region of interest in
> the image; application of the custom lookup table has no effect on the
> measurement values that
> come out of ImageJ, either for the Analyze -> Measure route or for a line
> profile that is drawn in the
> image.
>
> So, my question is: can a lookup table be applied to an image that
> actually has some effect on the
> measurements done on the image, or is the LUT a superficial way of viewing
> the images without
> altering the underlying data whatsoever?  I couldn't find any matching
> inquiries in the archive, except
> for one suggestion about going from grey to RGB and back (which didn't
> work, sadly).
>

[Christopher Lloyd-2]

Using a spectrum also does not actually change anything for the measurements in ImageJ, as the
region of interest yields the same mean when you measure under the spectrum map, the standard
map, or the custom LUT I've put in (along with a few others I tried for fun).

Yes, you could derive the mean from a re-mapped distribution using the histogram.  At 30 patients
(and counting), 6-10 movies per patient, and ~100 frames per patient, that's an awful lot of
histograms (an optimistic estimate is 40 manhours of time to extract those histograms only), when
changing the actual distribution to be analyzed is, in principle, so much easier, especially
considering the analysis was as simple as hitting "m" for each frame when the data is mapped
linearly from the ultrasonic field to the greyscale of the machine (which sadly is not the case on this
new machine).  Perhaps I was mistaken.

[Volker Baecker]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello Christopher,

| altering the underlying data whatsoever?  I couldn't find any matching
inquiries in the archive, except
| for one suggestion about going from grey to RGB and back (which didn't
work, sadly).
|

This works fine for me. Did you check that weighted rgb conversions is
unchecked under Edit>Options>Conversions?

Volker Baecker
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFIEEJwxZKX7A/4oMERAhxZAKC58p9cJg06s3Xa1MRYsY6sEbu81gCgvYiJ
rRdaKWyZ+ssFPASaUIg9wwE=
=xj9c
-----END PGP SIGNATURE-----

--
passerelle antivirus du campus CNRS de Montpellier
--

[Christopher Lloyd-2]

Good morning (here, at least),

The checkbox was unselected by default, but I was playing around with the problem some more and
it turns out that order of operations is important.  If you re-mask in the 8-bit greyscale to a custom
map, then do the conversion to RGB and back to 8-bit greys, the actual pixel values change when
using measurements.  Thanks for your suggestions as it pointed me in the right direction!

[Guerda Massillon]

Hi all,

I am using a new flat-bed scanner that generates images with 48-bit  
(16 bit per color) for radiochromic film dosimetry

I want to split the image to analyse them using each color component  
individualy. How i can do that using imageJ?

Thank you in advance!

Guerda

Quoting Christopher Lloyd <[hidden email]>:

> Good morning (here, at least),
>
> The checkbox was unselected by default, but I was playing around  
> with the problem some more and
> it turns out that order of operations is important.  If you re-mask  
> in the 8-bit greyscale to a custom
> map, then do the conversion to RGB and back to 8-bit greys, the  
> actual pixel values change when
> using measurements.  Thanks for your suggestions as it pointed me in  
>  the right direction!
>
> --
> This message has been scanned for viruses and
> dangerous content by MailScanner, and is
> believed to be clean.
>
>

--
Guerda Massillon-JL, Ph.D.    *
Instituto de Física UNAM      *
A.P. 20-364                   *
Mexico 01000 DF               *
Mexico                        *
Ph:([phone]
FAX:[phone]


----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.


--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

[Wayne Rasband]

> Hi all,
>
> I am using a new flat-bed scanner that generates images with 48-bit
> (16 bit per color) for radiochromic film dosimetry
>
> I want to split the image to analyse them using each color component
> individualy. How i can do that using imageJ?

Use Image>Color>RGB Split but you may not need to split the image since
commands such as Analyze>Measure only work with the current channel.
Check "Display Label" in Analyze>Set Measurements and the channel
("Red", "Green" or "Blue") will be shown in the "Label" column in the
Results table.

-wayne

[Daniel P Mills]

Good morning, list.

I have been trying to make a large image, consisting of about 30 smaller
images, to where when you click on any one of the images in the larger
image, it goes to that place in the image stack. Does anyone know an
easy way to do this? I am using ImageJ's Create Stack from Images if
that helps...

Thanks,
Daniel P. Mills