Display on a 12bit monochrome monitor

[Marcel H.]

Hi list members,

I have a 12bit monochrome Dicom image.
If I had a radiologic monitor and a professional graphics board that both support support 12 bit grayscale, would it be possible to display more than 8 bit with ImageJ?
Or will the image be sampled down to 8 bit also on a 12 bit screen?

If it is not possible with ImageJ, is there another way with Java to display more that 2^8 gray scales with that professional radiology hardware?

Or is that only possible with radiology workstations?

A lot of questions, thanks for your help!

Marcel H.

[Michael Schmid]

Hi Marcel,

with ImageJ, display grayscale resolution is 8 bits. The concept of  
Java seems to be open for a larger number of colors, but I am not  
aware of any implementation that would go beyond 8 bits.

Of course, by adjusting Image>Adjust>Brightness&Contrast, you can see  
finer grayscale shades of a 16-bit or float image, but that will make  
other areas saturated white or black.

Michael
________________________________________________________________

On 22 Jan 2010, at 09:56, Marcel H. wrote:

> Hi list members,
>
> I have a 12bit monochrome Dicom image.
> If I had a radiologic monitor and a professional graphics board  
> that both
> support support 12 bit grayscale, would it be possible to display  
> more than
> 8 bit with ImageJ?
> Or will the image be sampled down to 8 bit also on a 12 bit screen?
>
> If it is not possible with ImageJ, is there another way with Java  
> to display
> more that 2^8 gray scales with that professional radiology hardware?
>
> Or is that only possible with radiology workstations?
>
> A lot of questions, thanks for your help!
>
> Marcel H.

[Gluender-3]

Marcel,

I profoundly doubt that you will find a monitor that is able to
display 12 bit gray levels (of course you may find monitors that are
said to display 12 bit). 10 bit is already very difficult to achieve.

>Hi list members,
>
>I have a 12bit monochrome Dicom image.
>If I had a radiologic monitor and a professional graphics board that both
>support support 12 bit grayscale, would it be possible to display more than
>8 bit with ImageJ?
>Or will the image be sampled down to 8 bit also on a 12 bit screen?
>
>If it is not possible with ImageJ, is there another way with Java to display
>more that 2^8 gray scales with that professional radiology hardware?
>
>Or is that only possible with radiology workstations?
>
>A lot of questions, thanks for your help!
>
>Marcel H.
>--
>View this message in context:
>http://n2.nabble.com/Display-on-a-12bit-monochrome-monitor-tp4438960p4438960.html
>Sent from the ImageJ mailing list archive at Nabble.com.

Best
--

                   Herbie

          ------------------------
          <http://www.gluender.de>

[Gabriel Lapointe-2]

Hi Marcel,
The average human eyes are barely able to see the difference between
images with 128 and 256 levels of gray. Therefore I seriously doubt
there is a gain in going to a 12 bit grayscale monitor anyway.

Slide 14 of the following presentation contains images with various
level of gray (up to 256 after I just reused the 8 bit one). It's in
French but it's the numbers that are important in understanding the
slide.
http://gabriellapointe.ca/imagej/presentations/grsnc2009.php 

Gabriel Lapointe, M.Sc.
Laboratoire de Luc DesGroseillers, Ph.D.
Pavillon Roger-Gaudry Local A-538
Département de biochimie
Faculté de Médecine de l'Université de Montréal
2900 boul. Édouard-Montpetit,
Montréal, Qc, H3T 1J4
[hidden email]
http://gabriellapointe.ca 


On Fri, 2010-01-22 at 00:56 -0800, Marcel H. wrote:

> Hi list members,
>
> I have a 12bit monochrome Dicom image.
> If I had a radiologic monitor and a professional graphics board that both
> support support 12 bit grayscale, would it be possible to display more than
> 8 bit with ImageJ?
> Or will the image be sampled down to 8 bit also on a 12 bit screen?
>
> If it is not possible with ImageJ, is there another way with Java to display
> more that 2^8 gray scales with that professional radiology hardware?
>
> Or is that only possible with radiology workstations?
>
> A lot of questions, thanks for your help!
>
> Marcel H.

[Marcel H.]

Gabriel Lapointe-2 wrote

Hi Marcel,
The average human eyes are barely able to see the difference between
images with 128 and 256 levels of gray. Therefore I seriously doubt
there is a gain in going to a 12 bit grayscale monitor anyway.

Hi Gabriel,

I am doing investigations for the development of a radiologic tool. In radiology it is quite common to use 10 or 12 bit monitors (Eizo, Barco, etc.) and the radiologists swear that they can see the difference ;-)

My intention is to find out if I can use the common java graphic and imaging components for such a 10-12 bit radiology tool.

[Gluender-3]

Marcel,

monitors specified as 10 bit will display true 8 bit and this makes
the difference a trained radiologist may be able to recognize. 12 bit
means more than 4000 shades of gray (i.e. 244 ppm) and poses enormous
technical problems to realize (near to impossible today with LCDs),
not to speak of the human eye.

In general even radiologists work with a grayscale expansion or
"loup" and CT-radiologists did so from the very beginning of this
technique.

>Gabriel Lapointe-2 wrote:
>>
>>  Hi Marcel,
>>  The average human eyes are barely able to see the difference between
>>  images with 128 and 256 levels of gray. Therefore I seriously doubt
>>  there is a gain in going to a 12 bit grayscale monitor anyway.
>>
>
>
>Hi Gabriel,
>
>I am doing investigations for the development of a radiologic tool. In
>radiology it is quite common to use 10 or 12 bit monitors (Eizo, Barco,
>etc.) and the radiologists swear that they can see the difference ;-)
>
>My intention is to find out if I can use the common java graphic and imaging
>components for such a 10-12 bit radiology tool.

AFAIK ImageJ doesn't use SUN-Java image processing routines but deals
various image formats, 16 bit and 32 bit float included. That said,
you may consider ImageJ 16 bit routines for your purposes.

Best
--

                   Herbie

          ------------------------
          <http://www.gluender.de>

[Marcel H.]

Print products (e.g. the good old analogue BW photography paper) do have a lot wider gray scale ranges than monitors and you are able to see that. Therefore I do not see the point why the human eye should be a limiting factor.

After further investigations I found this article from NVidia. They have a solution for 12 bit but apparently only for OpenGl:

http://www.nvidia.com/docs/IO/40049/TB-04631-001_v01.pdf

[migue]

Even if it isn't strictly subject related

Sure that human vision can distinguish more than 8 bit grayscale images!!!. That what human vision can't do is to distinguish more than 128/256 gray levels simultaneously (i.e. brain has much more narrower dynamic range that eyes who must adapt them to the average luminance of the field of view).

[Gluender-3]

Hello,

it appears as if you're approaching the facts...

>Even if it isn't strictly subject related
>
>Sure that human vision can distinguish more than 8 bit grayscale images!!!.
>That what human vision can't do is to distinguish more than 128/256 gray
>levels simultaneously (i.e. brain has much more narrower dynamic range that
>eyes who must adapt them to the average luminance of the field of view).

According to the classic measurements of Steinhardt (1936), humans
show a threshold for light density (luminance) discrimination (just
noticeable difference) of at best 1% (Weber fraction).
"At best" here implies a mean density of around 10^4 Troland and a
test area of larger than 25 degrees of visual angle.
For test fields of 0.5 degrees of visual angle the Weber fraction
approaches 10%.

With respect to the earlier post of Marcel:

"Print products (e.g. the good old analogue BW photography paper) do have a
lot wider gray scale ranges than monitors and you are able to see that."

...I should like to add that in principle b&w-photographic material
shows essentially a binary behaviour at least on the microscopic
level. (Facts to think about...)

Best
--

                   Herbie

          ------------------------
          <http://www.gluender.de>

[Michael Schmid]

Hi everyone,

a few further remarks on this never-ending topic:

(1) The transfer curve is nonlinear, roughly quadratic (it would be  
quadratic at gamma=2, see http://en.wikipedia.org/wiki/ 
Gamma_correction). With a quadratic curve, the difference between  
gray levels 254 and 255 is roughly 0.8%, no problem. The difference  
between 127 and 128 is 1.6%, which is noticeable, and it becomes  
worse at lower levels.
That's also why sRGB uses a lower effective gamma at lower light levels.

(2) LCD monitors connected by analog (VGA) signals cause additional  
problems due to D/A and A/D conversion. If you create a grayscale map  
(new image, 8-bit, 512*512 pixels, fill with ramp) you can easily see  
artifacts on almost all monitors.

Even without the additional D/A+A/D conversion there are often  
artifacts. If I try on my iMac with built-in LCD screen, apart form  
the weak steps at low intensities (almost invisible above approx.  
gray level 130) I can clearly see vertical stripes on such a zoomed  
ramp image, especially when zooming in and panning the image (space  
bar down). Careful Monitor calibration (Control panel  
'Displays'>Color, expert mode) leads to only a slight improvement.

So in my view, the best one can do (apart from ultraexpensive  
radiology systems) is avoiding analog (VGA) connections and checking  
a ramp image before selecting a graphics card and monitor.


Michael
________________________________________________________________

On 26 Jan 2010, at 13:51, Gluender wrote:

> Hello,
>
> it appears as if you're approaching the facts...
>
>> Even if it isn't strictly subject related
>>
>> Sure that human vision can distinguish more than 8 bit grayscale  
>> images!!!.
>> That what human vision can't do is to distinguish more than  
>> 128/256 gray
>> levels simultaneously (i.e. brain has much more narrower dynamic  
>> range that
>> eyes who must adapt them to the average luminance of the field of  
>> view).
>
> According to the classic measurements of Steinhardt (1936), humans  
> show a threshold for light density (luminance) discrimination (just  
> noticeable difference) of at best 1% (Weber fraction).
> "At best" here implies a mean density of around 10^4 Troland and a  
> test area of larger than 25 degrees of visual angle.
> For test fields of 0.5 degrees of visual angle the Weber fraction  
> approaches 10%.
>
> With respect to the earlier post of Marcel:
>
> "Print products (e.g. the good old analogue BW photography paper)  
> do have a
> lot wider gray scale ranges than monitors and you are able to see  
> that."
>
> ...I should like to add that in principle b&w-photographic material  
> shows essentially a binary behaviour at least on the microscopic  
> level. (Facts to think about...)
>
> Best
> --
>
>                   Herbie
>
>          ------------------------
>          <http://www.gluender.de>

[thenork-2]

I know this thread is quite old, but it still comes up on a google search.  I
wrote a plugin for imagej which allows the use of all the bits on 10-bit or
higher monitors.
https://sites.imagej.net/JOGLCanvas/
https://github.com/aschain/JOGL_Canvas_DC3D



--
Sent from: http://imagej.1557.x6.nabble.com/

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html