Calibration with Image J

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Calibration with Image J

Soljanlahti Sami
Hi,

I have been trying to measure velocity of the blood flow from phase encoded
MRI images. On these images the different shades represent different
velocities. In our case black represent a flow velocity of 1.5 m/s and white
-1.5 m/s. On our last project we used NIH image but now don´t have Macs
anymore at our institution and we changed to Image J. I have done the
calibration for the measurements the same way I did with NIH image. First I
draw a white and a black spot to one of the images. Then I measure the mean
grey value of these. Then I use the calibrate command from analyze menu. I
use the straight line function.

My problem is that although white should represent one end of the scale of
pixel values in the picture and black the other there seems to be a lot more
values beyond this range. In other words when I measure my white spot it
represents mean gray value of 2558 and the black one represents 0. On the
calibration function the scale goes up 65535 which in thery is the maximum
of different shades in a 16-bit-picture. When I try to measure my pictures
after calibration the values are not even between 1.5 and -1.5.

If I change my images to 8-bit ones the calibration seems to work but my
images don´t look like they should anymore. So I think that the problem has
something to do with colour depth of the images (is it the right term when
talking about black& white images?).

I don´t know if I managed to describe my problem properly but I thank you in
advance for any suggestions that would solve the problem.

Yours,

           Sami Soljanlahti, MD
           Helsinki Medical Imaging Center
           Helsinki, Finland  
Reply | Threaded
Open this post in threaded view
|

Re: Calibration with Image J

Wayne Rasband
ImageJ supports density calibration of 16-bit images, but only using
straight line functions.

Here are the steps needed to calibrate an image so that pixels values
are in the range -1.5 m/s to 1.5 m/s:

1. Generate a histogram of the image and note that the range of values,
e.g., 1 - 4094
2. Open the Analyze>Calibrate dialog
3. Enter  "Straight Line" as the "Function:"
4. Enter "m/s" as the "Unit:"
5. Enter 1 and 4094 in the first column (raw values) and -1.5 and 1.5
in the second (values in m/s)
6. Press "OK".
7. Generate another  histogram and note that the range is now -1.5 -
1.5.

-wayne

On Nov 2, 2005, at 6:17 AM, Soljanlahti Sami wrote:

> Hi,
>
> I have been trying to measure velocity of the blood flow
> from phase encoded MRI images. On these images the different
> shades represent different velocities. In our case black
> represent a flow velocity of 1.5 m/s and white -1.5 m/s. On
> our last project we used NIH image but now don´t have Macs
> anymore at our institution and we changed to Image J. I have
> done the calibration for the measurements the same way I did
> with NIH image. First I draw a white and a black spot to one
> of the images. Then I measure the mean grey value of these.
> Then I use the calibrate command from analyze menu. I use
> the straight line function.
>
> My problem is that although white should represent one end
> of the scale of pixel values in the picture and black the
> other there seems to be a lot more values beyond this range.
> In other words when I measure my white spot it represents
> mean gray value of 2558 and the black one represents 0. On
> the calibration function the scale goes up 65535 which in
> thery is the maximum of different shades in a
> 16-bit-picture. When I try to measure my pictures after
> calibration the values are not even between 1.5 and -1.5.
>
> If I change my images to 8-bit ones the calibration seems to
> work but my images don´t look like they should anymore. So I
> think that the problem has something to do with colour depth
> of the images (is it the right term when talking about
> black& white images?).
>
> I don´t know if I managed to describe my problem properly
> but I thank you in advance for any suggestions that would
> solve the problem.
>
> Yours,
>
>            Sami Soljanlahti, MD
>            Helsinki Medical Imaging Center
>            Helsinki, Finland