jaded image programmer, ImageJ newbie, needs leg up

[George W. Sherouse, Ph.D.]

I am looking at the feasibility of implementing a set of image-based  
tools for Radiation Oncology in ImageJ.  I am an experienced(!)  
programmer in this arena but a novice with both ImageJ and Java.  
I've spent some time the last week or so trying to get the flavor of  
ImageJ for the following proof-of-concept application and have been  
stymied.  I'd appreciate any guidance as to whether ImageJ is the  
right platform for me and if so how to get it to do what I want.

My choice of a proof is to implement the so-called Winston-Lutz test  
for stereotactic radiosurgery.  The test involves a film image that  
has a darkened (exposed) circle about 2.5 cm in diameter and  
nominally centered in that the lighter shadow of a steel sphere,  
about 0.8 cm.  The evaluation requires determining the distance, or  
better the displacement vector, between the centers of two circles,  
one circumscribing the dark shadow and the other circumscribing the  
small light shadow.

I imagine a user interface in which the user places graphical circles  
by hand to visually match the image features and the distance is  
displayed.  Displaying the distance interactively, both graphically  
and in text, would be best, but reporting the distance out as a  
result triggered by some GUI element would be OK.

OvalROI has a lot of exactly the right functionality needed for  
placing the circles (except for the lack of "constrain" to keep the  
OvalROI circular).  Unfortunately ROI seems to be "one per image" and  
so I cannot place two independently on the same image.  I also don't  
see how to create an interactive feedback mechanism from a Plugin.

I can send a sample image to anyone who cares.

Help?

- GWS

=======================================================================
  Sherouse Systems, Inc., Chapel Hill, NC, <http://www.gwsherouse.com/>
      Medical Physics and Computing services for Radiation Oncology
         [phone] voice or FAX, <mailto:[hidden email]>

                           Tools not rules.
=======================================================================

[Duane and Julie]

GWS,

I don't know if this is what you're looking for, but I put lots of  
shapes on images as an overlay using a custom canvas.  You may be  
able to just do the ROI overlays yourself, trap the mouse and  
keyboard events, and provide exactly the GUI you want.

In your case you'll have only two shapes to overlay, both of which  
are circles.

cheers!

    duane

On Mar 17, 2006, at 4:12 AM, George W. Sherouse, Ph.D. wrote:

> I am looking at the feasibility of implementing a set of image-
> based tools for Radiation Oncology in ImageJ.  I am an experienced
> (!) programmer in this arena but a novice with both ImageJ and  
> Java.  I've spent some time the last week or so trying to get the  
> flavor of ImageJ for the following proof-of-concept application and  
> have been stymied.  I'd appreciate any guidance as to whether  
> ImageJ is the right platform for me and if so how to get it to do  
> what I want.
>
> My choice of a proof is to implement the so-called Winston-Lutz  
> test for stereotactic radiosurgery.  The test involves a film image  
> that has a darkened (exposed) circle about 2.5 cm in diameter and  
> nominally centered in that the lighter shadow of a steel sphere,  
> about 0.8 cm.  The evaluation requires determining the distance, or  
> better the displacement vector, between the centers of two circles,  
> one circumscribing the dark shadow and the other circumscribing the  
> small light shadow.
>
> I imagine a user interface in which the user places graphical  
> circles by hand to visually match the image features and the  
> distance is displayed.  Displaying the distance interactively, both  
> graphically and in text, would be best, but reporting the distance  
> out as a result triggered by some GUI element would be OK.
>
> OvalROI has a lot of exactly the right functionality needed for  
> placing the circles (except for the lack of "constrain" to keep the  
> OvalROI circular).  Unfortunately ROI seems to be "one per image"  
> and so I cannot place two independently on the same image.  I also  
> don't see how to create an interactive feedback mechanism from a  
> Plugin.
>
> I can send a sample image to anyone who cares.
>
> Help?
>
> - GWS
>
> ======================================================================
> =
>  Sherouse Systems, Inc., Chapel Hill, NC, <http://www.gwsherouse.com/>
>      Medical Physics and Computing services for Radiation Oncology
>         [phone] voice or FAX, <mailto:[hidden email]>
>
>                           Tools not rules.
> ======================================================================
> =

[Jonathan Hilmer]

I don't know much about progamming the type of interface you want for
ImageJ, but I don't think that circular ROI (based upon center and
radius) are the way to go for a user interface.  The problem is that
there are too many corrections necessary to get the placement just
right.

Although it would complicate your programming a little bit, I would
recommend spatial determination of your objects by having the user
place points on the circumferance: three clicks and it's done.



On 3/19/06, Duane & Julie <[hidden email]> wrote:

> GWS,
>
> I don't know if this is what you're looking for, but I put lots of
> shapes on images as an overlay using a custom canvas.  You may be
> able to just do the ROI overlays yourself, trap the mouse and
> keyboard events, and provide exactly the GUI you want.
>
> In your case you'll have only two shapes to overlay, both of which
> are circles.
>
> cheers!
>
>     duane
>
> On Mar 17, 2006, at 4:12 AM, George W. Sherouse, Ph.D. wrote:
>
> > I am looking at the feasibility of implementing a set of image-
> > based tools for Radiation Oncology in ImageJ.  I am an experienced
> > (!) programmer in this arena but a novice with both ImageJ and
> > Java.  I've spent some time the last week or so trying to get the
> > flavor of ImageJ for the following proof-of-concept application and
> > have been stymied.  I'd appreciate any guidance as to whether
> > ImageJ is the right platform for me and if so how to get it to do
> > what I want.
> >
> > My choice of a proof is to implement the so-called Winston-Lutz
> > test for stereotactic radiosurgery.  The test involves a film image
> > that has a darkened (exposed) circle about 2.5 cm in diameter and
> > nominally centered in that the lighter shadow of a steel sphere,
> > about 0.8 cm.  The evaluation requires determining the distance, or
> > better the displacement vector, between the centers of two circles,
> > one circumscribing the dark shadow and the other circumscribing the
> > small light shadow.
> >
> > I imagine a user interface in which the user places graphical
> > circles by hand to visually match the image features and the
> > distance is displayed.  Displaying the distance interactively, both
> > graphically and in text, would be best, but reporting the distance
> > out as a result triggered by some GUI element would be OK.
> >
> > OvalROI has a lot of exactly the right functionality needed for
> > placing the circles (except for the lack of "constrain" to keep the
> > OvalROI circular).  Unfortunately ROI seems to be "one per image"
> > and so I cannot place two independently on the same image.  I also
> > don't see how to create an interactive feedback mechanism from a
> > Plugin.
> >
> > I can send a sample image to anyone who cares.
> >
> > Help?
> >
> > - GWS
> >
> > ======================================================================
> > =
> >  Sherouse Systems, Inc., Chapel Hill, NC, <http://www.gwsherouse.com/>
> >      Medical Physics and Computing services for Radiation Oncology
> >         [phone] voice or FAX, <mailto:[hidden email]>
> >
> >                           Tools not rules.
> > ======================================================================
> > =
>