[Insight-users] mi registration problem

[Miller, James V (Research)]

You can apply any of the image metrics without using the
registration framework.  I have used this to evaluate the
various metrics in various contexts.  You give one of the
metrics two images and the regions within the images to 
apply the metric.  

In one application I was working on I had to tighten the 
region for MI to be as small as possible.  Then I moved this 
region around and saw how the metric changed.

Jim



> -----Original Message-----
> From: Dill, John [mailto:john-dill at uiowa . edu]
> Sent: Thursday, October 09, 2003 12:28 PM
> To: 'Miller, James V (Research)'; 'insight-users at itk . org'
> Subject: RE: [Insight-users] mi registration problem
> 
> 
> I was wondering about maybe taking a window from the drr 
> image that I can
> see has a counterpart in the portfilm, and moving that drr 
> window over the
> whole portfilm image, finding the one with the best match 
> according to the
> mutual information criteria.  Does that seem like a 
> reasonable possibility?
> The problem seems to stem from the different information in 
> the projections.
> Is there an algorithm like that I can try already in itk?
> 
> John
> 
> > -----Original Message-----
> > From: Miller, James V (Research) [mailto:millerjv at crd . ge . com]
> > Sent: Thursday, October 09, 2003 9:25 AM
> > To: 'Dill, John'; 'insight-users at itk . org'
> > Subject: RE: [Insight-users] mi registration problem
> > 
> > 
> > The pattern intensity metric is supposedly robust to devices in the
> > field.  It uses a non-quadratic error term so the mismatch 
> of the bar 
> > to CT should not affect the alignment. This metric allows the 
> > mismatch 
> > of intensities can be regional.  MI requires a globally 
> > consistent mapping
> > of intensities.
> > 
> > Jim
> > 
> > 
> > 
> > > -----Original Message-----
> > > From: Dill, John [mailto:john-dill at uiowa . edu]
> > > Sent: Wednesday, October 08, 2003 11:38 AM
> > > To: 'insight-users at itk . org'
> > > Subject: [Insight-users] mi registration problem
> > > 
> > > 
> > > I have been having a problem with registering portfilm and 
> > > drr images.  In
> > > my case, the algorithm registers the lateral projections just 
> > > fine, but in
> > > ap cases, I get mis-registrations.  The problem seems to be 
> > > from the fact
> > > that when the patient is on the treatment table, there is a 
> > metal bar
> > > attached to the table which is present in the portfilm, but 
> > > not in the CT,
> > > so the DRR doesn't have the corresponding bar in its 
> > > projection.  The bar
> > > shows up as roughly 10-15 pixels wide, and as tall as the 
> > > image.  I have
> > > tried some different filtering ideas, but can not come up 
> > with a good
> > > registration.  Does anyone have ideas on how to reduce this 
> > > bar's effects
> > > (automatically if possible)?
> > > 
> > > Thanks,
> > > John
> > > _______________________________________________
> > > Insight-users mailing list
> > > Insight-users at itk . org
> > > http://www . itk . org/mailman/listinfo/insight-users
> > > 
> > 
>