Hi Vikram,
these are very good questions.
One issue is where you plan to embed your dye phantom (e.g. water? agarose?),
since you can get all sorts of artifacts if you just stick it in the scanner on
its own, and, related to this, how thick you intend to make the wall, since the
resulting signal will be affected by it. Unless you tune things in such a way
that the wall has a similar signal as your embedding, you'll be validating a
situation where the wall is thicker and has more contrast with respect to the
lumen than in-vivo.
For your purposes silicone tubing would be probably easier to handle, but it
depends how thin a wall you can manufacture.
Note that it doesn't necessarily have to be the size of a mouse vessel: it's
enough that the ratio between the acquisition resolution (which is the raw
resolution, without interpolation or zero padding in k-space, of course) and
the diameter are correct.
Another way to go (way easier but not necessarily as compelling - although how
compelling an experiment may be is not necessarily related to the fact that
it's physical or virtual) is to simulate the effect of resolution on MR images.
There are ways to produce synthetic images, but you have to make sure what you
are simulating reproduces the features and point spread function of your actual
acquisition sequences.
There are several papers out there that describe these kinds of validation
studies. I'd make sure to read them thoroughly before taking your first steps.
Experiments take a lot of time and effort, and you want to make sure to plan
them out properly.
Luca
On Feb 10, 2011, at 12:18 PM, Vikram Mehta wrote:
> Hi,
>
> Thanks for your mail. I had it in the back of my mind to make a phantom but I
> thought I'd try and find out if there was another way to do it.
> I guess a simple phantom should give a much better insight into this matter
> and I will try and do this soon.
>
> what I was not sure about was what material to use for the phantom ? Any
> suggestions there ? do you think a glass capillary filled with dye would
> suffice or something more flexible, like silicone tubing ?
>
> Please let me know what you think.
>
> Regards,
> Vikram.
>
>
> On Wed, Feb 9, 2011 at 8:58 AM, Luca Antiga <luca.ant...@gmail.com> wrote:
> Hi Vikram,
> we are actually looking into the issue of small vessels as part of our
> research with David.
> It all depends on the point spread function of your acquisition system and
> how large is
> the vessel with respect to it.
> What we are observing (that is already well known to the field) is that the
> usual gradient-based
> deformable models tend to underestimate the size of vessels on MR as they
> approach the size
> of the acquisition resolution.
> The workaround with vmtk is to use the -featureimagetype upwind option in
> vmtklevelsetsegmentation,
> which tends to avoid severe underestimation of vessel size at the price of
> some overestimation.
>
> Having said this, Daniel is right, you really need a ground truth, and if you
> have a mouse scanner
> at your disposal the best is to build a simple phantom and scan it at
> different resolutions. This
> way you will be sure that you are tuning the segmentation method to your
> specific acquisition
> modality. Even talking about MR may not make sense in itself: you need to
> know what sequence
> you are using, since, e.g. phase contrast vs contrast-enhanced vs k-centric
> contrast-enhanced
> sequences may lead to different signals for the same shape.
>
> If you do anything like this it would be great if you let me know about it,
> as we are very interested
> in these kinds of issues.
>
> Best regards
>
>
> Luca
>
>
>
> On Feb 8, 2011, at 6:39 PM, Daniel Haehn wrote:
>
> > Hi Vikram,
> >
> > I did something similar during my Master's thesis. You can find it here
> > http://www.spl.harvard.edu/publications/item/view/1957
> >
> > You need an expert segmentation as a ground truth and a comparison metric
> > to do so.
> >
> > Cheers,
> > Daniel
> >
> > On Jan 31, 2011, at 6:43 AM, Vikram Mehta wrote:
> >
> >> Hello,
> >>
> >> I am learning to use vmtk to segment and extract carotid arteries from
> >> MRI/CT data of mice.
> >> I am managing to get the surface model but as you know, the diameters of
> >> mouse arteries are really small and I would like to know if the model that
> >> VMTK produces is accurate in terms of diameter and other dimensions ?
> >>
> >> Is there a way to measure the error (if any) in the diameter of the model
> >> with respect to the original ?
> >>
> >> Thank you in advance
> >> Vikram.
> >>
> >>
> >> ------------------------------------------------------------------------------
> >> Special Offer-- Download ArcSight Logger for FREE (a $49 USD value)!
> >> Finally, a world-class log management solution at an even better
> >> price-free!
> >> Download using promo code Free_Logger_4_Dev2Dev. Offer expires
> >> February 28th, so secure your free ArcSight Logger TODAY!
> >> http://p.sf.net/sfu/arcsight-sfd2d_______________________________________________
> >> vmtk-users mailing list
> >> vmtk-users@lists.sourceforge.net
> >> https://lists.sourceforge.net/lists/listinfo/vmtk-users
> >
> >
> >
> > The information contained in this e-mail message is intended only for the
> > personal and confidential use of the recipient(s) named above. If the
> > reader of this message is not the intended recipient or an agent
> > responsible for delivering it to the intended recipient, you are hereby
> > notified that you have received this document in error and that any review,
> > dissemination, distribution, or copying of this message is strictly
> > prohibited. If you have received this communication in error, please notify
> > us immediately by e-mail, and delete the original message.
> >
> > ------------------------------------------------------------------------------
> > The ultimate all-in-one performance toolkit: Intel(R) Parallel Studio XE:
> > Pinpoint memory and threading errors before they happen.
> > Find and fix more than 250 security defects in the development cycle.
> > Locate bottlenecks in serial and parallel code that limit performance.
> > http://p.sf.net/sfu/intel-dev2devfeb
> > _______________________________________________
> > vmtk-users mailing list
> > vmtk-users@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/vmtk-users
>
>
------------------------------------------------------------------------------
The ultimate all-in-one performance toolkit: Intel(R) Parallel Studio XE:
Pinpoint memory and threading errors before they happen.
Find and fix more than 250 security defects in the development cycle.
Locate bottlenecks in serial and parallel code that limit performance.
http://p.sf.net/sfu/intel-dev2devfeb
_______________________________________________
vmtk-users mailing list
vmtk-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/vmtk-users
