map model to volume

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

map model to volume

Dr. András Jakab
Dear Slicer users,

Recent Slicer binaries have the feature of "ModelIntoLabelvolume" which is exactly what I needed, as the subject of this message implies.

Albeit the application of this module seems to be problematic in my case. I would like to inquire if any of you know a better solution for the following problems.

When mapping a model into a volume of high resolution (very high indeed - I need this for thalamus nuclei, 0.25 x 0.25 x 0.5mm) unexpected crashes occur, without any error messages. This is even more pronounced when trying to modify the labelmap with Editor rendering it impossible to modify the results of the ModelIntoLAbelvolume. This must be a memory issue. (Linux binary, latest ones)

It would be even better if the algorithm would not exclusively map the vertices into the volume, as this gives a "hollow" result which is very hard to convert to "filled" labelmaps. This is especially true in the Z axis, I have a couple of slices which are more or less empty.

Is there a possibility (maybe outside Slicer) to "fill" a .vtk mesh and map what is inside onto a high res image without running into severee memory problems?

Thank you very much
András Jakab

_______________________________________________
slicer-users mailing list
[hidden email]
http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
To unsubscribe: send email to [hidden email] with unsubscribe as the subject
Reply | Threaded
Open this post in threaded view
|

Re: map model to volume

pieper
Administrator
Hi Andras -

Did you try the PolyData to Label Map module?  It's similar, but it does
try to fill in the interiors for you.  The two modules were written for
different purposes and probably should be merged into a single interface
at some point.

HTH,
Steve

On 02/22/2011 04:12 AM, András Jakab MD wrote:

> Dear Slicer users,
>
> Recent Slicer binaries have the feature of "ModelIntoLabelvolume" which
> is exactly what I needed, as the subject of this message implies.
>
> Albeit the application of this module seems to be problematic in my
> case. I would like to inquire if any of you know a better solution for
> the following problems.
>
> When mapping a model into a volume of high resolution (very high indeed
> - I need this for thalamus nuclei, 0.25 x 0.25 x 0.5mm) unexpected
> crashes occur, without any error messages. This is even more pronounced
> when trying to modify the labelmap with Editor rendering it impossible
> to modify the results of the ModelIntoLAbelvolume. This must be a memory
> issue. (Linux binary, latest ones)
>
> It would be even better if the algorithm would not exclusively map the
> vertices into the volume, as this gives a "hollow" result which is very
> hard to convert to "filled" labelmaps. This is especially true in the Z
> axis, I have a couple of slices which are more or less empty.
>
> Is there a possibility (maybe outside Slicer) to "fill" a .vtk mesh and
> map what is inside onto a high res image without running into severee
> memory problems?
>
> Thank you very much
> András Jakab
>
>
>
> _______________________________________________
> slicer-users mailing list
> [hidden email]
> http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
> To unsubscribe: send email to [hidden email] with unsubscribe as the subject
_______________________________________________
slicer-users mailing list
[hidden email]
http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
To unsubscribe: send email to [hidden email] with unsubscribe as the subject
Reply | Threaded
Open this post in threaded view
|

Re: map model to volume

Dr. András Jakab
Dear Steve / Slicer Users!

I have tried that module, but without much success. Either it gives a "completed with errors" message instantly after pressing the run button (log: failed to allocate memory), or it runs for a couple of minutes giving a bad result, i.e., a volume filled entirely with 255, without any structures.

I am attaching one model. Space is standard MNI152 space, I wish to map it to a high resolution T1 template (converted to 0.25 x 0.25 x 0.5 mm voxel size for this purpose). If the size of data remains a problem (10.000s of vertices and 500Mb volume files) then please suggest me a CLI-based method or any custom codes to resolve this.

Thank you very much
András Jakab
----------------------
University of Debrecen / ETH Zürich

On Tue, Feb 22, 2011 at 2:42 PM, Steve Pieper <[hidden email]> wrote:
Hi Andras -

Did you try the PolyData to Label Map module?  It's similar, but it does try to fill in the interiors for you.  The two modules were written for different purposes and probably should be merged into a single interface at some point.

HTH,
Steve


On 02/22/2011 04:12 AM, András Jakab MD wrote:
Dear Slicer users,

Recent Slicer binaries have the feature of "ModelIntoLabelvolume" which
is exactly what I needed, as the subject of this message implies.

Albeit the application of this module seems to be problematic in my
case. I would like to inquire if any of you know a better solution for
the following problems.

When mapping a model into a volume of high resolution (very high indeed
- I need this for thalamus nuclei, 0.25 x 0.25 x 0.5mm) unexpected
crashes occur, without any error messages. This is even more pronounced
when trying to modify the labelmap with Editor rendering it impossible
to modify the results of the ModelIntoLAbelvolume. This must be a memory
issue. (Linux binary, latest ones)

It would be even better if the algorithm would not exclusively map the
vertices into the volume, as this gives a "hollow" result which is very
hard to convert to "filled" labelmaps. This is especially true in the Z
axis, I have a couple of slices which are more or less empty.

Is there a possibility (maybe outside Slicer) to "fill" a .vtk mesh and
map what is inside onto a high res image without running into severee
memory problems?

Thank you very much
András Jakab



_______________________________________________
slicer-users mailing list
[hidden email]
http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
To unsubscribe: send email to [hidden email] with unsubscribe as the subject


_______________________________________________
slicer-users mailing list
[hidden email]
http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
To unsubscribe: send email to [hidden email] with unsubscribe as the subject

VLa.vtk (88K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: map model to volume

pieper
Administrator
Hi Andras -

Thanks for sending the example data - I can see why it's not coverting
to a label map - the model looks kind of ill-formed: you can see through
to the inside surface and there seem to be holes in the model.

A couple ideas:

- are you able to convert other models (say, ones generated in slicer)
into label maps?  This would confirm that the specific model needs some
work.

- How was the model generated?  It may be mixing the clockwise verses
counter clockwise conventions for triangles within the same file.  I
tried the Surface Toolbox module but wasn't able to get anything
coherent from this model.

HTH!
-Steve

On 02/23/2011 05:04 AM, András Jakab MD wrote:

> Dear Steve / Slicer Users!
>
> I have tried that module, but without much success. Either it gives a
> "completed with errors" message instantly after pressing the run button
> (log: failed to allocate memory), or it runs for a couple of minutes
> giving a bad result, i.e., a volume filled entirely with 255, without
> any structures.
>
> I am attaching one model. Space is standard MNI152 space, I wish to map
> it to a high resolution T1 template (converted to 0.25 x 0.25 x 0.5 mm
> voxel size for this purpose). If the size of data remains a problem
> (10.000s of vertices and 500Mb volume files) then please suggest me a
> CLI-based method or any custom codes to resolve this.
>
> Thank you very much
> András Jakab
> ----------------------
> University of Debrecen / ETH Zürich
>
> On Tue, Feb 22, 2011 at 2:42 PM, Steve Pieper <[hidden email]
> <mailto:[hidden email]>> wrote:
>
>     Hi Andras -
>
>     Did you try the PolyData to Label Map module?  It's similar, but it
>     does try to fill in the interiors for you.  The two modules were
>     written for different purposes and probably should be merged into a
>     single interface at some point.
>
>     HTH,
>     Steve
>
>
>     On 02/22/2011 04:12 AM, András Jakab MD wrote:
>
>         Dear Slicer users,
>
>         Recent Slicer binaries have the feature of
>         "ModelIntoLabelvolume" which
>         is exactly what I needed, as the subject of this message implies.
>
>         Albeit the application of this module seems to be problematic in my
>         case. I would like to inquire if any of you know a better
>         solution for
>         the following problems.
>
>         When mapping a model into a volume of high resolution (very high
>         indeed
>         - I need this for thalamus nuclei, 0.25 x 0.25 x 0.5mm) unexpected
>         crashes occur, without any error messages. This is even more
>         pronounced
>         when trying to modify the labelmap with Editor rendering it
>         impossible
>         to modify the results of the ModelIntoLAbelvolume. This must be
>         a memory
>         issue. (Linux binary, latest ones)
>
>         It would be even better if the algorithm would not exclusively
>         map the
>         vertices into the volume, as this gives a "hollow" result which
>         is very
>         hard to convert to "filled" labelmaps. This is especially true
>         in the Z
>         axis, I have a couple of slices which are more or less empty.
>
>         Is there a possibility (maybe outside Slicer) to "fill" a .vtk
>         mesh and
>         map what is inside onto a high res image without running into
>         severee
>         memory problems?
>
>         Thank you very much
>         András Jakab
>
>
>
>         _______________________________________________
>         slicer-users mailing list
>         [hidden email] <mailto:[hidden email]>
>         http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
>         To unsubscribe: send email to
>         [hidden email]
>         <mailto:[hidden email]> with
>         unsubscribe as the subject
>
>
_______________________________________________
slicer-users mailing list
[hidden email]
http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
To unsubscribe: send email to [hidden email] with unsubscribe as the subject
Reply | Threaded
Open this post in threaded view
|

Re: map model to volume

Xiaodong Tao-2
In reply to this post by Dr. András Jakab
Hello Andras,

I believe the problem is the high resolution of the volume (0.25X0.25X0.5 mm).

The module that converts a polydata into a label map works by first resampling
the input polydata to have fine triangles -- the edge length of the mesh no
bigger than the smallest voxel dimensions (0.25mm in this case), so that if we
label voxels in the volume that are close to the vertices of the resampled poltdata,
we get a label map of the object boundary (a shell) that does not have holes.
The next step of the process is a flood fill to label the interior of the object.

One suggestion is to downsample the input volume ( to 1mm^3 for example ),
create a label map in the grid space of the downsampled volume and then
upsample to the higher resolution grid sample -- this will certainly introduce
some error, but will solve the "memory allocation" error.

Hope this helps.
Xiaodong

-----Original Message-----
From: [hidden email] on behalf of András Jakab MD
Sent: Wed 2/23/2011 5:04 AM
To: Steve Pieper
Cc: [hidden email]
Subject: Re: [slicer-users] map model to volume
 
Dear Steve / Slicer Users!

I have tried that module, but without much success. Either it gives a
"completed with errors" message instantly after pressing the run button
(log: failed to allocate memory), or it runs for a couple of minutes giving
a bad result, i.e., a volume filled entirely with 255, without any
structures.

I am attaching one model. Space is standard MNI152 space, I wish to map it
to a high resolution T1 template (converted to 0.25 x 0.25 x 0.5 mm voxel
size for this purpose). If the size of data remains a problem (10.000s of
vertices and 500Mb volume files) then please suggest me a CLI-based method
or any custom codes to resolve this.

Thank you very much
András Jakab
----------------------
University of Debrecen / ETH Zürich

On Tue, Feb 22, 2011 at 2:42 PM, Steve Pieper <[hidden email]>wrote:

> Hi Andras -
>
> Did you try the PolyData to Label Map module?  It's similar, but it does
> try to fill in the interiors for you.  The two modules were written for
> different purposes and probably should be merged into a single interface at
> some point.
>
> HTH,
> Steve
>
>
> On 02/22/2011 04:12 AM, András Jakab MD wrote:
>
>> Dear Slicer users,
>>
>> Recent Slicer binaries have the feature of "ModelIntoLabelvolume" which
>> is exactly what I needed, as the subject of this message implies.
>>
>> Albeit the application of this module seems to be problematic in my
>> case. I would like to inquire if any of you know a better solution for
>> the following problems.
>>
>> When mapping a model into a volume of high resolution (very high indeed
>> - I need this for thalamus nuclei, 0.25 x 0.25 x 0.5mm) unexpected
>> crashes occur, without any error messages. This is even more pronounced
>> when trying to modify the labelmap with Editor rendering it impossible
>> to modify the results of the ModelIntoLAbelvolume. This must be a memory
>> issue. (Linux binary, latest ones)
>>
>> It would be even better if the algorithm would not exclusively map the
>> vertices into the volume, as this gives a "hollow" result which is very
>> hard to convert to "filled" labelmaps. This is especially true in the Z
>> axis, I have a couple of slices which are more or less empty.
>>
>> Is there a possibility (maybe outside Slicer) to "fill" a .vtk mesh and
>> map what is inside onto a high res image without running into severee
>> memory problems?
>>
>> Thank you very much
>> András Jakab
>>
>>
>>
>> _______________________________________________
>> slicer-users mailing list
>> [hidden email]
>> http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
>> To unsubscribe: send email to
>> [hidden email] with unsubscribe as the
>> subject
>>
>

_______________________________________________
slicer-users mailing list
[hidden email]
http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
To unsubscribe: send email to [hidden email] with unsubscribe as the subject
Reply | Threaded
Open this post in threaded view
|

Re: map model to volume

Alessandra P.

Hello,

 

I’m going to use Slicer 3D in order to do affine and non affine registration.

 

I tried to use the module “Fast Affine Registration” and in the module “transform” I can see the matrix of coefficients. I’m waiting to find the following coefficients: rotational, scaling, shear and translational. What can I understand with number rapresent the coefficients that I’m looking for?

 

Then I tried to use the module for “Fast Nonrigid Bspline Registration” but at the end I don’t have a matrix of number but just the identity matrix. The second questioni s: what can I do to have non affine registration?

 

Thank you very much for your attention.

 

Best regards

 

Alessandra



> Date: Wed, 23 Feb 2011 10:14:42 -0500
> From: [hidden email]
> To: [hidden email]; [hidden email]
> CC: [hidden email]
> Subject: Re: [slicer-users] map model to volume
>
> Hello Andras,
>
> I believe the problem is the high resolution of the volume (0.25X0.25X0.5 mm).
>
> The module that converts a polydata into a label map works by first resampling
> the input polydata to have fine triangles -- the edge length of the mesh no
> bigger than the smallest voxel dimensions (0.25mm in this case), so that if we
> label voxels in the volume that are close to the vertices of the resampled poltdata,
> we get a label map of the object boundary (a shell) that does not have holes.
> The next step of the process is a flood fill to label the interior of the object.
>
> One suggestion is to downsample the input volume ( to 1mm^3 for example ),
> create a label map in the grid space of the downsampled volume and then
> upsample to the higher resolution grid sample -- this will certainly introduce
> some error, but will solve the "memory allocation" error.
>
> Hope this helps.
> Xiaodong
>
> -----Original Message-----
> From: [hidden email] on behalf of András Jakab MD
> Sent: Wed 2/23/2011 5:04 AM
> To: Steve Pieper
> Cc: [hidden email]
> Subject: Re: [slicer-users] map model to volume
>
> Dear Steve / Slicer Users!
>
> I have tried that module, but without much success. Either it gives a
> "completed with errors" message instantly after pressing the run button
> (log: failed to allocate memory), or it runs for a couple of minutes giving
> a bad result, i.e., a volume filled entirely with 255, without any
> structures.
>
> I am attaching one model. Space is standard MNI152 space, I wish to map it
> to a high resolution T1 template (converted to 0.25 x 0.25 x 0.5 mm voxel
> size for this purpose). If the size of data remains a problem (10.000s of
> vertices and 500Mb volume files) then please suggest me a CLI-based method
> or any custom codes to resolve this.
>
> Thank you very much
> András Jakab
> ----------------------
> University of Debrecen / ETH Zürich
>
> On Tue, Feb 22, 2011 at 2:42 PM, Steve Pieper <[hidden email]>wrote:
>
> > Hi Andras -
> >
> > Did you try the PolyData to Label Map module? It's similar, but it does
> > try to fill in the interiors for you. The two modules were written for
> > different purposes and probably should be merged into a single interface at
> > some point.
> >
> > HTH,
> > Steve
> >
> >
> > On 02/22/2011 04:12 AM, András Jakab MD wrote:
> >
> >> Dear Slicer users,
> >>
> >> Recent Slicer binaries have the feature of "ModelIntoLabelvolume" which
> >> is exactly what I needed, as the subject of this message implies.
> >>
> >> Albeit the application of this module seems to be problematic in my
> >> case. I would like to inquire if any of you know a better solution for
> >> the following problems.
> >>
> >> When mapping a model into a volume of high resolution (very high indeed
> >> - I need this for thalamus nuclei, 0.25 x 0.25 x 0.5mm) unexpected
> >> crashes occur, without any error messages. This is even more pronounced
> >> when trying to modify the labelmap with Editor rendering it impossible
> >> to modify the results of the ModelIntoLAbelvolume. This must be a memory
> >> issue. (Linux binary, latest ones)
> >>
> >> It would be even better if the algorithm would not exclusively map the
> >> vertices into the volume, as this gives a "hollow" result which is very
> >> hard to convert to "filled" labelmaps. This is especially true in the Z
> >> axis, I have a couple of slices which are more or less empty.
> >>
> >> Is there a possibility (maybe outside Slicer) to "fill" a .vtk mesh and
> >> map what is inside onto a high res image without running into severee
> >> memory problems?
> >>
> >> Thank you very much
> >> András Jakab
> >>
> >>
> >>
> >> _______________________________________________
> >> slicer-users mailing list
> >> [hidden email]
> >> http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
> >> To unsubscribe: send email to
> >> [hidden email] with unsubscribe as the
> >> subject
> >>
> >
>
> _______________________________________________
> slicer-users mailing list
> [hidden email]
> http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
> To unsubscribe: send email to [hidden email] with unsubscribe as the subject

_______________________________________________
slicer-users mailing list
[hidden email]
http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
To unsubscribe: send email to [hidden email] with unsubscribe as the subject