Re: 3D Eroded Object Masks *commercial response*

Posted by Bitplane Support on
URL: http://confocal-microscopy-list.275.s1.nabble.com/Re-3D-Eroded-Object-Masks-commercial-response-tp3272647.html

Dear Cam,

I'll have to give your question a bit more thought and maybe talk it
over with some others, so I might have more info later, but for now Eric
is right, for the current version of Imaris, at least some of this would
require an ImarisXT "XTension".

For "subtractive image mathematics" there is one that already exists
called "channel arithmetics".  It's actually one of the XTensions that
is included in the Imaris installer.  It requires that you have MatLab
also installed (this is not a general requirement of XTensions, but it
is required for most of the ones that I know of).

I'm not sure you really need channel arithmetics, because I got the
impression you only need to apply a binary mask.  This is done in Imaris
by creating a "Surfaces" object, and then on the "Edit" tab, you'll find
buttons called "Mask All" and "Mask Sel" (use all of the objects or just
selected one(s) )

That still does not solve your 3D erosion problem (which could also be
solved by an XTension, but I don't know of an existing one).  As I said,
I might have more ideas for you later, but for now, since you said all
of the MetaMorph masks worked fairly well except the membrane, you might
try running a Gaussian smoothing filter in Imaris on your "stack of
rubber bands".  Because the Gaussian smoothing in Imaris operates in 3D,
this should lessen the stepping effect along the z direction.  Note that
the smoothing width value is in um, not pixels/voxels, so you may have
to be fairly aggressive with the value if your z-steps are relatively large.

Tip:  you can force the smoothing units to operate in the pixel/voxel
domain by temporarily setting your x-y-z voxel sizes to 1-1-1.  Your
image might look quite distorted, but after you have run the filter, set
the values back to their original size, and viola! your image shape is
normal again, and you have just forced an anisotropic Gaussian
smoothing - because the "real" Z dimension is actually not the same as
XY, you are applying more smoothing along Z (in spatial/object terms,
not in terms of voxels).

You can use ratios other than 1:1:1 if you want to "tune" the anisotropy
in a particular way (decreasing the Z voxel size will cause the Z
dimension to be more smoothed than X and Y - in terms of voxels).  I
suggest experimenting on a cropped version of your dataset so you can
quickly see the results of different variations.
* drastic changes to voxel dimensions will require you to re-center the
dataset with the "100%" button, followed by "fit"

Best regards,
-Kevin

Kevin Frischmann
Head of Technical Support, US & Canada
Bitplane, Inc.
tel: [phone], ext. 11
fax: [phone]

Join the Bitplane Facebook Group!
http://www.facebook.com/group.php?gid=22023791027
Be a part of our effort build an active community for advanced 3D imaging,
for Imaris users, or anyone else with related interests.
Gather and share information, images and movies in a fun, casual forum.



Eric Scarfone wrote:

> "is it possible in Imaris to do
> subtractive image mathematics?"
>
> not with the standard modules I din't think.
>
> But there is a ImarisXT modulethat lest you access, from Imaris,
> applications developped with other software. That gives also access to a
> whole library of customer developped appications where you could find
> what you are looking for!
>
> Good luck
>
> Eric
>
>
> Eric Scarfone, PhD, CNRS,
> Center for Hearing and communication Research
> Department of Clinical Neuroscience
> Karolinska Institutet
>
> Postal Address:
> CFH, M1:02
> Karolinska Hospital,
> SE-171 76 Stockholm, Sweden
>
> Work: [phone],
> Cell: [phone]
> Fax: [phone]
>
> email: [hidden email]
> http://www.ki.se/cfh/
>
>
> ----- Original Message -----
> From: Cameron Nowell <[hidden email]>
> Date: Wednesday, July 15, 2009 8:15 am
> Subject: 3D Eroded Object Masks
> To: [hidden email]
>
>  > I am trying to segment out membrane, cytoplasm and nuclei from a
>  > confocal image set. This is so we can measure translocation of a
>  > proteinfrom the membrane to the cytoplasm/nucleus over time. I can
>  > do this in
>  > MetaMorph but would like to be able to do it in 3D in Imaris if
>  > possible.
>  >
>  > The sample is stained as follows
>  > - Nuclei - DAPI
>  > - Whole Cell - CellMask (Invitrogen)
>  > - Protein of Interest (Alexa Antibody)
>  >
>  > Since there is not a specific stain from membrane or cytoplasm i have
>  > been doing the following
>  > - Segment out whole cells and create a binary mask
>  > - Erode that mask by 4 or so pixels, this represents the cytoplasm
>  > and nuclei of the cell
>  > - Subtract the eroded mask from the whole cell mask. This leave a
>  > ring that represents the membrane of the cell.
>  > - Segment and subtract the nuclei from the combined cytoplasm and
>  > nuclei mask to give a cytoplasm mask.
>  > - The end result is three masks; one each representing membrane,
>  > nuclei and cytoplasm.
>  >
>  > I can create these masks for each slice of a confocal set and get
>  > intensity etc information out from MetaMorph. I have tried
>  > exporting the
>  > mask stacks out to Imaris to create 3D masks but it doesn't work very
>  > well, especially for the membrane mask, as the slices don't
>  > necessarilyoverlap so there are gaps in the mask. My 3D model of
>  > the membrane masks
>  > looks more like a badly piled up lot of rubber bands. The 3D masks for
>  > the whole cell or nucleus work fine.
>  >
>  > So i guess the main question is: is it possible in Imaris to do
>  > subtractive image mathematics?
>  >
>  >
>  >
>  > Thanks
>  >
>  >
>  > Cam
>  >
>  >
>  >
>  > Cameron J. Nowell
>  > Microscopy Manager
>  > Centre for Advanced Microscopy
>  > Ludwig Institute for Cancer Research
>  > PO Box 2008
>  > Royal Melbourne Hospital
>  > Victoria, 3050
>  > AUSTRALIA
>  > Office: [phone]
>  > Mobile: [phone]
>  > Fax: [phone]
>  > Facility Website
>  >
>