newbie needs to know about 3D

HOW TO MAKE A 3-D RENDERING OF OUTLINES OF NUCLEARBOUNDARIES IN AN EVENLY SPACED SET OF STAINED SECTION THROUGH A MAMMALIANBRAIN.
 
What we thought would be a reasonable approach:


1.     1.  Scan or photograph each section in the series,


2.     2.  Align the sections, ideally by reference toblockface photos taken during the sectioning process. Alterniatively, open thesections one at a time in photoshop, copying each section and placing it a alayer of a .psd file. By manipulating the images of two layers so as tonear-exactly place each layer in line with its predecessor.


3.     3,  Then crop the .psd file so each image frame isthe same, and the positions of the sections in the frames define the boundariesof the eventual 3D rendering.


4.    4.   Save each layer as a .tif file, and place all ofthem in a folder.


5.    5.    Drop the folder into the ImageJ window so as tomake a stack.


6.    6.   In each image in the stack, using the freehandtrace the outlines of the nucleus of interest.


HERE THE WHOLE APPROACH COLLAPSES IN CHAOS, SINCE WHEN WE TRACE ANOUTLINE ON AN IMAGE, THEN SCROLL TO THE NEXT IMAGE IN THE STACK, THE OUTLINE LEAVES THE IMAGE AND COMES WITH THESCROLL. ATTEMPTING TO DRAW ANYTHING CAUSES THE OUTLINE TO VANISH AND ONE CANNOT DRAW AN OUTLINE AROUND THE NUCLEUS IN THE NEXT IMAGE.
 
Where did we go wrong, and where can we find any instructions at allabout how to get the images and or outlines into Image J so as to continue withthe 3D rendering.?
 
Any information will be greatly appreciated,
 
John I. Johnson
[hidden email].

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Stephan Saalfeld]

Hi John,

please read about and check the plugin TrakEM2

http://fiji.sc/TrakEM2

which is included in Fiji.  It's designed for exactly what you are
describing, reconstructing 3D structures from section series, some
learning curve involved.  It also enables automatic section series
alignment about which you can read here:

http://fiji.sc/Elastic_Alignment_and_Montage

AND PLEASE STOP SCREAMING!!!! ;)

Cheers,
Stephan



On Tue, 2013-07-23 at 23:56 -0400, [hidden email] wrote:

> HOW TO MAKE A 3-D RENDERING OF OUTLINES OF NUCLEARBOUNDARIES IN AN EVENLY SPACED SET OF STAINED SECTION THROUGH A MAMMALIANBRAIN.
>  
> What we thought would be a reasonable approach:
>
>
> 1.     1.  Scan or photograph each section in the series,
>
>
> 2.     2.  Align the sections, ideally by reference toblockface photos taken during the sectioning process. Alterniatively, open thesections one at a time in photoshop, copying each section and placing it a alayer of a .psd file. By manipulating the images of two layers so as tonear-exactly place each layer in line with its predecessor.
>
>
> 3.     3,  Then crop the .psd file so each image frame isthe same, and the positions of the sections in the frames define the boundariesof the eventual 3D rendering.
>
>
> 4.    4.   Save each layer as a .tif file, and place all ofthem in a folder.
>
>
> 5.    5.    Drop the folder into the ImageJ window so as tomake a stack.
>
>
> 6.    6.   In each image in the stack, using the freehandtrace the outlines of the nucleus of interest.
>
>
> HERE THE WHOLE APPROACH COLLAPSES IN CHAOS, SINCE WHEN WE TRACE ANOUTLINE ON AN IMAGE, THEN SCROLL TO THE NEXT IMAGE IN THE STACK, THE OUTLINE LEAVES THE IMAGE AND COMES WITH THESCROLL. ATTEMPTING TO DRAW ANYTHING CAUSES THE OUTLINE TO VANISH AND ONE CANNOT DRAW AN OUTLINE AROUND THE NUCLEUS IN THE NEXT IMAGE.
>  
> Where did we go wrong, and where can we find any instructions at allabout how to get the images and or outlines into Image J so as to continue withthe 3D rendering.?
>  
> Any information will be greatly appreciated,
>  
> John I. Johnson
> [hidden email].
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[José María Mateos]

2013/7/24 [hidden email] <[hidden email]>:
> HERE THE WHOLE APPROACH COLLAPSES IN CHAOS, SINCE WHEN WE TRACE ANOUTLINE ON AN IMAGE, THEN SCROLL TO THE NEXT IMAGE IN THE STACK, THE OUTLINE LEAVES THE IMAGE AND COMES WITH THESCROLL. ATTEMPTING TO DRAW ANYTHING CAUSES THE OUTLINE TO VANISH AND ONE CANNOT DRAW AN OUTLINE AROUND THE NUCLEUS IN THE NEXT IMAGE.

I understand you would have to use the ROI manager (Analyze > Tools >
ROI Manager...) to draw different selections on different slices of
your stack.

Best,

José María Mateos.

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Robert Kirmse-2]

Hi John,
I don't know if I understood the imaging process of your sample correctly
but here are my thought for your process.
 
If you haven't already, download the Fiji package
it is essentially ImageJ with a bunch of pre-installed Plugins,
including a lot that deal with stack registration i.e. alignment
and segmentation. Of course you can also download the plugins
for ImageJ seperately
 
Without having seen your images
I would already start in ImageJ/Fiji,
and import the whole series into a stack
 
1. Fiji > File > Import > Image Sequence (your files need to be numbered)
2. Fiji > Plugins > Registration > Stackreg (this should already help you with a decent alignment, but also try other methods from the registration plugins to see what is best for you data.
This should give you a nicely aligned stack that you can movie through.
Crop the stack to your needs in ImageJ/Fiji
 
3. Segmentation,
You need a good Segmentation editor that can handle different objects that are traced
either manually or automatically. In these editors you won't have the problem you described
because the program takes care of the outlines you draw and to which object they belong.
Afterwards you can render the outlines that belong to each object.
My experience with Segmentation editors in ImageJ is limited so I can't give a good recommendation
I normally used the imod package (developed for electron microscopy) for manual segmentation.
However even with good tools manual segmentation is quite time intensive.
 
Also if you "just" want to generate a 3D rendering try Fiji > Plugins > Volume Viewer
Here you have to experiment with the display option and the transfer curves to bring out
the feature you want and surpress the unwanted features. I guess for what you want
the option "Volume" under Mode is probably what you want.
 
You could also have a look at the Ilastik package in addition to the Segmentation plugins in
Fiji which is a good trainable segmentation tool that also offers 3D rendering options.
 
Anyway, Segmentation will require some learning time, so be prepared to spent some time
researching and trying different tools.
 
Cheers
Robert
 
 
 
 
 
 
 
 
 

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html