Posted by
Katinka Wouters on
Jul 26, 2010; 6:39am
URL: http://imagej.273.s1.nabble.com/multichannel-3D-viewing-problem-with-Fluoview500-tiffs-tp3687386p3687390.html
Hi Glen,
Dear Katinka,
Scrolling through the stack is simply stepping between the images. The voxel depth value is assigned at acquisition and written into the file metadata. The TIFF does not have a native tag for storing the axial dimension. Your macro may be converting the Fluoview file to tiff in a manner which causes that value to be lost.
-- Indeed, that's what I thought, thanks for clarifying
ImageJ writes voxel depth to the TIFF comments tag and reads it back out, if you re-open in ImageJ. Have you tried the current stable version of LOCI, 4.2.0 to open your files?
-- yes I did
If that does not correctly set channel order, then you can select the correct channel order by the Stack Order dropdown menu in the upper left region of the plugin window.
-- tried this, but there's no change in my outcome whether I choose xyzct, xyczt, etc... when I want to import with separating channels
Importing as Hyperstack with Composite color will maintain voxel properties and does a better job of channel color than does the Standard ImageJ Viewer option. Use Hyperstacks>Channels Tool>More to change channel colors. Use Hyperstacks>Channels Tool>More>Convert to RGB will create an 8-bit/channel RGB with voxel depth (record this once and save as a macro) for the 3D viewer.
-- I'll try this one, thanks!
Katinka
Regards,Glen
On Jul 23, 2010, at 4:21 AM, Katinka Wouters wrote:
> Hi Bene,
>
> Thanks for your very fast reply. I checked the image properties, and indeed, the voxel depth is at 0.0000000. So, problem solved! Any idea why the macro changes the voxel depths for multichannel images this dramatically?
>
> I will now try to use a SetVoxelDepth macro, in order to keep the process automated.
>
> Thanks a lot, you made my day!
>
> Katinka
>
> -----Original Message-----
> From: ImageJ Interest Group [mailto:
[hidden email]] On Behalf Of Benjamin Schmid
> Sent: Friday, July 23, 2010 1:09 PM
> To:
[hidden email]
> Subject: Re: multichannel 3D viewing problem with Fluoview500 tiffs
>
> Hi Katinka,
>>
>> For a while, I've been using ImageJ 3D viewer or volume viewer to visualize bacterial biofilms scanned with an Olympus Fluoview 500. It works perfectly for single channel images, thanks a lot to the developers!
>>
>> But, when I'm using it for multichannel images (so in my case: multispecies biofilms), my '3D' images appear 'flat' (like 2D), in both the 3D viewer and the volume viewer. When I rotate them, I get just one plane. This is particularly strange, since I can still browse through the Z-slices in the stack from which the 3D-image originate: all the (over 40) z-slices are there. So, I have a perfect 8 bit z-stack of each channel, but get a 2D image from them, both from the separate channels as from a stack in which the channels have been merged first.
>>
>> Because this problem occurs in two separate plugins (both 3D Viewer as Volume Viewer), I wonder whether it's a problem with my tiff-series. An image stack originates from a Fluoview500 tiff file, which was converted for use with ImageJ using the Save_8bit_MultiCH plugin in combination with the LOCI plugin. This generates a separate tiff stack for each channel. So everything seems perfect, untill I try to render them to 3D. I used the same plugin to convert the single channel images to a tiff stack, without any 3D viewing problems.
>>
>> I noticed before that I can not use the LOCI plugin as such to import my FV tiffs, because channels and z-position get mixed up: I always get a stack with repeatedly CH1; CH2; CH3; CH1; CH2; CH3... as if these are the z-slices, and this divided over 3 'channels', which are in fact not the channels. This happens with every setting I tried (e.g. changing the xyzct order). Luckily, this problem was solved when using the Save_8bit_MultiCH plugin, but now I wonder there is maybe some error left in the metadata that gets the 3D/Volume Viewer confused about the stack.
>>
>> Does anyone recognize this problem? Or is there something obvious that I'm missing here? I would be very gratefull if I could find a solution for it, since I'm promoting ImageJ towards my colleagues who are now using the very expensive program from Zurich...
>>
>
> My first thought is that the calibration in the z direction may be wrong/to
> small. Can you check this? (->Image->Properties: Voxel depth).
>
> If the voxel depth seems to make sense, could you post the dimensions of
> the image, i.e. what you see in the dialog when clicking
> Image->Hyperstacks->Stack to Hyperstack.
>
>
> Best wishes,
> Bene
Glen MacDonald
Core for Communication Research
Virginia Merrill Bloedel Hearing Research Center
Box 357923
University of Washington
Seattle, WA 98195-7923 USA
(206) 616-4156
[hidden email]