Opening Mosaix zvi files with Bio-Formats

[Daniel James White]

Hi Jan,


On Jun 15, 2011, at 6:00 AM, IMAGEJ automatic digest system wrote:

>
> Dear Bio-Formats specialists,
>
> I used to open zvi files that were acquired by Zeiss AxioVision software
> in MosaiX mode (without overlap, and without using the software's
> stitching). With previous versions of bio-formats, each file was opened
> as a single tiled image containing all tiles in the correct order.
> After recent updates, the Bio-Formats importer recognizes the tiles as
> separate series, and asks for the series to be opened.
>
> How can I open the file (all tiles) as a single image now?
>
> (In the dialog, I checked "Open all series" and "Concatenate series when
> compatible", but that results in a hyperstack with the series as time
> points. "Stacks>Make Montage..." doesn't help in this case, because it
> does not work with hyperstacks.)
>
> Any help is appreciated.
> Thanks,
> Jan

>
> Date:    Tue, 14 Jun 2011 10:55:42 +0200
> From:    Jan Eglinger <[hidden email]>
> Subject: Re: Opening Mosaix zvi files with Bio-Formats
>
> On 14.06.2011 10:48 AM, Jan Eglinger wrote:
>>
>> How can I open the file (all tiles) as a single image now?
>
> I forgot to add that I have z-stacks for each tile image.
> (2 channels, 12 z slices, 3x3 tile mosaic)

We asked for the change in behaviour to make it consistent with how
tile scans from the Zeiss LSM 7xx come our as single .lsm files.

This is so one can then use the stitching plugin:
Stitch multiple series or tile scan file
to stitch the images properly according to the stage positions,
fine tuning the overlap and blending of the edges,
so the result image is segmentable etc.

The Zeiss stitching does not adjust the position of the tiles
and you can easily se the edges are not well matched,
esp when the camera is not perfectly square with the stage.

So this is a feature indeed.

Let us know if that plugin version can stitch your
axiovision .zvi mosaic scans.

We need to make this work, there are many users here also.

Dan




>
> Jan

Dr. Daniel James White BSc. (Hons.) PhD
Senior Microscopist / Image Visualisation, Processing and Analysis
Light Microscopy and Image Processing Facilities
Max Planck Institute of Molecular Cell Biology and Genetics
Pfotenhauerstrasse 108
01307 DRESDEN
Germany

[phone] (German Mobile)
[phone] (Work phone at MPI-CBG)
[phone] (Fax MPI-CBG LMF)

http://www.bioimagexd.net  BioImageXD
http://pacific.mpi-cbg.de                Fiji -  is just ImageJ (Batteries Included)
http://www.chalkie.org.uk                Dan's Homepages
https://ifn.mpi-cbg.de  Dresden Imaging Facility Network
dan (at) chalkie.org.uk
( white (at) mpi-cbg.de )

[Jan Eglinger-3]

Hi Dan,

thanks for your reply.

On 15.06.2011 8:59 AM, Daniel James White wrote:
> We asked for the change in behaviour to make it consistent with how
> tile scans from the Zeiss LSM 7xx come our as single .lsm files.

I see. In my case, all mosaic positions are saved in a single zvi file.
What's the best way to open these now? Whould I open all 64 series (in
case of an 8x8 mosaic) in separate windows and then stitch them using
the plugin? The command "Stitch Grid of Images" takes a directory as
input, but I have a single file only.

> This is so one can then use the stitching plugin:
> Stitch multiple series or tile scan file
> to stitch the images properly according to the stage positions,
> fine tuning the overlap and blending of the edges,
> so the result image is segmentable etc.
>
> The Zeiss stitching does not adjust the position of the tiles
> and you can easily se the edges are not well matched,
> esp when the camera is not perfectly square with the stage.
>
> So this is a feature indeed.
>
> Let us know if that plugin version can stitch your
> axiovision .zvi mosaic scans.
>

In my case, the image data contains 90% black area and only very few
elongated objects. That's why I have to rely on a good alignment of
camera and stage. In my experience, stitching algorithms don't help much
in that case, and I simply want to put the tiles together with 0% overlap.

How can I achieve this with the current plugin?

Thanks for your help,
Jan

[Jan Eglinger-3]

Hi again,

On 15.06.2011 2:47 PM, Jan Eglinger wrote:
> In my case, the image data contains 90% black area and only very few
> elongated objects. That's why I have to rely on a good alignment of
> camera and stage. In my experience, stitching algorithms don't help much
> in that case, and I simply want to put the tiles together with 0% overlap.
>
> How can I achieve this with the current plugin?

Please excuse my slightly premature question. I now found that I can
open the mosaic file using "Stitch Multiple Series or Tile Scan File",
then unchecking "Compute overlap", setting the overlap to 0% and the
Fusion Method to "None".

However, this opens my 2-channel stack in RGB mode, thereby consuming an
unnecessary amount of memory.

Any hint how I can directly open as 2-channel hyperstack?


In my opinion, it would be nice to have an intuitive way to open this
kind of zvi file just as the acquisition software, i.e. AxioVision, does.

Cheers,
Jan

[Daniel James White]

Hi Jan,

On Jun 15, 2011, at 3:16 PM, Jan Eglinger wrote:

> Hi again,
>
> On 15.06.2011 2:47 PM, Jan Eglinger wrote:
>> In my case, the image data contains 90% black area and only very few
>> elongated objects. That's why I have to rely on a good alignment of
>> camera and stage. In my experience, stitching algorithms don't help much
>> in that case, and I simply want to put the tiles together with 0% overlap.
>>
>> How can I achieve this with the current plugin?
>
> Please excuse my slightly premature question. I now found that I can
> open the mosaic file using "Stitch Multiple Series or Tile Scan File",
> then unchecking "Compute overlap", setting the overlap to 0% and the
> Fusion Method to "None".

Good, that how it should work!!!

>
> However, this opens my 2-channel stack in RGB mode, thereby consuming an
> unnecessary amount of memory.

Yes, and if its a 4 channel image... we are in trouble!

so Steffi and I talked about the need to use composite image instead of RGB,
so you can have many more "channels"

>
> Any hint how I can directly open as 2-channel hyperstack?

split them, then merge as a composite
throw away the empty channel....
i guess a script would be needed to make that efficient.

>
> In my opinion, it would be nice to have an intuitive way to open this
> kind of zvi file just as the acquisition software, i.e. AxioVision, does.

how do you mean exactly?

This is a job for the bio-formats importer.....
which maybe you can suggest how you might like it to be able to work as an option?

Dan



>
> Cheers,
> Jan

Dr. Daniel James White BSc. (Hons.) PhD
Senior Microscopist / Image Visualisation, Processing and Analysis
Light Microscopy and Image Processing Facilities
Max Planck Institute of Molecular Cell Biology and Genetics
Pfotenhauerstrasse 108
01307 DRESDEN
Germany

[phone] (German Mobile)
[phone] (Work phone at MPI-CBG)
[phone] (Fax MPI-CBG LMF)

http://www.bioimagexd.net  BioImageXD
http://pacific.mpi-cbg.de                Fiji -  is just ImageJ (Batteries Included)
http://www.chalkie.org.uk                Dan's Homepages
https://ifn.mpi-cbg.de  Dresden Imaging Facility Network
dan (at) chalkie.org.uk
( white (at) mpi-cbg.de )

[Daniel James White]

Hi Jan,

On Jun 15, 2011, at 2:47 PM, Jan Eglinger wrote:

> Hi Dan,
>
> thanks for your reply.
>
> On 15.06.2011 8:59 AM, Daniel James White wrote:
>> We asked for the change in behaviour to make it consistent with how
>> tile scans from the Zeiss LSM 7xx come our as single .lsm files.
>
> I see. In my case, all mosaic positions are saved in a single zvi file.
> What's the best way to open these now? Whould I open all 64 series (in
> case of an 8x8 mosaic) in separate windows and then stitch them using
> the plugin? The command "Stitch Grid of Images" takes a directory as
> input, but I have a single file only.

nope  - use the plugin:
Stitch multiple series or tile scan file

which takes a single file input.

>
>> This is so one can then use the stitching plugin:
>> Stitch multiple series or tile scan file
>> to stitch the images properly according to the stage positions,
>> fine tuning the overlap and blending of the edges,
>> so the result image is segmentable etc.
>>
>> The Zeiss stitching does not adjust the position of the tiles
>> and you can easily se the edges are not well matched,
>> esp when the camera is not perfectly square with the stage.
>>
>> So this is a feature indeed.
>>
>> Let us know if that plugin version can stitch your
>> axiovision .zvi mosaic scans.
>>
>
> In my case, the image data contains 90% black area and only very few
> elongated objects. That's why I have to rely on a good alignment of
> camera and stage.

it wil never be perfect enough, as the lenses never have perfectly flat and distortion free field of view,
and even a tiny camera rotation will show up over a large tile scan.

> In my experience, stitching algorithms don't help much
> in that case, and I simply want to put the tiles together with 0% overlap.

you will still have problems at the edges of the tiles,
they wil not line up quite right.

Best to have overlapping info and use that to align the tile correctly.

>
> How can I achieve this with the current plugin?

Use some overlap.. enough for the plugin to see enough of the same stuff in overlapping areas.

the images with no information will be ignored if you use the right settings.

why not try using multiple points instead of tile scan if the image is so sparse,
making sure there is overlap between the fields of view...?

Dan



>
> Thanks for your help,
> Jan
>
>

Dr. Daniel James White BSc. (Hons.) PhD
Senior Microscopist / Image Visualisation, Processing and Analysis
Light Microscopy and Image Processing Facilities
Max Planck Institute of Molecular Cell Biology and Genetics
Pfotenhauerstrasse 108
01307 DRESDEN
Germany

[phone] (German Mobile)
[phone] (Work phone at MPI-CBG)
[phone] (Fax MPI-CBG LMF)

http://www.bioimagexd.net  BioImageXD
http://pacific.mpi-cbg.de                Fiji -  is just ImageJ (Batteries Included)
http://www.chalkie.org.uk                Dan's Homepages
https://ifn.mpi-cbg.de  Dresden Imaging Facility Network
dan (at) chalkie.org.uk
( white (at) mpi-cbg.de )

[Jan Eglinger-3]

Hi Dan,

thanks for your comments, I'll retry with images using some overlap.
However, the results I got with just a well aligned camera and without
stitching calculations - although not yielding perfect images - _were_
indeed sufficient for my purpose.
And now I'd like to be able to work with those images containing 0% overlap.

On 15.06.2011 4:21 PM, Daniel James White wrote:
> On Jun 15, 2011, at 3:16 PM, Jan Eglinger wrote:
>> In my opinion, it would be nice to have an intuitive way to open this
>> kind of zvi file just as the acquisition software, i.e. AxioVision, does.
>
> how do you mean exactly?

In AxioVision, a file acquired with Mosaix mode but not processed by the
built-in Stitching will be opened as a full X*Y tiled image (with the
tile borders possibly visible due to imperfect illumination and lens
aberrations) ...

>
> This is a job for the bio-formats importer.....
> which maybe you can suggest how you might like it to be able to work as an option?

... and older versions of Bio-Formats used to open those files in
exactly that same way, hence my very first mail in this thread.

Best,
Jan

[Daniel James White]

Hi Jan,

On Jun 15, 2011, at 4:55 PM, Jan Eglinger wrote:

> Hi Dan,
>
> thanks for your comments, I'll retry with images using some overlap.
> However, the results I got with just a well aligned camera and without
> stitching calculations - although not yielding perfect images - _were_
> indeed sufficient for my purpose.
> And now I'd like to be able to work with those images containing 0% overlap.

you can do it, but remember that the plugin is expecting overlap
to be able to optimize the tile positions..
but it will in any case read the stage position to get the first initial guess..
which for you might be close enough.

>
> On 15.06.2011 4:21 PM, Daniel James White wrote:
>> On Jun 15, 2011, at 3:16 PM, Jan Eglinger wrote:
>>> In my opinion, it would be nice to have an intuitive way to open this
>>> kind of zvi file just as the acquisition software, i.e. AxioVision, does.
>>
>> how do you mean exactly?
>
> In AxioVision, a file acquired with Mosaix mode but not processed by the
> built-in Stitching will be opened as a full X*Y tiled image (with the
> tile borders possibly visible due to imperfect illumination and lens
> aberrations) ...

right....
but thats just eye candy... so you can "see" an overview.

... but there is no good reason for actually destructively saving the images like that.
The real image info is the individual tiles,  and thats what Zeiss gives us. ...
We versatility in how to handle the tiles...
not some poorly stitched and then impossible to fix single overview image...

In the general case, the tile images need to be accessible individually - would you agree?
In your case you want to get straight to the roughly stitched overview image.... that has "zero" overlap and ugly edges.

Presently imageJ has no way to do that automagically: using the stage positions to make a rough tile view.

If you want to write an extension to bio-formats importer that does that pre processing on file import,
and spits the roughly splatted together overview image into an imageJ window,
then its for sure possible... feel free to implement it.

But the general case is that tiles should have overlap, and should be stitched properly according to the overlapping info close to the edges
as this output is then suitable for measurement and segmentation, whereas the rough splatted together image is not.
>
>>
>> This is a job for the bio-formats importer.....
>> which maybe you can suggest how you might like it to be able to work as an option?
>
> ... and older versions of Bio-Formats used to open those files in
> exactly that same way, hence my very first mail in this thread.

Yes, we changed it as it was inconsistent with the way Zeiss .lsm files are handled,
which also have tile scan s that need to be stitched... as does any tile scan from a system that can do mosaic/tile-scans.

Remember we are trying to make the whole of Fiji work using sensible defaults
that cover most cases, but still allow edge cases (like zero overlap) to be doable.

Most folks are doing a 10% or so overlap, as thats a sensible thing to do in most cases:
so you dont miss some parts at the edges, and also so that the stitching algorithms can work.

The stage position info is in the metadata an dis read by bio-formats,
so you can assemble the roughly stitched image overview any way you like from that info.

It could be another plugin in the stitching menu:
Stitch roughly using stage position data only.
Fast but ugly... but maybe sometimes good enough, as in your case.


Or you can just export the roughly stitched image from Axiovision.... maybe.

cheers

Dan


>
> Best,
> Jan

Dr. Daniel James White BSc. (Hons.) PhD
Senior Microscopist / Image Visualisation, Processing and Analysis
Light Microscopy and Image Processing Facilities
Max Planck Institute of Molecular Cell Biology and Genetics
Pfotenhauerstrasse 108
01307 DRESDEN
Germany

[phone] (German Mobile)
[phone] (Work phone at MPI-CBG)
[phone] (Fax MPI-CBG LMF)

http://www.bioimagexd.net  BioImageXD
http://pacific.mpi-cbg.de                Fiji -  is just ImageJ (Batteries Included)
http://www.chalkie.org.uk                Dan's Homepages
https://ifn.mpi-cbg.de  Dresden Imaging Facility Network
dan (at) chalkie.org.uk
( white (at) mpi-cbg.de )