Posted by
ctrueden on
Jan 09, 2008; 7:03pm
URL: http://imagej.273.s1.nabble.com/ImageJ-driving-AVT-Guppy-CCD-cam-tp3697467p3697473.html
Hi Toby,
Thanks for the warning. Fortunately, Bio-Formats already supports
nearly everything you mention in a modular fashion. It can already
decode JPEG2000 using the JAI ImageIO tools library, works with JPEG
and TIFF, including many forms of compressed TIFF (though we may need
to code an LZH filter -- I don't think we handle LZH yet).
To be perfectly clear: we are completely serious about Bio-Formats
being a complete library for the exchange of microscopy data between
both software packages and scientific organizations. We want to
support conversion of every major microscopy file format, including
both pixels and metadata, into a common data standard.
As for the size of the image planes, Bio-Formats recently added an API
for reading "sub-images" -- specific portions of an image plane within
a bounding rectangle. So at worst, for gigantic image planes, the
plugin will allow the user to specify a subregion to load in. If there
is enough demand, we can add other sorts of "tile on import" types of
functionality to the plugin as well. Depending on whether the planes
are compressed as tiles, or as one gigantic compressed block, a large
amount of system RAM may be required to read the files, but we have
largely overcome the 2 GB (32-bit) RAM barrier at this point, so 1.5
gigapixels is becoming more manageable.
Regarding improving ImageJ's support for working with huge images in
general, that is something we can hopefully collaborate on with Wayne,
since such functionality would be of wider applicability than just
Aperio format. Ideally, Bio-Formats won't need any custom display
logic, but rather ImageJ would have a "large image" mode that could be
invoked, or something like that.
The most useful thing people can do to help us in the immediate future
would be to send us some SVS samples. Ideally, we would have one or
more samples in each of the internal formats within the container
(i.e., one in JPEG2000, one JPEG, one TIFF, one TIFF-LZH, etc.). If
you can provide download links to some samples, that would be great.
Otherwise, if you need a place to upload files, I can send you our FTP
information privately -- just let me know.
This same offer goes for ANY file format people would like to see
added to Bio-Formats -- send us some sample data, and we will do our
best to add support for the format, as time allows. Depending on the
complexity of the format, I cannot promise a particular time frame,
but most formats end up being pretty simple to add, so chances are
good we can do it reasonably quickly.
-Curtis
On Jan 9, 2008 12:13 AM, Toby Cornish <
[hidden email]> wrote:
> Melissa,
>
> It would be great if we could get some an imageJ plugin for the Aperio file format. I would like to warn you somewhat about what you are getting into, though. As far as I understand it, the .svs file is a container format that can contain tiff, LZH compressed tiff, jpeg or jpeg2000 images. These are generally whole slide images and typically run around 120 MB per file for a 20x, highly compressed jpeg2000 image. That is somewhere around 1.5 gigapixels for a typical 20x slide. At 40x, the situation is even worse.
>
> So, there are really two issues with opening these files in imageJ. First,jpeg2000 (which b/c of image size, is the defacto default encoding for whole slide images) isn's supported by imageJ, although java libraries exist for handling the encoding. More concerning, is the massive size of these images. This will probably necessitate a lot of display tricks such as paging portions of the image into memory and using prescaled lower resolution images (included in the svs file) for zooming. I am not sure how much of this imageJ would support without a lot of new code or some serious hacks. Of course, I could be wrong and maybe large image support was added to imageJ at some point.
>
> So, I think adding .svs file support in considerably more than coding up import/export filters. That stated, there are MANY people that would love to see new tools for working with these files. The lock-in with Aperio is frustrating, especially when all of your analysis methods are imageJ-centric. If you are serious about getting some form of support for .svs files in imageJ as a plugin or otherwise, I would be glad to do whatever I could to aid you.
>
> toby
>
>
>
> > Date: Mon, 7 Jan 2008 14:01:44 -0500
> > From: Yan Gao <
[hidden email]>
> > Subject: Re: I would like doing image analysis of svs file by using imageJ
> >
> > Hi, Melissa.
> >
> > Thank you very much for your offer. Yes, I can send you some files. The
> > files are too big, you need let me know how I can put the files in your
> > server. I really appreciated your help.
> >
> > Best Regards,
> >
> > Yan
> >=====================================
> >
> >Yan Gao, Doctoral of Science, HTL (ASCP)
> >Novartis Institutes for BioMedical Research
>
>
> Toby C. Cornish, M.D., Ph.D.
> Pathology Resident
> Johns Hopkins Medical Institutions
>
[hidden email]
>