Hello, maybe my question could be stupid. Is it possible to use bonej from hematoxylin and eosin staining? Thanks
could you please explain a bit more in detail, what you want to do? I mean, ImageJ is made for processing images, e.g. images which result from hematoxylin and eosin staining. But why do you think BoneJ makes sense and what exactly do you want to do? Furthermore, it may be interesting to see an example image you would like to process.
I apologize but the histomorphometry is not my expertise. I have to calculate various bone parameters (such as trabecular number, volume, SMI etc.) from hematoxylin and eosin figures.
BoneJ doesn’t yet handle histological sections - it’s much more geared towards 3D images from X-ray microtomography and clinical CT scans.
You might like to try Rob van ‘t Hof’s bone histomorphometry software. We’re aiming to bring Rob’s software into BoneJ2 over the next couple of years.
Final note: you will not be able to calculate SMI, which is a 3D parameter (and which doesn’t really do what it’s meant to do).
My 2 cents, after playing with H&E stainings a bit. If you have a sequence of serial sections, then you can start to take 3D measurements, but you’ll have to go through an image registration process first. This aligns the sections, because as we know even automated sectioning isn’t always perfect. There are some tools for that in ImageJ, like http://fiji.sc/Linear_Stack_Alignment_with_SIFT. I’m not the person to speak about any of BoneJ’s morphometric wonderments, but volume, for example, is a matter of:
- Register your serial sections (Plugins>Registration [consider Linear Stack Alignment with SIFT, TurboReg, and Elastic>Elastic stack alignment])
- Segment the relevant part of your stains (see http://fiji.sc/Trainable_Weka_Segmentation or, hop over to Ilastik to do your segmentation externally, then re-import your segmented image)
- Then you should actually be able to just use the 3D object counter (Analyze>3D Object Counter) to measure volume. Keep an eye on the calibration information for your images, especially if you’re using external programs for some operations, you may need to manually re-enter this before taking your measurements.
Note that registration and segmentation are hard problems (they also might be slow), but depending on your data, you might be able to get some usable results with this approach.
- It might also be worth trying both orders of operation: Register then Segment, and Segment then Register
I know this is quite a late response. Something to take into consideration is that while microct images are quite accurate in terms of dimensions, histological sections from paraffin embedded blocks suffer from a number of non-controllable distortions generated during the cutting that cannot be properly compensated when ‘floating’ the sections on warm water to ‘fish’ them out with a glass slide. Image registration from consecutive paraffin sections is problematic.
So it you are after accuracy, plastic embedding is probably better, although the sectioning is still a problem. In any event, there is shrinkage that happens during tissue fixation. Look around in the literature to horrify yourself of the % of shrinkage samples sometimes experience.
Just an update on my histomorphometry software: Our institute has changed its web pages, and although I was told this would not affect my web page for this software, it has. The new page is: