BoneJ create ROI of cortical bone in femoral cross section stack


I have a binary stack (80 slices) of the femoral shaft and wound like to use the BV/TV function within bonej. However I cannot to create an ROI that includes only the cortical bone (bone and cortical spaces) throughout a stack so would like exclude the trabecular bone and background.

Any help much appreciated.

Thank youBinary 55cm0008.tif (164.5 KB)

1 Like

If this is the only image that you have, the easiest thing to do is manually trace using the ROI tools and delete the trabecular bone by filling with a background pixel value (use the eye-dropper tool to pick a background value).

Bear in mind that the BV/TV values that you get will be calculated from BV = cortical bone volume (minus any pores: Haversian canals and osteocyte lacunae if resolved) and TV = volume of the image stack.

Thank you for your reply.

We have a small stack of images (80 slices) of a 1cm thick section of the femoral shaft. Ideally we wanted to be able to recognize the cotical bone throughout the stack to use the bv/tv function. The cortex changes in shape and size over the 80 images so an initial function to detect the cortex would be ideal.

We then have to apply this to multiple femoral shaft stacks and compare the bv/tv. We realise other users have selected a small volume or sphere of cortical bone but we would, if possible, like to select the whole cortex. Thank you

How many stacks? Unless it’s hundreds, it will probably be quicker to remove the trabeculae by manual drawing than setting up a new image processing pipeline. These solutions tend not to generalise well so it’s hard to get your process development time paid back, unless your total dataset is really big.

See e.g. this from XKCD

(Also - it’s really hard even for trained human operators to decide consistently where the boundary is between trabeculae and cortex, how are you going to tell whether your algorithm is working properly?)