First of all I hope this is the correct category as I was unable to find a BoneJ category.
I have been trying to use the newly developed ellipsoid factor instead of SMI in BoneJ to obtain the plate-to-rod composition of a cube of trabecular bone. boneJ is able to produce lovely graphs describing the composition. However, I am unable to obtain the descriptive statistics or plots presented in this paper: http://journal.frontiersin.org/article/10.3389/fendo.2015.00015/full.
Could anyone advise me how to obtain the mean value and standard deviation of the EF?
Descriptive statistics for EF can be obtained by running a histogram ([h] on your keyboard, uncheck pixel value range, X min = -1, X max = 1, check ‘stack histogram’ ) on the EF-image.
You can get the numbers into your Results table by setting the measurements you want (e.g. StdDev, mean, min, max, mode, count) in Analyze > Set Measurements… then running Image > Stacks > Statistics.
Let us know how you get on with EF - it’s in active development.
Ellipsoid Factor - BoneJ
Ellipsoid factor problem getting values -1 to +1
I am trying to extract the a/b and b/c values from the EF-images, is it possible to export the data points of Flinn diagram so that I could plot it in another software like excel or origin?
Yes, but it’s a bit difficult. You would first have to eliminate Gaussian blurring by setting Gaussian sigma to 0, then convert each pixel position into ratio value, then get the intensity at each location.
It would be more straightforward to get Ellipsoid Factor to produce axis ratio images by checking that option in the EF dialog, then getting the histogram values for a/b and b/c from those images and exporting them into your statistics software.
@alessandrofelder is now working on EF full time so please tag him on further questions.
Thanks for the quick reply Michael! Really appreciate!
The thing is i want to do is plot a same scatter chart in Origin as Flinn diagram, i need to know the x value (b/c) and y value (a/b) at the same time at the same ellipsoid, so that i could do a curve fitting. I have tried to get the histogram for a/b from short-mid images and b/c from mid-long images, i could access the values of these two ratio, however, these two values might not match each other, which means the order of histogram is not based on ellipsoid, it is based on the value.
do you have any clue to exact these values?
Yes, of course you are right here.
You could try exporting the pixel values as text, in the format:
x y pixelValue
by running the built-in function Analyze > Tools > Save XY Coordinates… using 0 as the background value and leaving the two other boxes unchecked.
Note that the xy coordinates are in pixels so they have to be normalised by dividing by 511, and that you have to run Ellipsoid Factor with the Gaussian blurring (sigma) on the Flinn plot set to 0.