I have two images of the same sample with different x,y,z dimensions. Can anyone explain - in order to quantify the same object - why do I need to enter larger number of voxels in the size filter (for an image with higher voxel size) than compared to a different image of the same sample (with lesser voxel size)?
http://rsbweb.nih.gov/ij/plugins/track/objects.html
What you might want to do in first place is to put the correct x/y/z dimensions, i.e. the voxel size into Image>>Properties before setting your threshold for the size filter - or are those loaded into ImageJ automatically?
Typically one wants to filter locally; e.g. in a 6mm neighborhood. If the voxel-size in one scan is 1mm (for all three axes) and 2mm in the other scan, then the filter sizes would be 13 and 7 respectively.
@ Michael A. Sandholzer - yes the dimensions are already loaded in the imageJ for the images
@ Lambert Zijp - That is exactly my concern. I realise the same way as you have mentioned and logically that should be the case but when I try to filter according to the 13 no. of voxels (for 1mm x,y,z) and 7 voxels (2mm for x,y,z) the object map does not seem to filter or quantify the objects correctly. I have no clue as to why this is happening.
I don't use ImageJ, but just read the documentation of the object counter. It doesn't say anything about voxel-sizes, nor about filtering. Maybe you get the objects back where their volumes are expressed in number of voxels instead of mm^3.
@ Lambert Zijp- I read about the size filter in another documentation where they have explained how the plugin works. In that document it is clearly mentioned that it takes into account the no. of voxels
That filter-size is a kind of threshold. It is passed as the minimum or maximum number of voxels that an object is allowed to have. It has nothing to do with the kernel-size of e.g. a convolution filter.
Ask on the imagej mailing list where those who know the answers are mist like to be
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
- Image Processing