Jerman's 3D and 2D Hessian based tubular (vessel/vesselness) and spherical (blob/blobness) enhancement filters.
The MATLAB code is the implementation of the next two journal publications:
and a conference paper (please refer to journal publication [1] for more details):
The code is based on Dirk-Jan Kroon's implementation of Frangi's vesselness filter. (https://www.mathworks.com/matlabcentral/fileexchange/24409-hessian-based-frangi-vesselness-filter)
Make sure that the objects of interest have the highest (if bright compared to the background) or lowest (if dark compared to background) intensities in the image/volume. Scale/normalize the images appropriately.
The 3D method contains a c-code file that needs to be compiled with "mex eig3volume.c". (For more info visit: https://www.mathworks.com/matlabcentral/fileexchange/24409-hessian-based-frangi-vesselness-filter)
Threshold the filter response to remove any remaining enhanced noise
2D enhancement of vessel/tube-like structures:
3D enhancement of vessel/tube-like structures:
3D enhancement of blob/sphere-like structures:
blobness3D.m - main function
eig3volume.c - (as above)
example_blobness3D.m - aneurysm enhancement in a 3D cerebral vasculature
volume.mat - (as above)
2D enhancement of blob/disk-like structures (experimental - not published):
blobness2D.m - main function
example_blobness2D.m - enhancement of an aneurysm on a vessel (synthetically generated image)
blob2D.png - image for the example