[Insight-users] destinguishing sheets from fibers

[Jeff Donner]

Are 'fibers' tube-like in shape (longer, than wide-around)? You might
want to look into the structure tensor as well. There was a recent
exchange on this list about that. Someone's still (afaik) working on
adding a filter to ITK but, there was a project that I did,
referenced, too.

On Wed, Aug 11, 2010 at 11:32 AM, Kishore Mosaliganti
<kishoreraom at gmail.com> wrote:
> I encountered the problem of separating sheets from fibers before. I
> developed some ITK methods using the Hessian and Frangi's measures for
> this very purpose and I can send you that code. I also wrote a paper:
>
> Anisotropic Plate Diffusion Filtering for Detection of Cell Membranes
> in 3D Microscopy Images". IEEE International Symposium on Biomedical
> Imaging: From Nano to Macro, 2010. ISBI '10, pp. 588-591
>
> On Wed, Aug 11, 2010 at 10:05 AM, Bradley Lowekamp
> <blowekamp at mail.nih.gov> wrote:
>> Helo,
>> Please be aware that there is an unfortunate bug in release 3.20 with regard
>> to this filter:
>> http://www.itk.org/Bug/view.php?id=11098
>>
>> On Aug 10, 2010, at 12:46 PM, lynx.abraxas at freenet.de wrote:
>>
>> On 20/07/10 17:41:32, Sergio Vera wrote:
>>
>> Using the hessian you can look for tubular structures easily. ITK has some
>>
>> functions related oriented do enhance vessels (that are similar to your
>>
>> fibers).
>>
>> There are also minor modifications of the hessian filter that sould be able
>>
>> to respond well to plate-like structures as far as I remember...
>>
>> Look for Alejandro Frangi Vesselness measure. His paper explains the
>>
>> combination of eigenvalues of the Hessian that respond well to tubes and to
>>
>> plates.
>>
>>
>> Thanks a lot Sergio! This seems to be a very promising solution to my
>> problem.
>> I'll try the HessianObjectnessFilter.
>>
>> I wonder though if it would be better to apply the filter before
>>  binarization
>> (thresholding) of a grey image or on the binarized (thresholded) image?