[Insight-users] Re: 2D vesselness code submission

[Moti Freiman]

Hi,

I took a look at your code, and i have a design remark whihc reltaed 
also to the original itk vesselness filter:

You follow the design of the itk vesselness filter (proposed by sato), 
which did not incorporate the multi-scale strategy into the filter, and 
leave it to the user. I Prefer another strategy, which incorporate the 
multi-scale strategy to the filter itself. For instance: add the 
functions: setSigma, setNumberOfScales. setStartScale.

 

Another practical issue, that i found when i tried to use that filter:

I implemented the frangi 3D filter, and he worked well for areas with 
vessels and a uniform background. However, when i tried to apply this 
filter onto liver images (where the background is the liver and the air 
- therefore there are gradients in the background also) I encounter a 
noisy result, in the border of the liver and the air. The vessels result 
by itself was good.



You wrote:

Message: 2
Date: Sun, 25 Jun 2006 16:14:26 -0700 (PDT)
From: Seniha Esen Yuksel <eseny99 at yahoo.com>
Subject: [Insight-users] 2D vesselness code submission
To: insight-users at itk.org
Message-ID: <20060625231426.27625.qmail at web54513.mail.yahoo.com>
Content-Type: text/plain; charset="iso-8859-1"

Hi all,
 
 I have implemented the 2D vesselness code by 
 
 A.F. Frangi, W.J. Niessen, K.L.Vincken, M.A. Viergever
 "Multiscale vessel enhancement filtering" 
 MICCAI 1998, pp. 130-137.
 
 The results look fine, but I guess I was expecting more. I was wondering if anyone would be interested to take a look at the code, make suggesstions if I am skipping sth from the paper, or contribute to ITK.
 
 The paper, code, a test image, and its results are given at:
 http://www.cise.ufl.edu/~seyuksel/MyCodes/Vesselness2DCode/ 
 
 The test results are obtained by setting the sigma from 2 to 6 with 0.5 increments.
 
 The code is written by getting help mostly from itkHessian3DToVesselnessMeasureImageFilter, but I needed a 2D version; and this paper was easy to implement and it had a 2D version described in it. The paper I mentioned also has  a 3D version; so I am wondering if anyone has any experience in which one would perform better; and if it would be valuable to implement the aforomentioned paper vs. the itkHessian3DToVesselnessMeasureImageFilter.
 
 Thanks!
 Esen
 
 
 
 

Seniha Esen Yuksel
 University of Florida


-- 
__
Moti Freiman, Ph.D Student.
Medical Image Processing and Computer-Assisted Surgery Laboratory.
School of Computer Science and Engineering.
The Hebrew University of Jerusalem Givat Ram, Jerusalem 91904, Israel
Phone: +(972)-2-658-5371 (laboratory)
E-mail: freiman at cs.huji.ac.il
WWW site: http://www.cs.huji.ac.il/~freiman