[Insight-users] Suitable parameters for Level set algorithms needed.

[Iván Macía]

Hi Philippe,

The values you have to choose for the different filters depending totally of
your application. For example, you don't mention anything about your data
(image type, modality, what are you trying to segment, etc). Please,
understand that setting all the parameters for the different filters in the
itk Level Set examples at once without understanding their influence in
order to obtain a good segmentation is like trying to choose some words
randomly and pretend they make sense together.

I would recommend you to run the examples in the itk Software Guide with the
data used there, trying to understand the effect and purpose of the
different filters/parameters. For this you don't need to know how the
differential equation evolves under the hood. The Software Guide will give
you a general understanding. There, it is explained for example how a good
feature image or an initial Level-Set should look like (maybe you have
already done this).

You can also run some of the programs in InsightApplications (i.e.
ShapeDetectionLevelSet application) that may give you a better visual
feedback on what is going on when you change the value of a certain
parameter.  Then try to reproduce that behaviour with you own data, step by
step. For example a bad feature image will give you a bad segmentation
result no matter the parameters you use in the Level-Set filter itself.
Then, if you have any further questions don't hesitate to ask.

Hope that helps

Iván


__________________________________
Ivan Macia Oliver
VICOMTech - Visual Interaction and
Communication Technologies Center
Paseo Mikeletegi 57
20009 San Sebastian
Spain
Tel: +34 943 30 92 30
Fax: +34 943 30 93 93
e-mail: imacia at vicomtech.es
http://www.vicomtech.es
 
*** member of INI-GraphicsNet ***
http://www.inigraphics.net

VICOMTech is an ISO-9001:2000
certified institute







-----Mensaje original-----
De: insight-users-bounces+imacia=vicomtech.es at itk.org
[mailto:insight-users-bounces+imacia=vicomtech.es at itk.org] En nombre de
Rainbowfrog at web.de
Enviado el: lunes, 17 de octubre de 2005 21:37
Para: insight-users at itk.org
Asunto: [Insight-users] Suitable parameters for Level set algorithms needed.



Hello 

I took the code found in :

FastMarchingImageFilter.cxx
ShapeDetectionLevelSetFilter.cxx
ThresholdSegmentationLevelSetImageFilter.cxx

of the Itk examples, and modified it to fit into my segmentation program. I
allow the user to set the most important parameters for these algorithms
over a dialog via sliders. I don't know much about the mathematical
background of those algorithms though (for example I couldn't solve
differential equations by myself, but I have read in the Itk software guide
that the algorithms use them) that's why I don't know for some of the
variables which standard values make sens, and which range of values I
should allow the user to set (for a slider bar this means,depending on the
parameter: which values I should take as minimum 
and maximum for the user to set). I have listed the default values for the
parameter sliders that I use for each of the algorithms, some of them were
in the examples , and for some of them I just guessed a value.

I don't get good results with them, for the ShapeDetectionLevelSet I don't
even 
get a valid result. Which standard values , minimum values , and maximum
values would make sens for them ? Does anyone know good starting values ?

I'd be glad if someone could help me with this.

Regards
Philipp

Standart parameters I'm using:


FastMarching:

	For the Smoothingfilter
		(int) Iterations = 5
		(double) Conductance = 9.0

	For the main algorithm:
		(double) timeTreshold = 3.0
		(double) sigma =3.0
		(double) alpha =-3.0
		(double) beta  =3.0
		(double) stoppingTime = 15.0

		+ at least one seed point


Shape Detection:

	For the smoothingfilter
		(int) Iterations = 5;
		(double) Conductance = 9.0;

	For the main algorithm
		(double) sigma =3.0;
		(double) alpha =(-3.0);
		(double) beta  =3.0;
		(double) initDistance=0.0;
		(double) curvatureScaling = 10.0;
		(double) propagationScaling = 1.0;
		(double) maximumRmsError =0.02;
		(int)    Iterations =800;

		+ at least 1 seed point

Threshold Level Set:

	(double) initDistance=0.0;
	(double) curvatureScaling = 1.0;
	(double) propagationScaling = 1.0;
	(double) maximumRmsError =0.02;
	(double) upperTreshold =250.0;
	(double) lowerTreshold=5.0;
	(int)    Iterations =1200;

	+ at least 1 seed point



______________________________________________________________
Verschicken Sie romantische, coole und witzige Bilder per SMS! Jetzt bei
WEB.DE FreeMail: http://f.web.de/?mc=021193

_______________________________________________
Insight-users mailing list
Insight-users at itk.org http://www.itk.org/mailman/listinfo/insight-users

-- 
No virus found in this incoming message.
Checked by AVG Anti-Virus.
Version: 7.0.344 / Virus Database: 267.12.2/140 - Release Date: 18/10/2005
 

-- 
No virus found in this outgoing message.
Checked by AVG Anti-Virus.
Version: 7.0.344 / Virus Database: 267.12.2/140 - Release Date: 18/10/2005