<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=us-ascii">
<META NAME="Generator" CONTENT="MS Exchange Server version 6.5.6944.0">
<TITLE>RE: [Insight-users] ITK ROAD MAP 2005-2006 : Call for feedback</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/rtf format -->
<P><FONT SIZE=2 FACE="Arial">Hi Zach,Martin and other interested users/developers,</FONT>
<BR><FONT SIZE=2 FACE="Arial">It is great to hear about emerging interest in enhancing support for MRF filters in ITK. I would like to add a couple of things to Jim's mail.</FONT> </P>
<P><FONT SIZE=2 FACE="Arial">The</FONT> <FONT SIZE=2 FACE="Courier New">itk::MRFImageFilter</FONT> <FONT SIZE=2 FACE="Arial"> class implements Besag's classical MRF filter where the classification labels are iteratively updated . This class was envisioned to be a base implementation, which could be extended to other MRF realizations. One idea would be to capture the different classes of MRFs the ITK users community would like to have. It would be very helpful to hear from someone in the community who has worked in this area and is willing to share their experiences.</FONT></P>
<P><FONT SIZE=2 FACE="Arial">Thanks,<BR>
Sayan</FONT>
</P>
<P><FONT SIZE=2 FACE="Arial"></FONT> <FONT SIZE=2 FACE="Courier New">Date: Tue, 31 May 2005 11:56:07 -0400</FONT>
<BR><FONT SIZE=2 FACE="Courier New">From: "Miller, James V (Research)" <millerjv@crd.ge.com></FONT>
<BR><FONT SIZE=2 FACE="Courier New">Subject: RE: [Insight-users] ITK ROAD MAP 2005-2006 : Call for</FONT>
<BR> <FONT SIZE=2 FACE="Courier New">feedback</FONT>
<BR><FONT SIZE=2 FACE="Courier New">To: "Zachary Pincus" <zpincus@stanford.edu>, "ITK mailing"</FONT>
<BR> <FONT SIZE=2 FACE="Courier New"><insight-users@itk.org></FONT>
<BR><FONT SIZE=2 FACE="Courier New">Message-ID:</FONT>
<BR> <FONT SIZE=2 FACE="Courier New"><FA26BEF1EA775E4584FB34B91E14A1C4B419A2@SCHMLVEM01.e2k.ad.ge.com></FONT>
<BR><FONT SIZE=2 FACE="Courier New">Content-Type: text/plain; charset="Windows-1252"</FONT>
</P>
<P><FONT SIZE=2 FACE="Courier New">Zachary, </FONT>
</P>
<P><FONT SIZE=2 FACE="Courier New">Is the current itk::MRFImageFilter not sufficient? Does it need to be</FONT>
<BR><FONT SIZE=2 FACE="Courier New">refactored to accomadate different MRF algorithms?</FONT>
</P>
<P><FONT SIZE=2 FACE="Courier New">I only used the MRFImageFilter once. It seemed to perform as I would have</FONT>
<BR><FONT SIZE=2 FACE="Courier New">expected (from reading some of the literature).</FONT>
</P>
<P><FONT SIZE=2 FACE="Courier New">One thing I think ITK probably needs are more techniques for learning the </FONT>
<BR><FONT SIZE=2 FACE="Courier New">pdf's for each class of material.</FONT>
</P>
<P><FONT SIZE=2 FACE="Courier New">Jim</FONT>
</P>
<BR>
<BR>
<BR>
<P><FONT SIZE=2 FACE="Courier New">-----Original Message-----</FONT>
<BR><FONT SIZE=2 FACE="Courier New">From: insight-users-bounces+millerjv=crd.ge.com@itk.org</FONT>
<BR><FONT SIZE=2 FACE="Courier New">[</FONT><A HREF="mailto:insight-users-bounces+millerjv=crd.ge.com@itk.org]On"><U><FONT COLOR="#0000FF" SIZE=2 FACE="Courier New">mailto:insight-users-bounces+millerjv=crd.ge.com@itk.org]On</FONT></U></A><FONT SIZE=2 FACE="Courier New"> Behalf Of</FONT>
<BR><FONT SIZE=2 FACE="Courier New">Zachary Pincus</FONT>
<BR><FONT SIZE=2 FACE="Courier New">Sent: Saturday, May 28, 2005 7:37 PM</FONT>
<BR><FONT SIZE=2 FACE="Courier New">To: ITK mailing</FONT>
<BR><FONT SIZE=2 FACE="Courier New">Subject: Re: [Insight-users] ITK ROAD MAP 2005-2006 : Call for feedback</FONT>
</P>
<BR>
<P><FONT SIZE=2 FACE="Courier New">Hi all,</FONT>
</P>
<P><FONT SIZE=2 FACE="Courier New">After looking over the 2005-2006 ITK roadmap, I've also got a couple of </FONT>
<BR><FONT SIZE=2 FACE="Courier New">questions/comments on the machine learning aspects.</FONT>
</P>
<P><FONT SIZE=2 FACE="Courier New">Specifically, to what ends are classification algorithms (e.g. gaussian </FONT>
<BR><FONT SIZE=2 FACE="Courier New">mixture models, k-nearest neighbors, putative neural networks or SVMs) </FONT>
<BR><FONT SIZE=2 FACE="Courier New">present in ITK? It strikes me that one major use of such algorithms in </FONT>
<BR><FONT SIZE=2 FACE="Courier New">medical imaging is for classification of image pixels into various </FONT>
<BR><FONT SIZE=2 FACE="Courier New">tissue types, e.g. grey matter vs. white matter.</FONT>
</P>
<P><FONT SIZE=2 FACE="Courier New">If this is the case, I would think that adding Markov Random Field </FONT>
<BR><FONT SIZE=2 FACE="Courier New">capabilities to ITK would be a big win. Basically, MRFs allow users to </FONT>
<BR><FONT SIZE=2 FACE="Courier New">add priors about the *spatial* distribution of various pixel types into </FONT>
<BR><FONT SIZE=2 FACE="Courier New">the classification process. For example, a single isolated pixel </FONT>
<BR><FONT SIZE=2 FACE="Courier New">initially labeled as "grey matter" in a blob of white matter might </FONT>
<BR><FONT SIZE=2 FACE="Courier New">(depending on the priors) be considered an unlikely configuration and </FONT>
<BR><FONT SIZE=2 FACE="Courier New">thus be re-labeled in the final MRF configuration. Such spatial </FONT>
<BR><FONT SIZE=2 FACE="Courier New">considerations are ignored by traditional classifiers.</FONT>
</P>
<P><FONT SIZE=2 FACE="Courier New">Because spatial information is so important, and MRFs are a relatively </FONT>
<BR><FONT SIZE=2 FACE="Courier New">easy way to add simple spatial priors, they have become quite popular </FONT>
<BR><FONT SIZE=2 FACE="Courier New">in the image processing literature. I think filters to estimate the MAP </FONT>
<BR><FONT SIZE=2 FACE="Courier New">MRF given an input "label images" (e.g the results of pixel-wise </FONT>
<BR><FONT SIZE=2 FACE="Courier New">classification) would be a very valuable addition, especially if the </FONT>
<BR><FONT SIZE=2 FACE="Courier New">stable of pixel classification methods in ITK is to expand.</FONT>
</P>
<P><FONT SIZE=2 FACE="Courier New">Now, I haven't described in too much detail how MRF models actually </FONT>
<BR><FONT SIZE=2 FACE="Courier New">work. A Google Scholar search for "Markov random field image" will show </FONT>
<BR><FONT SIZE=2 FACE="Courier New">the breadth of utilization of MRFs in the imaging literature. Here is a </FONT>
<BR><FONT SIZE=2 FACE="Courier New">good introduction to MRF segmentation, with specific reference to MRI </FONT>
<BR><FONT SIZE=2 FACE="Courier New">images:</FONT>
<BR><FONT SIZE=2 FACE="Courier New">Segmentation of brain MR images through a hidden Markov random field </FONT>
<BR><FONT SIZE=2 FACE="Courier New">model and the expectation-maximization algorithm.</FONT>
<BR><FONT SIZE=2 FACE="Courier New">Y Zhang, M Brady, S Smith - IEEE Trans Med Imaging, 2001</FONT>
<BR><A HREF="http://www.cvmt.dk/~hja/teaching/cv/HMRF_EM_BRAIN.pdf"><U><FONT COLOR="#0000FF" SIZE=2 FACE="Courier New">http://www.cvmt.dk/~hja/teaching/cv/HMRF_EM_BRAIN.pdf</FONT></U></A>
</P>
<P><FONT SIZE=2 FACE="Courier New">I would be happy to discuss at (much) more length how a MRF </FONT>
<BR><FONT SIZE=2 FACE="Courier New">"classification cleanup" filters could be implemented in ITK, if there </FONT>
<BR><FONT SIZE=2 FACE="Courier New">is any interest in these methods.</FONT>
</P>
<P><FONT SIZE=2 FACE="Courier New">Zach</FONT>
</P>
<BR>
<BR>
<P><FONT SIZE=2 FACE="Courier New">On May 27, 2005, at 2:46 PM, Lino Ramirez wrote:</FONT>
</P>
<P><FONT SIZE=2 FACE="Courier New">> Hi Luis and ITK Users/Developers,</FONT>
<BR><FONT SIZE=2 FACE="Courier New">></FONT>
<BR><FONT SIZE=2 FACE="Courier New">> I had a brief look at the ITK roadmap 2005-2006. It looks quite</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> impressive. I cannot wait until having available all these tools in one</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> single package ;-)</FONT>
<BR><FONT SIZE=2 FACE="Courier New">></FONT>
<BR><FONT SIZE=2 FACE="Courier New">> I have some small comments/questions about functionalities I would </FONT>
<BR><FONT SIZE=2 FACE="Courier New">> like to</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> see in the toolkit.</FONT>
<BR><FONT SIZE=2 FACE="Courier New">></FONT>
<BR><FONT SIZE=2 FACE="Courier New">> I noticed that Neural Networks will be added to the toolkit. Are there </FONT>
<BR><FONT SIZE=2 FACE="Courier New">> any</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> plans for adding a Support Vector Machines (SVM) [1] implementation? </FONT>
<BR><FONT SIZE=2 FACE="Courier New">> SVM</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> have been used successfully in a variety of applications that could be </FONT>
<BR><FONT SIZE=2 FACE="Courier New">> of</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> interest to the ITK community (see [2] for some sample applications).</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> Moreover, it is always good to have a machine learning approach that is</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> similar to the neural networks in architecture but that uses a </FONT>
<BR><FONT SIZE=2 FACE="Courier New">> different</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> learning strategy. In this way, one could try the two of them and</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> determine which one is more appropriate for a particular dataset.</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> Sometimes, in datasets in which the neural networks fail the SVM </FONT>
<BR><FONT SIZE=2 FACE="Courier New">> succeed</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> and vice versa.</FONT>
<BR><FONT SIZE=2 FACE="Courier New">></FONT>
<BR><FONT SIZE=2 FACE="Courier New">> Are there any plans (even in the very long term) to add support for </FONT>
<BR><FONT SIZE=2 FACE="Courier New">> Fuzzy</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> Sets [3], Fuzzy Geometry [4], and Fuzzy Spatial Relations [5] between</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> objects in an image. I think these concepts would be invaluable in the</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> future of medical image analysis. For example, when we want to measure</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> geometric properties in objects in an image, we find that generally the</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> objects are not crisply defined (due to errors during the segmentation,</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> errors in the acquisition of the image, or errors in the definition of </FONT>
<BR><FONT SIZE=2 FACE="Courier New">> the</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> object</FONT> <FONT SIZE=2 FACE="Courier New">–</FONT><FONT SIZE=2 FACE="Courier New">where do the ribs start and the vertebrae end in a spine </FONT>
<BR><FONT SIZE=2 FACE="Courier New">> X-ray).</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> In this case, fuzzy geometry could be used to compute the object</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> properties. Another example would be in the identification of objects </FONT>
<BR><FONT SIZE=2 FACE="Courier New">> in</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> the images. For instance, in the internal brain structures, the right</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> caudate nucleus should be closer to the right lateral ventricle than to</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> the left lateral ventricle. Fuzzy spatial relations with the help of </FONT>
<BR><FONT SIZE=2 FACE="Courier New">> fuzzy</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> logic [6] could be used to develop a system that makes use of that </FONT>
<BR><FONT SIZE=2 FACE="Courier New">> piece</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> of information to identify right lateral ventricle.</FONT>
<BR><FONT SIZE=2 FACE="Courier New">></FONT>
<BR><FONT SIZE=2 FACE="Courier New">> Well, those are my two picks ;-)</FONT>
<BR><FONT SIZE=2 FACE="Courier New">></FONT>
<BR><FONT SIZE=2 FACE="Courier New">> I am looking forward to any comment</FONT>
<BR><FONT SIZE=2 FACE="Courier New">></FONT>
<BR><FONT SIZE=2 FACE="Courier New">> Take care</FONT>
<BR><FONT SIZE=2 FACE="Courier New">></FONT>
<BR><FONT SIZE=2 FACE="Courier New">> Lino</FONT>
<BR><FONT SIZE=2 FACE="Courier New">></FONT>
<BR><FONT SIZE=2 FACE="Courier New">> [1] C. Cortes and V. Vapnik, "Support-Vector Networks," Machine </FONT>
<BR><FONT SIZE=2 FACE="Courier New">> Learning,</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> vol. 20, pp. 273-297, 1995</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> [2] </FONT><A HREF="http://www.clopinet.com/isabelle/Projects/SVM/applist.html"><U><FONT COLOR="#0000FF" SIZE=2 FACE="Courier New">http://www.clopinet.com/isabelle/Projects/SVM/applist.html</FONT></U></A>
<BR><FONT SIZE=2 FACE="Courier New">> [3] L.A. Zadeh, "Fuzzy sets," Information and Control, vol. 8, pp. </FONT>
<BR><FONT SIZE=2 FACE="Courier New">> 38-352,</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> 1965</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> [4] A. Rosenfeld, "Fuzzy geometry: An updated overview," Information</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> Sciences, vol. 110, pp. 127-133, 1998</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> [5] I. Bloch, "Fuzzy spatial relationships for image processing and</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> interpretation: a review," Image and Vision Computing, vol. 23, pp.</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> 89-110, 2005</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> [6] L.A. Zadeh, "Outline of a new approach to the analysis of complex</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> systems and decision processes," IEEE Transactions on Systems, Man, and</FONT>
<BR><FONT SIZE=2 FACE="Courier New">> Cybernetics, vol. SMC-3, no. 1, pp. 28-44, 1973</FONT>
<BR><FONT SIZE=2 FACE="Courier New">></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> A first draft of the road map for ITK development/maintenance has</FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> been crafted for the period of September 2005 - September 2006.</FONT>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> You will find this draft as a link to the Oversight Committee page</FONT>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> </FONT><A HREF="http://www.itk.org/Wiki/ITK_Oversight_Committee"><U><FONT COLOR="#0000FF" SIZE=2 FACE="Courier New">http://www.itk.org/Wiki/ITK_Oversight_Committee</FONT></U></A>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> More specifically at</FONT>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> </FONT><A HREF="http://www.itk.org/Wiki/ITK_Roadmap_2005_2006"><U><FONT COLOR="#0000FF" SIZE=2 FACE="Courier New">http://www.itk.org/Wiki/ITK_Roadmap_2005_2006</FONT></U></A>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> The purpose of this road map is to plan for features and </FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> functionalities</FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> to be included in ITK in the near/medium term (1 to 2 years).</FONT>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> The addition of these features should make of ITK a better tool for</FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> supporting your efforts in medical research, and development of </FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> medical</FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> applications.</FONT>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> The road map also includes the maintenance tasks to be undertaken in</FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> ITK. This may involve refactoring of classes, deprecation of classes,</FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> additional testing, additional coverage, improvements on tutorials and</FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> so on.</FONT>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> Please let us know of the features that you would like to see in ITK</FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> in the upcoming future, and what points of the toolkit you consider</FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> that can be improved in order to better server the community.</FONT>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> Thanks</FONT>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>></FONT>
<BR><FONT SIZE=2 FACE="Courier New">>> Luis</FONT>
</P>
<BR>
</BODY>
</HTML>