User Tools

Site Tools


netscape_list

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
netscape_list [2014/05/31 17:33]
admin created
netscape_list [2014/05/31 17:41] (current)
admin
Line 1: Line 1:
-!!!Evaluation of Appearance Models of the Brain+!!!Netscape candidates
  
-!!Abstract 
  
-Appearance models are an applicable approach to the +* jeremytoday (+3) - http://www.netscape.com/​member/​jeremytoday
-analysis of anatomical variabilityThey are able to +
-distinguish between groups, e.g. normal and diseased, +
-as a model encapsulates the properties of a group from +
-which it was derived. The construction of such models +
-is closely-related to the task of registration and they +
-require the correspondence,​ which registration is able +
-to obtain.+
  
-We developed a framework which evaluates appearance +* pagey (+2/-1) - http://www.netscape.com/​member/​pagey
-models, based on the statistics of large sets of +
-imagesThe framework is capable of distinguishing +
-between good models of the brain and worse ones. +
-Furthermore,​ it provides a method of validating models. +
-It does so by measuring how well a model and its data +
-fit together.+
  
-Two measures are defined which reflect on the quality 
-of a model. The first of these -- specificity -- 
-approximates the level to which data generated by the 
-model fits data from which the model was constructed. 
-The complementary measure -- generalisation -- is able 
-to quantify '​distance'​ between data from which the 
-model was constructed and model-generated data. 
  
-Results show that as models degrade in quality, their +!!!Digg candidates
-specificity and generalisation ability rise, as +
-expected. The algorithms are used to compare models of +
-the brains, which were built automatically by +
-independent approaches of registration. This greatly +
-helps in identifying better model construction +
-algorithms, which are analogous to registration algorithms.+
  
-!!1 Introduction 
  
-A powerful method for the modelling of anatomy was +* aidenag ( +4 ) http://digg.com/​users/​aidenag
-introduced by Edwards et al. [Edwards] and it is known as +
-appearance models -- a natural successor to shape +
-models [Cootes]. This method requires a large enough set of +
-data, which is representative of a population and +
-ideally spans its full variability. Appearance models +
-are able to learn what characterises inter-subject +
-changes and determine the prominence of the main +
-characteristics. Hence, it is able to identify changes +
-and derive a model that encapsulates change -- all in a +
-data-driven manner.+
  
-Non-rigid image registration is ubiquitously used as +* curtissthompson ( + 3 ) http://digg.com/​users/​curtissthompson
-the basis for analysis of medical images. The results +
-of registration can be used for structural analysis, +
-atlas matching, and analysis of change. Methods for +
-obtaining registration are are well-established and +
-quite uniform. The goal is achieved by warping pairs of +
-images so that they appear more similar. The similarity +
-leads to overlap, which allows corresponding structures +
-to be identified. This problem is complementary to that +
-of modelling groups of images. A statistical model of a +
-group of images needs dense correspondence to be +
-defined across the group; non-rigid registration +
-provides exactly that.+
  
-Since the emergence of appearance models, attempts have +* GregD (+2) http://www.digg.com/​users/​gregd ​ &    http://www.treelimb.org
-been made to reproduce and improve it. To name a few +
-such efforts, Stegmann [Stegmann] built 4-dimensional cardiac +
-models and Reuckert et al[Rueckert] derived statistical +
-deformation models from several registrations of the +
-brainModels have been built in a variety of ways, but +
-what is yet lacked is the ability to compare them. It +
-becomes clear from experience that attempts to +
-distinguish between them by eyesight is hopeless. More +
-recently, appearance models were built automatically +
-using piece-wise affine registration [IPMI - YET TO +
-ADD]Evaluation of models in this particular case +
-enables evaluation of registration algorithms.+
  
-The idea of evaluating models is not unexampled. Davies +!!!Newsvine candidates 
-et al. [Davies] explored the evaluation of shape models and + 
-ultimately developed a robust framework. This paper +* 
-outlines a principled approach to the evaluation of + 
-appearance models, which is a challenging task since +!!!Reddit candidates 
-their complexity is very high. The approach is shown to + 
-be reliable in evaluation of brain models ​(FOOTERExamples from non-medical domains are beyond the remit +* Jeff (+5) - http://​reddit.com/​user/​Fedquip/​ 
-of this paper, but they have been very successful.) and it is then used to learn about registration + 
-algorithms, from which appearance models have been derived.+* Jeroen (+3) http://​reddit.com/​user/​BioGeek/​ 
 + 
 + 
 +!!!Blogger candidates 
 + 
 + 
 +* Steven(insidesocialnews.blogspot.com) (+1/-1
 +* Chris (http://www.windingroad.com/​) (+2) 
 + 
 +!!!Delicious candidates 
 + 
 +*
netscape_list.txt · Last modified: 2014/05/31 17:41 by admin