Imperial College London

ProfessorDanielRueckert

Faculty of EngineeringDepartment of Computing

Head of Department of Computing
 
 
 
//

Contact

 

+44 (0)20 7594 8333d.rueckert Website

 
 
//

Location

 

568Huxley BuildingSouth Kensington Campus

//

Summary

 

Publications

Citation

BibTex format

@article{Cardoso:2015:10.1109/TMI.2015.2418298,
author = {Cardoso, MJ and Modat, M and Wolz, R and Melbourne, A and Cash, D and Rueckert, D and Ourselin, S},
doi = {10.1109/TMI.2015.2418298},
journal = {IEEE Transactions on Medical Imaging},
pages = {1976--1988},
title = {Geodesic Information Flows: Spatially-Variant Graphs and Their Application to Segmentation and Fusion},
url = {http://dx.doi.org/10.1109/TMI.2015.2418298},
volume = {34},
year = {2015}
}

RIS format (EndNote, RefMan)

TY  - JOUR
AB - Clinical annotations, such as voxel-wise binary orprobabilistic tissue segmentations, structural parcellations, pathologicalregions-of-interest and anatomical landmarks are key tomany clinical studies. However, due to the time consuming natureof manually generating these annotations, they tend to be scarceand limited to small subsets of data. This work explores a novelframework to propagate voxel-wise annotations between morphologicallydissimilar images by diffusing and mapping the availableexamples through intermediate steps. A spatially-variant graphstructure connecting morphologically similar subjects is introducedover a database of images, enabling the gradual diffusion ofinformation to all the subjects, even in the presence of large-scalemorphological variability. We illustrate the utility of the proposedframework on two example applications: brain parcellation usingcategorical labels and tissue segmentation using probabilistic features.The application of the proposed method to categorical labelfusion showed highly statistically significant improvements whencompared to state-of-the-art methodologies. Significant improvementswere also observed when applying the proposed frameworkto probabilistic tissue segmentation of both synthetic and real data,mainly in the presence of large morphological variability.
AU - Cardoso,MJ
AU - Modat,M
AU - Wolz,R
AU - Melbourne,A
AU - Cash,D
AU - Rueckert,D
AU - Ourselin,S
DO - 10.1109/TMI.2015.2418298
EP - 1988
PY - 2015///
SN - 1558-254X
SP - 1976
TI - Geodesic Information Flows: Spatially-Variant Graphs and Their Application to Segmentation and Fusion
T2 - IEEE Transactions on Medical Imaging
UR - http://dx.doi.org/10.1109/TMI.2015.2418298
UR - http://hdl.handle.net/10044/1/30755
VL - 34
ER -