Extraction of airways from CT (EXACT’09)
Research output: Contribution to journal › Journal article › Research › peer-review
Standard
Extraction of airways from CT (EXACT’09). / Lo, Pechin Chien Pau; Ginneken, Bram van; Reinhardt, Joseph M.; Yavarna, Tarunashree; Jong, Pim A. de; Irving, Benjamin; Fetita, Catalin; Ortner, Margarete; Pinho, Rômulo; Sijbers, Jan; Feuerstein, Marco; Fabijanska, Anna; Bauer, Christian; Beichel, Reinhard; Mendoza, Carlos S.; Wiemker, Rafael; Lee, Jaesung; Reeves, Anthony P.; Born, Silvia; Weinheimer, Oliver; Rikxoort, Eva M. van; Tschirren, Juerg; Mori, Ken; Odry, Benjamin; Naidich, David P.; Hartmann, Ieneke; Hoffman, Eric A.; Prokop, Mathias; Pedersen, Jesper H.; de Bruijne, Marleen.
In: I E E E Transactions on Medical Imaging, Vol. 31, No. 11, 2012, p. 2093-2107.Research output: Contribution to journal › Journal article › Research › peer-review
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - JOUR
T1 - Extraction of airways from CT (EXACT’09)
AU - Lo, Pechin Chien Pau
AU - Ginneken, Bram van
AU - Reinhardt, Joseph M.
AU - Yavarna, Tarunashree
AU - Jong, Pim A. de
AU - Irving, Benjamin
AU - Fetita, Catalin
AU - Ortner, Margarete
AU - Pinho, Rômulo
AU - Sijbers, Jan
AU - Feuerstein, Marco
AU - Fabijanska, Anna
AU - Bauer, Christian
AU - Beichel, Reinhard
AU - Mendoza, Carlos S.
AU - Wiemker, Rafael
AU - Lee, Jaesung
AU - Reeves, Anthony P.
AU - Born, Silvia
AU - Weinheimer, Oliver
AU - Rikxoort, Eva M. van
AU - Tschirren, Juerg
AU - Mori, Ken
AU - Odry, Benjamin
AU - Naidich, David P.
AU - Hartmann, Ieneke
AU - Hoffman, Eric A.
AU - Prokop, Mathias
AU - Pedersen, Jesper H.
AU - de Bruijne, Marleen
PY - 2012
Y1 - 2012
N2 - This paper describes a framework for establishing areference airway tree segmentation, which was used to quantitativelyevaluate 15 different airway tree extraction algorithms ina standardized manner. Because of the sheer difficulty involvedin manually constructing a complete reference standard fromscratch, we propose to construct the reference using results fromall algorithms that are to be evaluated. We start by subdividingeach segmented airway tree into its individual branch segments.Each branch segment is then visually scored by trained observersto determine whether or not it is a correctly segmented part of theairway tree. Finally, the reference airway trees are constructed bytaking the union of all correctly extracted branch segments. Fifteenairway tree extraction algorithms from different research groupsare evaluated on a diverse set of 20 chest computed tomography(CT) scans of subjects ranging from healthy volunteers to patientswith severe pathologies, scanned at different sites, with differentCT scanner brands, models, and scanning protocols. Three performancemeasures covering different aspects of segmentationquality were computed for all participating algorithms. Resultsfrom the evaluation showed that no single algorithm could extractmore than an average of 74% of the total length of all branches inthe reference standard, indicating substantial differences between the algorithms. A fusion scheme that obtained superior results ispresented, demonstrating that there is complementary informationprovided by the different algorithms and there is still roomfor further improvements in airway segmentation algorithms.
AB - This paper describes a framework for establishing areference airway tree segmentation, which was used to quantitativelyevaluate 15 different airway tree extraction algorithms ina standardized manner. Because of the sheer difficulty involvedin manually constructing a complete reference standard fromscratch, we propose to construct the reference using results fromall algorithms that are to be evaluated. We start by subdividingeach segmented airway tree into its individual branch segments.Each branch segment is then visually scored by trained observersto determine whether or not it is a correctly segmented part of theairway tree. Finally, the reference airway trees are constructed bytaking the union of all correctly extracted branch segments. Fifteenairway tree extraction algorithms from different research groupsare evaluated on a diverse set of 20 chest computed tomography(CT) scans of subjects ranging from healthy volunteers to patientswith severe pathologies, scanned at different sites, with differentCT scanner brands, models, and scanning protocols. Three performancemeasures covering different aspects of segmentationquality were computed for all participating algorithms. Resultsfrom the evaluation showed that no single algorithm could extractmore than an average of 74% of the total length of all branches inthe reference standard, indicating substantial differences between the algorithms. A fusion scheme that obtained superior results ispresented, demonstrating that there is complementary informationprovided by the different algorithms and there is still roomfor further improvements in airway segmentation algorithms.
U2 - 10.1109/TMI.2012.2209674
DO - 10.1109/TMI.2012.2209674
M3 - Journal article
C2 - 22855226
VL - 31
SP - 2093
EP - 2107
JO - I E E E Transactions on Medical Imaging
JF - I E E E Transactions on Medical Imaging
SN - 0278-0062
IS - 11
ER -
ID: 38290021