Asymmetric similarity-weighted ensembles for image segmentation
Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review
Standard
Asymmetric similarity-weighted ensembles for image segmentation. / Cheplygina, V.; Van Opbroek, A.; Ikram, M. A.; Vernooij, M. W.; de Bruijne, Marleen.
2016 IEEE International Symposium on Biomedical Imaging: from Nano to Macro. IEEE, 2016. p. 273-277.Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - GEN
T1 - Asymmetric similarity-weighted ensembles for image segmentation
AU - Cheplygina, V.
AU - Van Opbroek, A.
AU - Ikram, M. A.
AU - Vernooij, M. W.
AU - de Bruijne, Marleen
PY - 2016
Y1 - 2016
N2 - Supervised classification is widely used for image segmentation. To work effectively, these techniques need large amounts of labeled training data, that is representative of the test data. Different patient groups, different scanners or different scanning protocols can lead to differences between the images, thus representative data might not be available. Transfer learning techniques can be used to account for these differences, thus taking advantage of all the available data acquired with different protocols. We investigate the use of classifier ensembles, where each classifier is weighted according to the similarity between the data it is trained on, and the data it needs to segment. We examine 3 asymmetric similarity measures that can be used in scenarios where no labeled data from a newly introduced scanner or scanning protocol is available. We show that the asymmetry is informative and the direction of measurement needs to be chosen carefully. We also show that a point set similarity measure is robust across different studies, and outperforms state-of-the-art results on a multi-center brain tissue segmentation task.
AB - Supervised classification is widely used for image segmentation. To work effectively, these techniques need large amounts of labeled training data, that is representative of the test data. Different patient groups, different scanners or different scanning protocols can lead to differences between the images, thus representative data might not be available. Transfer learning techniques can be used to account for these differences, thus taking advantage of all the available data acquired with different protocols. We investigate the use of classifier ensembles, where each classifier is weighted according to the similarity between the data it is trained on, and the data it needs to segment. We examine 3 asymmetric similarity measures that can be used in scenarios where no labeled data from a newly introduced scanner or scanning protocol is available. We show that the asymmetry is informative and the direction of measurement needs to be chosen carefully. We also show that a point set similarity measure is robust across different studies, and outperforms state-of-the-art results on a multi-center brain tissue segmentation task.
KW - asymmetry
KW - similarity measure
KW - tissue segmentation
KW - Transfer learning
U2 - 10.1109/ISBI.2016.7493262
DO - 10.1109/ISBI.2016.7493262
M3 - Article in proceedings
AN - SCOPUS:84978434492
SP - 273
EP - 277
BT - 2016 IEEE International Symposium on Biomedical Imaging
PB - IEEE
T2 - 2016 IEEE 13th International Symposium on Biomedical Imaging: From Nano to Macro, ISBI 2016
Y2 - 13 April 2016 through 16 April 2016
ER -
ID: 167101667