Block-based statistics for robust non-parametric morphometry

Geng Chen, Pei Zhang, Ke Li, Chong Yaw Wee, Yafeng Wu, Dinggang Shen, Pew Thian Yap

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Citations (Scopus)

Abstract

Automated algorithms designed for comparison of medical images are generally dependent on a sufficiently large dataset and highly accurate registration as they implicitly assume that the comparison is being made across a set of images with locally matching structures. However, very often sample size is limited and registration methods are not perfect and may be prone to errors due to noise, artifacts, and complex variations of brain topology. In this paper, we propose a novel statistical group comparison algorithm, called block-based statistics (BBS), which reformulates the conventional comparison framework from a non-local means perspective in order to learn what the statistics would have been, given perfect correspondence. Through this formulation, BBS (1) explicitly considers image registration errors to reduce reliance on high-quality registrations, (2) increases the number of samples for statistical estimation by collapsing measurements from similar signal distributions, and (3) diminishes the need for large image sets. BBS is based on permutation test and hence no assumption, such as Gaussianity, is imposed on the distribution. Experimental results indicate that BBS yields markedly improved lesion detection accuracy especially with limited sample size, is more robust to sample imbalance, and converges faster to results expected for large sample size.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
PublisherSpringer Verlag
Pages62-70
Number of pages9
Volume9467
ISBN (Print)9783319281933
DOIs
Publication statusPublished - 2015
Externally publishedYes
Event1st International Workshop on Patch-Based Techniques in Medical Imaging, Patch-MI 2015 - Munich, Germany
Duration: 2015 Oct 92015 Oct 9

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9467
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other1st International Workshop on Patch-Based Techniques in Medical Imaging, Patch-MI 2015
CountryGermany
CityMunich
Period15/10/915/10/9

Fingerprint

Morphometry
Statistics
Registration
Sample Size
Block Algorithm
Permutation Test
Statistical Estimation
Collapsing
Image registration
Image Registration
Medical Image
Large Data Sets
Brain
Correspondence
Topology
Converge
Formulation
Dependent
Experimental Results

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Chen, G., Zhang, P., Li, K., Wee, C. Y., Wu, Y., Shen, D., & Yap, P. T. (2015). Block-based statistics for robust non-parametric morphometry. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 9467, pp. 62-70). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9467). Springer Verlag. https://doi.org/10.1007/978-3-319-28194-0_8

Block-based statistics for robust non-parametric morphometry. / Chen, Geng; Zhang, Pei; Li, Ke; Wee, Chong Yaw; Wu, Yafeng; Shen, Dinggang; Yap, Pew Thian.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9467 Springer Verlag, 2015. p. 62-70 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9467).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Chen, G, Zhang, P, Li, K, Wee, CY, Wu, Y, Shen, D & Yap, PT 2015, Block-based statistics for robust non-parametric morphometry. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 9467, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9467, Springer Verlag, pp. 62-70, 1st International Workshop on Patch-Based Techniques in Medical Imaging, Patch-MI 2015, Munich, Germany, 15/10/9. https://doi.org/10.1007/978-3-319-28194-0_8
Chen G, Zhang P, Li K, Wee CY, Wu Y, Shen D et al. Block-based statistics for robust non-parametric morphometry. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9467. Springer Verlag. 2015. p. 62-70. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-28194-0_8
Chen, Geng ; Zhang, Pei ; Li, Ke ; Wee, Chong Yaw ; Wu, Yafeng ; Shen, Dinggang ; Yap, Pew Thian. / Block-based statistics for robust non-parametric morphometry. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9467 Springer Verlag, 2015. pp. 62-70 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{e11bd1ca2dcd4142942293e4d78de22b,
title = "Block-based statistics for robust non-parametric morphometry",
abstract = "Automated algorithms designed for comparison of medical images are generally dependent on a sufficiently large dataset and highly accurate registration as they implicitly assume that the comparison is being made across a set of images with locally matching structures. However, very often sample size is limited and registration methods are not perfect and may be prone to errors due to noise, artifacts, and complex variations of brain topology. In this paper, we propose a novel statistical group comparison algorithm, called block-based statistics (BBS), which reformulates the conventional comparison framework from a non-local means perspective in order to learn what the statistics would have been, given perfect correspondence. Through this formulation, BBS (1) explicitly considers image registration errors to reduce reliance on high-quality registrations, (2) increases the number of samples for statistical estimation by collapsing measurements from similar signal distributions, and (3) diminishes the need for large image sets. BBS is based on permutation test and hence no assumption, such as Gaussianity, is imposed on the distribution. Experimental results indicate that BBS yields markedly improved lesion detection accuracy especially with limited sample size, is more robust to sample imbalance, and converges faster to results expected for large sample size.",
author = "Geng Chen and Pei Zhang and Ke Li and Wee, {Chong Yaw} and Yafeng Wu and Dinggang Shen and Yap, {Pew Thian}",
year = "2015",
doi = "10.1007/978-3-319-28194-0_8",
language = "English",
isbn = "9783319281933",
volume = "9467",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "62--70",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Block-based statistics for robust non-parametric morphometry

AU - Chen, Geng

AU - Zhang, Pei

AU - Li, Ke

AU - Wee, Chong Yaw

AU - Wu, Yafeng

AU - Shen, Dinggang

AU - Yap, Pew Thian

PY - 2015

Y1 - 2015

N2 - Automated algorithms designed for comparison of medical images are generally dependent on a sufficiently large dataset and highly accurate registration as they implicitly assume that the comparison is being made across a set of images with locally matching structures. However, very often sample size is limited and registration methods are not perfect and may be prone to errors due to noise, artifacts, and complex variations of brain topology. In this paper, we propose a novel statistical group comparison algorithm, called block-based statistics (BBS), which reformulates the conventional comparison framework from a non-local means perspective in order to learn what the statistics would have been, given perfect correspondence. Through this formulation, BBS (1) explicitly considers image registration errors to reduce reliance on high-quality registrations, (2) increases the number of samples for statistical estimation by collapsing measurements from similar signal distributions, and (3) diminishes the need for large image sets. BBS is based on permutation test and hence no assumption, such as Gaussianity, is imposed on the distribution. Experimental results indicate that BBS yields markedly improved lesion detection accuracy especially with limited sample size, is more robust to sample imbalance, and converges faster to results expected for large sample size.

AB - Automated algorithms designed for comparison of medical images are generally dependent on a sufficiently large dataset and highly accurate registration as they implicitly assume that the comparison is being made across a set of images with locally matching structures. However, very often sample size is limited and registration methods are not perfect and may be prone to errors due to noise, artifacts, and complex variations of brain topology. In this paper, we propose a novel statistical group comparison algorithm, called block-based statistics (BBS), which reformulates the conventional comparison framework from a non-local means perspective in order to learn what the statistics would have been, given perfect correspondence. Through this formulation, BBS (1) explicitly considers image registration errors to reduce reliance on high-quality registrations, (2) increases the number of samples for statistical estimation by collapsing measurements from similar signal distributions, and (3) diminishes the need for large image sets. BBS is based on permutation test and hence no assumption, such as Gaussianity, is imposed on the distribution. Experimental results indicate that BBS yields markedly improved lesion detection accuracy especially with limited sample size, is more robust to sample imbalance, and converges faster to results expected for large sample size.

UR - http://www.scopus.com/inward/record.url?scp=84955287169&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84955287169&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-28194-0_8

DO - 10.1007/978-3-319-28194-0_8

M3 - Conference contribution

AN - SCOPUS:84955287169

SN - 9783319281933

VL - 9467

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 62

EP - 70

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

PB - Springer Verlag

ER -