Machine learning on high dimensional shape data from subcortical brain surfaces: A comparison of feature selection and classification methods

Benjamin S.C. Wade; Shantanu H. Joshi; Boris A. Gutman; Paul M. Thompson

doi:10.1007/978-3-319-24888-2_5

Machine learning on high dimensional shape data from subcortical brain surfaces: A comparison of feature selection and classification methods

Benjamin S.C. Wade
, Shantanu H. Joshi
, Boris A. Gutman
, Paul M. Thompson

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

3 Scopus citations

Abstract

Recently, high-dimensional shape data (HDSD) has been demonstrated to be informative in describing subcortical brain morphometry in several disorders. While HDSD may serve as a biomarker of disease, its high dimensionality may require careful treatment in its application to machine learning. Here, we compare several possible approaches for feature selection and pattern classification using HDSD. We explore the efficacy of three candidate feature selection (FS) methods: Guided Random Forest (GRF), LASSO and no feature selection (NFS). Each feature set was applied to three classifiers: Random Forest (RF), Support Vector Machines (SVM) and Naïve Bayes (NB). Each model was cross-validated using two diagnostic contrasts: Alzheimer’s Disease and mild cognitive impairment; each relative to matched controls. GRF and NFS outperformed LASSO as FS methods and were comparably competitive. NB underperformed relative to RF and SVM, which were comparable in performance. Our results advocate the NFS-RF approach for its speed, simplicity and interpretability.

Original language	English
Title of host publication	Machine Learning in Medical Imaging - 6th International Workshop, MLMI 2015 Held in Conjunction with MICCAI 2015, Proceedings
Editors	Luping Zhou, Yinghuan Shi, Li Wang, Qian Wang
Publisher	Springer Verlag
Pages	36-43
Number of pages	8
ISBN (Print)	9783319248875
DOIs	https://doi.org/10.1007/978-3-319-24888-2_5
State	Published - 2015
Externally published	Yes
Event	6th International Workshop on Machine Learning in Medical Imaging, MLMI 2015 and Held in Conjunction with 18th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2015 - Munich, Germany Duration: 5 Oct 2015 → 5 Oct 2015

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	9352
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	6th International Workshop on Machine Learning in Medical Imaging, MLMI 2015 and Held in Conjunction with 18th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2015
Country/Territory	Germany
City	Munich
Period	5/10/15 → 5/10/15

Access to Document

10.1007/978-3-319-24888-2_5

Cite this

Wade, B. S. C., Joshi, S. H., Gutman, B. A., & Thompson, P. M. (2015). Machine learning on high dimensional shape data from subcortical brain surfaces: A comparison of feature selection and classification methods. In L. Zhou, Y. Shi, L. Wang, & Q. Wang (Eds.), Machine Learning in Medical Imaging - 6th International Workshop, MLMI 2015 Held in Conjunction with MICCAI 2015, Proceedings (pp. 36-43). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9352). Springer Verlag. https://doi.org/10.1007/978-3-319-24888-2_5

Wade, Benjamin S.C. ; Joshi, Shantanu H. ; Gutman, Boris A. et al. / Machine learning on high dimensional shape data from subcortical brain surfaces : A comparison of feature selection and classification methods. Machine Learning in Medical Imaging - 6th International Workshop, MLMI 2015 Held in Conjunction with MICCAI 2015, Proceedings. editor / Luping Zhou ; Yinghuan Shi ; Li Wang ; Qian Wang. Springer Verlag, 2015. pp. 36-43 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{d513afe73deb4327ad14f90e662645d5,

title = "Machine learning on high dimensional shape data from subcortical brain surfaces: A comparison of feature selection and classification methods",

abstract = "Recently, high-dimensional shape data (HDSD) has been demonstrated to be informative in describing subcortical brain morphometry in several disorders. While HDSD may serve as a biomarker of disease, its high dimensionality may require careful treatment in its application to machine learning. Here, we compare several possible approaches for feature selection and pattern classification using HDSD. We explore the efficacy of three candidate feature selection (FS) methods: Guided Random Forest (GRF), LASSO and no feature selection (NFS). Each feature set was applied to three classifiers: Random Forest (RF), Support Vector Machines (SVM) and Na{\"i}ve Bayes (NB). Each model was cross-validated using two diagnostic contrasts: Alzheimer{\textquoteright}s Disease and mild cognitive impairment; each relative to matched controls. GRF and NFS outperformed LASSO as FS methods and were comparably competitive. NB underperformed relative to RF and SVM, which were comparable in performance. Our results advocate the NFS-RF approach for its speed, simplicity and interpretability.",

author = "Wade, \{Benjamin S.C.\} and Joshi, \{Shantanu H.\} and Gutman, \{Boris A.\} and Thompson, \{Paul M.\}",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing Switzerland 2015.; 6th International Workshop on Machine Learning in Medical Imaging, MLMI 2015 and Held in Conjunction with 18th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2015 ; Conference date: 05-10-2015 Through 05-10-2015",

year = "2015",

doi = "10.1007/978-3-319-24888-2\_5",

language = "English",

isbn = "9783319248875",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "36--43",

editor = "Luping Zhou and Yinghuan Shi and Li Wang and Qian Wang",

booktitle = "Machine Learning in Medical Imaging - 6th International Workshop, MLMI 2015 Held in Conjunction with MICCAI 2015, Proceedings",

address = "Germany",

}

Wade, BSC, Joshi, SH, Gutman, BA & Thompson, PM 2015, Machine learning on high dimensional shape data from subcortical brain surfaces: A comparison of feature selection and classification methods. in L Zhou, Y Shi, L Wang & Q Wang (eds), Machine Learning in Medical Imaging - 6th International Workshop, MLMI 2015 Held in Conjunction with MICCAI 2015, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9352, Springer Verlag, pp. 36-43, 6th International Workshop on Machine Learning in Medical Imaging, MLMI 2015 and Held in Conjunction with 18th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2015, Munich, Germany, 5/10/15. https://doi.org/10.1007/978-3-319-24888-2_5

Machine learning on high dimensional shape data from subcortical brain surfaces: A comparison of feature selection and classification methods. / Wade, Benjamin S.C.; Joshi, Shantanu H.; Gutman, Boris A. et al.
Machine Learning in Medical Imaging - 6th International Workshop, MLMI 2015 Held in Conjunction with MICCAI 2015, Proceedings. ed. / Luping Zhou; Yinghuan Shi; Li Wang; Qian Wang. Springer Verlag, 2015. p. 36-43 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9352).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Machine learning on high dimensional shape data from subcortical brain surfaces

T2 - 6th International Workshop on Machine Learning in Medical Imaging, MLMI 2015 and Held in Conjunction with 18th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2015

AU - Wade, Benjamin S.C.

AU - Joshi, Shantanu H.

AU - Gutman, Boris A.

AU - Thompson, Paul M.

N1 - Publisher Copyright: © Springer International Publishing Switzerland 2015.

PY - 2015

Y1 - 2015

N2 - Recently, high-dimensional shape data (HDSD) has been demonstrated to be informative in describing subcortical brain morphometry in several disorders. While HDSD may serve as a biomarker of disease, its high dimensionality may require careful treatment in its application to machine learning. Here, we compare several possible approaches for feature selection and pattern classification using HDSD. We explore the efficacy of three candidate feature selection (FS) methods: Guided Random Forest (GRF), LASSO and no feature selection (NFS). Each feature set was applied to three classifiers: Random Forest (RF), Support Vector Machines (SVM) and Naïve Bayes (NB). Each model was cross-validated using two diagnostic contrasts: Alzheimer’s Disease and mild cognitive impairment; each relative to matched controls. GRF and NFS outperformed LASSO as FS methods and were comparably competitive. NB underperformed relative to RF and SVM, which were comparable in performance. Our results advocate the NFS-RF approach for its speed, simplicity and interpretability.

AB - Recently, high-dimensional shape data (HDSD) has been demonstrated to be informative in describing subcortical brain morphometry in several disorders. While HDSD may serve as a biomarker of disease, its high dimensionality may require careful treatment in its application to machine learning. Here, we compare several possible approaches for feature selection and pattern classification using HDSD. We explore the efficacy of three candidate feature selection (FS) methods: Guided Random Forest (GRF), LASSO and no feature selection (NFS). Each feature set was applied to three classifiers: Random Forest (RF), Support Vector Machines (SVM) and Naïve Bayes (NB). Each model was cross-validated using two diagnostic contrasts: Alzheimer’s Disease and mild cognitive impairment; each relative to matched controls. GRF and NFS outperformed LASSO as FS methods and were comparably competitive. NB underperformed relative to RF and SVM, which were comparable in performance. Our results advocate the NFS-RF approach for its speed, simplicity and interpretability.

UR - https://www.scopus.com/pages/publications/84951948808

U2 - 10.1007/978-3-319-24888-2_5

DO - 10.1007/978-3-319-24888-2_5

M3 - Conference contribution

AN - SCOPUS:84951948808

SN - 9783319248875

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

SP - 36

EP - 43

BT - Machine Learning in Medical Imaging - 6th International Workshop, MLMI 2015 Held in Conjunction with MICCAI 2015, Proceedings

A2 - Zhou, Luping

A2 - Shi, Yinghuan

A2 - Wang, Li

A2 - Wang, Qian

PB - Springer Verlag

Y2 - 5 October 2015 through 5 October 2015

ER -

Wade BSC, Joshi SH, Gutman BA, Thompson PM. Machine learning on high dimensional shape data from subcortical brain surfaces: A comparison of feature selection and classification methods. In Zhou L, Shi Y, Wang L, Wang Q, editors, Machine Learning in Medical Imaging - 6th International Workshop, MLMI 2015 Held in Conjunction with MICCAI 2015, Proceedings. Springer Verlag. 2015. p. 36-43. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-24888-2_5

Machine learning on high dimensional shape data from subcortical brain surfaces: A comparison of feature selection and classification methods

Abstract

Publication series

Conference

Access to Document

Fingerprint

Cite this