Random Forest-Based Manifold Learning for Classification of Imaging Data in Dementia

Katherine R. Gray; Paul Aljabar; Rolf A. Heckemann; Alexander Hammers; Daniel Rueckert

Random Forest-Based Manifold Learning for Classification of Imaging Data in Dementia

Katherine R. Gray
, Paul Aljabar
, Rolf A. Heckemann
, Alexander Hammers
, Daniel Rueckert

Biomedical Engineering Department

Imperial College London

Research output: Chapter in Book/Report/Conference proceeding › Conference paper

23 Citations (Scopus)

Abstract

Neurodegenerative disorders are characterized by changes in multiple biomarkers, which may provide complementary information for diagnosis and prognosis. We present a framework in which proximities derived from random forests are used to learn a low-dimensional manifold from labelled training data and then to infer the clinical labels of test data mapped to this space. The proposed method facilitates the combination of embeddings from multiple datasets, resulting in the generation of a joint embedding that simultaneously encodes information about all the available features. It is possible to combine different types of data without additional processing, and we demonstrate this key feature by application to voxel-based FDG-PET and region-based MR imaging data from the ADNI study. Classification based on the joint embedding coordinates out-performs classification based on either modality alone. Results are impressive compared with other state-of-the-art machine learning techniques applied to multi-modality imaging data.

Original language	English
Title of host publication	Machine learning in medical imaging
Editors	K Suzuki, F Wang, DG Shen, PK Yan
Place of Publication	BERLIN
Publisher	Springer
Pages	159-166
Number of pages	8
Volume	7009 LNCS
ISBN (Print)	9783642243189
Publication status	Published - 2011
Event	2nd International Workshop on Machine Learning in Medical Imaging (MLMI 2011) - Toronto Duration: 18 Sept 2011 → …

Conference

Conference	2nd International Workshop on Machine Learning in Medical Imaging (MLMI 2011)
City	Toronto
Period	18/09/2011 → …

Cite this

@inbook{eb42ea82792d4d32b2689a4a01489670,

title = "Random Forest-Based Manifold Learning for Classification of Imaging Data in Dementia",

abstract = "Neurodegenerative disorders are characterized by changes in multiple biomarkers, which may provide complementary information for diagnosis and prognosis. We present a framework in which proximities derived from random forests are used to learn a low-dimensional manifold from labelled training data and then to infer the clinical labels of test data mapped to this space. The proposed method facilitates the combination of embeddings from multiple datasets, resulting in the generation of a joint embedding that simultaneously encodes information about all the available features. It is possible to combine different types of data without additional processing, and we demonstrate this key feature by application to voxel-based FDG-PET and region-based MR imaging data from the ADNI study. Classification based on the joint embedding coordinates out-performs classification based on either modality alone. Results are impressive compared with other state-of-the-art machine learning techniques applied to multi-modality imaging data.",

author = "Gray, \{Katherine R.\} and Paul Aljabar and Heckemann, \{Rolf A.\} and Alexander Hammers and Daniel Rueckert",

year = "2011",

language = "English",

isbn = "9783642243189 ",

volume = "7009 LNCS",

pages = "159--166",

editor = "K Suzuki and F Wang and DG Shen and PK Yan",

booktitle = "Machine learning in medical imaging",

publisher = "Springer",

note = "2nd International Workshop on Machine Learning in Medical Imaging (MLMI 2011) ; Conference date: 18-09-2011",

}

Gray, KR, Aljabar, P, Heckemann, RA, Hammers, A & Rueckert, D 2011, Random Forest-Based Manifold Learning for Classification of Imaging Data in Dementia. in K Suzuki, F Wang, DG Shen & PK Yan (eds), Machine learning in medical imaging. vol. 7009 LNCS, Springer, BERLIN, pp. 159-166, 2nd International Workshop on Machine Learning in Medical Imaging (MLMI 2011), Toronto, 18/09/2011.

TY - CHAP

T1 - Random Forest-Based Manifold Learning for Classification of Imaging Data in Dementia

AU - Gray, Katherine R.

AU - Aljabar, Paul

AU - Heckemann, Rolf A.

AU - Hammers, Alexander

AU - Rueckert, Daniel

PY - 2011

Y1 - 2011

N2 - Neurodegenerative disorders are characterized by changes in multiple biomarkers, which may provide complementary information for diagnosis and prognosis. We present a framework in which proximities derived from random forests are used to learn a low-dimensional manifold from labelled training data and then to infer the clinical labels of test data mapped to this space. The proposed method facilitates the combination of embeddings from multiple datasets, resulting in the generation of a joint embedding that simultaneously encodes information about all the available features. It is possible to combine different types of data without additional processing, and we demonstrate this key feature by application to voxel-based FDG-PET and region-based MR imaging data from the ADNI study. Classification based on the joint embedding coordinates out-performs classification based on either modality alone. Results are impressive compared with other state-of-the-art machine learning techniques applied to multi-modality imaging data.

AB - Neurodegenerative disorders are characterized by changes in multiple biomarkers, which may provide complementary information for diagnosis and prognosis. We present a framework in which proximities derived from random forests are used to learn a low-dimensional manifold from labelled training data and then to infer the clinical labels of test data mapped to this space. The proposed method facilitates the combination of embeddings from multiple datasets, resulting in the generation of a joint embedding that simultaneously encodes information about all the available features. It is possible to combine different types of data without additional processing, and we demonstrate this key feature by application to voxel-based FDG-PET and region-based MR imaging data from the ADNI study. Classification based on the joint embedding coordinates out-performs classification based on either modality alone. Results are impressive compared with other state-of-the-art machine learning techniques applied to multi-modality imaging data.

M3 - Conference paper

SN - 9783642243189

VL - 7009 LNCS

SP - 159

EP - 166

BT - Machine learning in medical imaging

A2 - Suzuki, K

A2 - Wang, F

A2 - Shen, DG

A2 - Yan, PK

PB - Springer

CY - BERLIN

T2 - 2nd International Workshop on Machine Learning in Medical Imaging (MLMI 2011)

Y2 - 18 September 2011

ER -

Random Forest-Based Manifold Learning for Classification of Imaging Data in Dementia

Abstract

Conference

Fingerprint

Cite this