Using deep Siamese neural networks for detection of brain asymmetries associated with Alzheimer's Disease and Mild Cognitive Impairment

the BIOCARD Research Team, for the Alzheimer's Disease Neuroimaging Initiative

Research output: Contribution to journalArticle

Abstract

In recent studies, neuroanatomical volume and shape asymmetries have been seen during the course of Alzheimer's Disease (AD) and could potentially be used as preclinical imaging biomarkers for the prediction of Mild Cognitive Impairment (MCI) and AD dementia. In this study, a deep learning framework utilizing Siamese neural networks trained on paired lateral inter-hemispheric regions is used to harness the discriminative power of whole-brain volumetric asymmetry. The method uses the MRICloud pipeline to yield low-dimensional volumetric features of pre-defined atlas brain structures, and a novel non-linear kernel trick to normalize these features to reduce batch effects across datasets and populations. By working with the low-dimensional features, Siamese networks were shown to yield comparable performance to studies that utilize whole-brain MR images, with the advantage of reduced complexity and computational time, while preserving the biological information density. Experimental results also show that Siamese networks perform better in certain metrics by explicitly encoding the asymmetry in brain volumes, compared to traditional prediction methods that do not use the asymmetry, on the ADNI and BIOCARD datasets.

Original languageEnglish (US)
JournalMagnetic Resonance Imaging
DOIs
StateAccepted/In press - Jan 1 2019

Fingerprint

Brain
Alzheimer Disease
Atlases
Biomarkers
Pipelines
Learning
Neural networks
Imaging techniques
Cognitive Dysfunction
Deep neural networks
Population
Datasets

Keywords

  • Alzheimer's Disease
  • Deep learning
  • Machine learning
  • Mild Cognitive Impairment
  • Siamese networks
  • Structural magnetic resonance imaging

ASJC Scopus subject areas

  • Biophysics
  • Biomedical Engineering
  • Radiology Nuclear Medicine and imaging

Cite this

Using deep Siamese neural networks for detection of brain asymmetries associated with Alzheimer's Disease and Mild Cognitive Impairment. / the BIOCARD Research Team; for the Alzheimer's Disease Neuroimaging Initiative.

In: Magnetic Resonance Imaging, 01.01.2019.

Research output: Contribution to journalArticle

@article{a8206f1ea9a745949d9c1d7d58bff533,
title = "Using deep Siamese neural networks for detection of brain asymmetries associated with Alzheimer's Disease and Mild Cognitive Impairment",
abstract = "In recent studies, neuroanatomical volume and shape asymmetries have been seen during the course of Alzheimer's Disease (AD) and could potentially be used as preclinical imaging biomarkers for the prediction of Mild Cognitive Impairment (MCI) and AD dementia. In this study, a deep learning framework utilizing Siamese neural networks trained on paired lateral inter-hemispheric regions is used to harness the discriminative power of whole-brain volumetric asymmetry. The method uses the MRICloud pipeline to yield low-dimensional volumetric features of pre-defined atlas brain structures, and a novel non-linear kernel trick to normalize these features to reduce batch effects across datasets and populations. By working with the low-dimensional features, Siamese networks were shown to yield comparable performance to studies that utilize whole-brain MR images, with the advantage of reduced complexity and computational time, while preserving the biological information density. Experimental results also show that Siamese networks perform better in certain metrics by explicitly encoding the asymmetry in brain volumes, compared to traditional prediction methods that do not use the asymmetry, on the ADNI and BIOCARD datasets.",
keywords = "Alzheimer's Disease, Deep learning, Machine learning, Mild Cognitive Impairment, Siamese networks, Structural magnetic resonance imaging",
author = "{the BIOCARD Research Team} and {for the Alzheimer's Disease Neuroimaging Initiative} and Liu, {Chin Fu} and Shreyas Padhy and Sandhya Ramachandran and Wang, {Victor X.} and Andrew Efimov and Alonso Bernal and Linyuan Shi and Marc Vaillant and Ratnanather, {J. Tilak} and Faria, {Andreia Vasconcellos} and Caffo, {Brian S} and Marilyn Albert and Miller, {Michael I.}",
year = "2019",
month = "1",
day = "1",
doi = "10.1016/j.mri.2019.07.003",
language = "English (US)",
journal = "Magnetic Resonance Imaging",
issn = "0730-725X",
publisher = "Elsevier Inc.",

}

TY - JOUR

T1 - Using deep Siamese neural networks for detection of brain asymmetries associated with Alzheimer's Disease and Mild Cognitive Impairment

AU - the BIOCARD Research Team

AU - for the Alzheimer's Disease Neuroimaging Initiative

AU - Liu, Chin Fu

AU - Padhy, Shreyas

AU - Ramachandran, Sandhya

AU - Wang, Victor X.

AU - Efimov, Andrew

AU - Bernal, Alonso

AU - Shi, Linyuan

AU - Vaillant, Marc

AU - Ratnanather, J. Tilak

AU - Faria, Andreia Vasconcellos

AU - Caffo, Brian S

AU - Albert, Marilyn

AU - Miller, Michael I.

PY - 2019/1/1

Y1 - 2019/1/1

N2 - In recent studies, neuroanatomical volume and shape asymmetries have been seen during the course of Alzheimer's Disease (AD) and could potentially be used as preclinical imaging biomarkers for the prediction of Mild Cognitive Impairment (MCI) and AD dementia. In this study, a deep learning framework utilizing Siamese neural networks trained on paired lateral inter-hemispheric regions is used to harness the discriminative power of whole-brain volumetric asymmetry. The method uses the MRICloud pipeline to yield low-dimensional volumetric features of pre-defined atlas brain structures, and a novel non-linear kernel trick to normalize these features to reduce batch effects across datasets and populations. By working with the low-dimensional features, Siamese networks were shown to yield comparable performance to studies that utilize whole-brain MR images, with the advantage of reduced complexity and computational time, while preserving the biological information density. Experimental results also show that Siamese networks perform better in certain metrics by explicitly encoding the asymmetry in brain volumes, compared to traditional prediction methods that do not use the asymmetry, on the ADNI and BIOCARD datasets.

AB - In recent studies, neuroanatomical volume and shape asymmetries have been seen during the course of Alzheimer's Disease (AD) and could potentially be used as preclinical imaging biomarkers for the prediction of Mild Cognitive Impairment (MCI) and AD dementia. In this study, a deep learning framework utilizing Siamese neural networks trained on paired lateral inter-hemispheric regions is used to harness the discriminative power of whole-brain volumetric asymmetry. The method uses the MRICloud pipeline to yield low-dimensional volumetric features of pre-defined atlas brain structures, and a novel non-linear kernel trick to normalize these features to reduce batch effects across datasets and populations. By working with the low-dimensional features, Siamese networks were shown to yield comparable performance to studies that utilize whole-brain MR images, with the advantage of reduced complexity and computational time, while preserving the biological information density. Experimental results also show that Siamese networks perform better in certain metrics by explicitly encoding the asymmetry in brain volumes, compared to traditional prediction methods that do not use the asymmetry, on the ADNI and BIOCARD datasets.

KW - Alzheimer's Disease

KW - Deep learning

KW - Machine learning

KW - Mild Cognitive Impairment

KW - Siamese networks

KW - Structural magnetic resonance imaging

UR - http://www.scopus.com/inward/record.url?scp=85071137520&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85071137520&partnerID=8YFLogxK

U2 - 10.1016/j.mri.2019.07.003

DO - 10.1016/j.mri.2019.07.003

M3 - Article

C2 - 31319126

AN - SCOPUS:85071137520

JO - Magnetic Resonance Imaging

JF - Magnetic Resonance Imaging

SN - 0730-725X

ER -