Patch-Based Brain Age Estimation from MR Images

Kyriaki Margarita Bintsi; Vasileios Baltatzis; Arinbjörn Kolbeinsson; Alexander Hammers; Daniel Rueckert

doi:10.1007/978-3-030-66843-3_10

Patch-Based Brain Age Estimation from MR Images

Kyriaki Margarita Bintsi^*, Vasileios Baltatzis, Arinbjörn Kolbeinsson, Alexander Hammers, Daniel Rueckert

^*Corresponding author for this work

PET Imaging Centre Facility

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

9 Citations (Scopus)

Abstract

Brain age estimation from Magnetic Resonance Images (MRI) derives the difference between a subject’s biological brain age and their chronological age. This is a potential biomarker for neurodegeneration, e.g. as part of Alzheimer’s disease. Early detection of neurodegeneration manifesting as a higher brain age can potentially facilitate better medical care and planning for affected individuals. Many studies have been proposed for the prediction of chronological age from brain MRI using machine learning and specifically deep learning techniques. Contrary to most studies, which use the whole brain volume, in this study, we develop a new deep learning approach that uses 3D patches of the brain as well as convolutional neural networks (CNNs) to develop a localised brain age estimator. In this way, we can obtain a visualization of the regions that play the most important role for estimating brain age, leading to more anatomically driven and interpretable results, and thus confirming relevant literature which suggests that the ventricles and the hippocampus are the areas that are most informative. In addition, we leverage this knowledge in order to improve the overall performance on the task of age estimation by combining the results of different patches using an ensemble method, such as averaging or linear regression. The network is trained on the UK Biobank dataset and the method achieves state-of-the-art results with a Mean Absolute Error of 2.46 years for purely regional estimates, and 2.13 years for an ensemble of patches before bias correction, while 1.96 years after bias correction.

Original language	English
Title of host publication	Machine Learning in Clinical Neuroimaging and Radiogenomics in Neuro-oncology - 3rd International Workshop, MLCN 2020, and Second International Workshop, RNO-AI 2020, Held in Conjunction with MICCAI 2020, Proceedings
Editors	Seyed Mostafa Kia, Hassan Mohy-ud-Din, Ahmed Abdulkadir, Cher Bass, Mohamad Habes, Jane Maryam Rondina, Chantal Tax, Hongzhi Wang, Thomas Wolfers, Saima Rathore, Madhura Ingalhalikar
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	98-107
Number of pages	10
ISBN (Print)	9783030668426
DOIs	https://doi.org/10.1007/978-3-030-66843-3_10
Publication status	Published - 2020
Event	3rd International Workshop on Machine Learning in Clinical Neuroimaging, MLCN 2020, and 2nd International Workshop on Radiogenomics in Neuro-oncology, RNO-AI 2020, held in conjunction with MICCAI 2020 - Lima, Peru Duration: 4 Oct 2020 → 8 Oct 2020

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	12449 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	3rd International Workshop on Machine Learning in Clinical Neuroimaging, MLCN 2020, and 2nd International Workshop on Radiogenomics in Neuro-oncology, RNO-AI 2020, held in conjunction with MICCAI 2020
Country/Territory	Peru
City	Lima
Period	4/10/2020 → 8/10/2020

Keywords

Brain age estimation
Deep learning
Localization
MR Images

Access to Document

10.1007/978-3-030-66843-3_10

Cite this

Bintsi, K. M., Baltatzis, V., Kolbeinsson, A., Hammers, A., & Rueckert, D. (2020). Patch-Based Brain Age Estimation from MR Images. In S. M. Kia, H. Mohy-ud-Din, A. Abdulkadir, C. Bass, M. Habes, J. M. Rondina, C. Tax, H. Wang, T. Wolfers, S. Rathore, & M. Ingalhalikar (Eds.), Machine Learning in Clinical Neuroimaging and Radiogenomics in Neuro-oncology - 3rd International Workshop, MLCN 2020, and Second International Workshop, RNO-AI 2020, Held in Conjunction with MICCAI 2020, Proceedings (pp. 98-107). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12449 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-030-66843-3_10

Bintsi, Kyriaki Margarita ; Baltatzis, Vasileios ; Kolbeinsson, Arinbjörn et al. / Patch-Based Brain Age Estimation from MR Images. Machine Learning in Clinical Neuroimaging and Radiogenomics in Neuro-oncology - 3rd International Workshop, MLCN 2020, and Second International Workshop, RNO-AI 2020, Held in Conjunction with MICCAI 2020, Proceedings. editor / Seyed Mostafa Kia ; Hassan Mohy-ud-Din ; Ahmed Abdulkadir ; Cher Bass ; Mohamad Habes ; Jane Maryam Rondina ; Chantal Tax ; Hongzhi Wang ; Thomas Wolfers ; Saima Rathore ; Madhura Ingalhalikar. Springer Science and Business Media Deutschland GmbH, 2020. pp. 98-107 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inbook{be9f97124641471ab2e89be866de20d8,

title = "Patch-Based Brain Age Estimation from MR Images",

abstract = "Brain age estimation from Magnetic Resonance Images (MRI) derives the difference between a subject{\textquoteright}s biological brain age and their chronological age. This is a potential biomarker for neurodegeneration, e.g. as part of Alzheimer{\textquoteright}s disease. Early detection of neurodegeneration manifesting as a higher brain age can potentially facilitate better medical care and planning for affected individuals. Many studies have been proposed for the prediction of chronological age from brain MRI using machine learning and specifically deep learning techniques. Contrary to most studies, which use the whole brain volume, in this study, we develop a new deep learning approach that uses 3D patches of the brain as well as convolutional neural networks (CNNs) to develop a localised brain age estimator. In this way, we can obtain a visualization of the regions that play the most important role for estimating brain age, leading to more anatomically driven and interpretable results, and thus confirming relevant literature which suggests that the ventricles and the hippocampus are the areas that are most informative. In addition, we leverage this knowledge in order to improve the overall performance on the task of age estimation by combining the results of different patches using an ensemble method, such as averaging or linear regression. The network is trained on the UK Biobank dataset and the method achieves state-of-the-art results with a Mean Absolute Error of 2.46 years for purely regional estimates, and 2.13 years for an ensemble of patches before bias correction, while 1.96 years after bias correction.",

keywords = "Brain age estimation, Deep learning, Localization, MR Images",

author = "Bintsi, {Kyriaki Margarita} and Vasileios Baltatzis and Arinbj{\"o}rn Kolbeinsson and Alexander Hammers and Daniel Rueckert",

note = "Funding Information: KMB would like to acknowledge funding from the EPSRC Centre for Doctoral Training in Medical Imaging (EP/L015226/1). Publisher Copyright: {\textcopyright} 2020, Springer Nature Switzerland AG. Copyright: Copyright 2021 Elsevier B.V., All rights reserved.; 3rd International Workshop on Machine Learning in Clinical Neuroimaging, MLCN 2020, and 2nd International Workshop on Radiogenomics in Neuro-oncology, RNO-AI 2020, held in conjunction with MICCAI 2020 ; Conference date: 04-10-2020 Through 08-10-2020",

year = "2020",

doi = "10.1007/978-3-030-66843-3_10",

language = "English",

isbn = "9783030668426",

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

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "98--107",

editor = "Kia, {Seyed Mostafa} and Hassan Mohy-ud-Din and Ahmed Abdulkadir and Cher Bass and Mohamad Habes and Rondina, {Jane Maryam} and Chantal Tax and Hongzhi Wang and Thomas Wolfers and Saima Rathore and Madhura Ingalhalikar",

booktitle = "Machine Learning in Clinical Neuroimaging and Radiogenomics in Neuro-oncology - 3rd International Workshop, MLCN 2020, and Second International Workshop, RNO-AI 2020, Held in Conjunction with MICCAI 2020, Proceedings",

address = "Germany",

}

Bintsi, KM, Baltatzis, V, Kolbeinsson, A, Hammers, A & Rueckert, D 2020, Patch-Based Brain Age Estimation from MR Images. in SM Kia, H Mohy-ud-Din, A Abdulkadir, C Bass, M Habes, JM Rondina, C Tax, H Wang, T Wolfers, S Rathore & M Ingalhalikar (eds), Machine Learning in Clinical Neuroimaging and Radiogenomics in Neuro-oncology - 3rd International Workshop, MLCN 2020, and Second International Workshop, RNO-AI 2020, Held in Conjunction with MICCAI 2020, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12449 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 98-107, 3rd International Workshop on Machine Learning in Clinical Neuroimaging, MLCN 2020, and 2nd International Workshop on Radiogenomics in Neuro-oncology, RNO-AI 2020, held in conjunction with MICCAI 2020, Lima, Peru, 4/10/2020. https://doi.org/10.1007/978-3-030-66843-3_10

Patch-Based Brain Age Estimation from MR Images. / Bintsi, Kyriaki Margarita; Baltatzis, Vasileios; Kolbeinsson, Arinbjörn et al.
Machine Learning in Clinical Neuroimaging and Radiogenomics in Neuro-oncology - 3rd International Workshop, MLCN 2020, and Second International Workshop, RNO-AI 2020, Held in Conjunction with MICCAI 2020, Proceedings. ed. / Seyed Mostafa Kia; Hassan Mohy-ud-Din; Ahmed Abdulkadir; Cher Bass; Mohamad Habes; Jane Maryam Rondina; Chantal Tax; Hongzhi Wang; Thomas Wolfers; Saima Rathore; Madhura Ingalhalikar. Springer Science and Business Media Deutschland GmbH, 2020. p. 98-107 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12449 LNCS).

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

TY - CHAP

T1 - Patch-Based Brain Age Estimation from MR Images

AU - Bintsi, Kyriaki Margarita

AU - Baltatzis, Vasileios

AU - Kolbeinsson, Arinbjörn

AU - Hammers, Alexander

AU - Rueckert, Daniel

N1 - Funding Information: KMB would like to acknowledge funding from the EPSRC Centre for Doctoral Training in Medical Imaging (EP/L015226/1). Publisher Copyright: © 2020, Springer Nature Switzerland AG. Copyright: Copyright 2021 Elsevier B.V., All rights reserved.

PY - 2020

Y1 - 2020

N2 - Brain age estimation from Magnetic Resonance Images (MRI) derives the difference between a subject’s biological brain age and their chronological age. This is a potential biomarker for neurodegeneration, e.g. as part of Alzheimer’s disease. Early detection of neurodegeneration manifesting as a higher brain age can potentially facilitate better medical care and planning for affected individuals. Many studies have been proposed for the prediction of chronological age from brain MRI using machine learning and specifically deep learning techniques. Contrary to most studies, which use the whole brain volume, in this study, we develop a new deep learning approach that uses 3D patches of the brain as well as convolutional neural networks (CNNs) to develop a localised brain age estimator. In this way, we can obtain a visualization of the regions that play the most important role for estimating brain age, leading to more anatomically driven and interpretable results, and thus confirming relevant literature which suggests that the ventricles and the hippocampus are the areas that are most informative. In addition, we leverage this knowledge in order to improve the overall performance on the task of age estimation by combining the results of different patches using an ensemble method, such as averaging or linear regression. The network is trained on the UK Biobank dataset and the method achieves state-of-the-art results with a Mean Absolute Error of 2.46 years for purely regional estimates, and 2.13 years for an ensemble of patches before bias correction, while 1.96 years after bias correction.

AB - Brain age estimation from Magnetic Resonance Images (MRI) derives the difference between a subject’s biological brain age and their chronological age. This is a potential biomarker for neurodegeneration, e.g. as part of Alzheimer’s disease. Early detection of neurodegeneration manifesting as a higher brain age can potentially facilitate better medical care and planning for affected individuals. Many studies have been proposed for the prediction of chronological age from brain MRI using machine learning and specifically deep learning techniques. Contrary to most studies, which use the whole brain volume, in this study, we develop a new deep learning approach that uses 3D patches of the brain as well as convolutional neural networks (CNNs) to develop a localised brain age estimator. In this way, we can obtain a visualization of the regions that play the most important role for estimating brain age, leading to more anatomically driven and interpretable results, and thus confirming relevant literature which suggests that the ventricles and the hippocampus are the areas that are most informative. In addition, we leverage this knowledge in order to improve the overall performance on the task of age estimation by combining the results of different patches using an ensemble method, such as averaging or linear regression. The network is trained on the UK Biobank dataset and the method achieves state-of-the-art results with a Mean Absolute Error of 2.46 years for purely regional estimates, and 2.13 years for an ensemble of patches before bias correction, while 1.96 years after bias correction.

KW - Brain age estimation

KW - Deep learning

KW - Localization

KW - MR Images

UR - http://www.scopus.com/inward/record.url?scp=85101529420&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-66843-3_10

DO - 10.1007/978-3-030-66843-3_10

M3 - Conference paper

AN - SCOPUS:85101529420

SN - 9783030668426

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

SP - 98

EP - 107

BT - Machine Learning in Clinical Neuroimaging and Radiogenomics in Neuro-oncology - 3rd International Workshop, MLCN 2020, and Second International Workshop, RNO-AI 2020, Held in Conjunction with MICCAI 2020, Proceedings

A2 - Kia, Seyed Mostafa

A2 - Mohy-ud-Din, Hassan

A2 - Abdulkadir, Ahmed

A2 - Bass, Cher

A2 - Habes, Mohamad

A2 - Rondina, Jane Maryam

A2 - Tax, Chantal

A2 - Wang, Hongzhi

A2 - Wolfers, Thomas

A2 - Rathore, Saima

A2 - Ingalhalikar, Madhura

PB - Springer Science and Business Media Deutschland GmbH

T2 - 3rd International Workshop on Machine Learning in Clinical Neuroimaging, MLCN 2020, and 2nd International Workshop on Radiogenomics in Neuro-oncology, RNO-AI 2020, held in conjunction with MICCAI 2020

Y2 - 4 October 2020 through 8 October 2020

ER -

Bintsi KM, Baltatzis V, Kolbeinsson A, Hammers A, Rueckert D. Patch-Based Brain Age Estimation from MR Images. In Kia SM, Mohy-ud-Din H, Abdulkadir A, Bass C, Habes M, Rondina JM, Tax C, Wang H, Wolfers T, Rathore S, Ingalhalikar M, editors, Machine Learning in Clinical Neuroimaging and Radiogenomics in Neuro-oncology - 3rd International Workshop, MLCN 2020, and Second International Workshop, RNO-AI 2020, Held in Conjunction with MICCAI 2020, Proceedings. Springer Science and Business Media Deutschland GmbH. 2020. p. 98-107. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-66843-3_10