Regional abnormality representation learning in structural MRI for AD/MCI diagnosis

Jun Sik Choi, Eunho Lee, Heung-Il Suk

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

Abstract

In this paper, we propose a novel method for MRI-based AD/MCI diagnosis that systematically integrates voxel-based, region-based, and patch-based approaches in a unified framework. Specifically, we parcellate a brain into predefined regions by using anatomical knowledge, i.e., template, and find complex nonlinear relations among voxels, whose intensity denotes the volumetric measure in our case, within each region. Unlike the existing methods that mostly use a cubical or rectangular shape, we regard the anatomical shape of regions as atypical forms of patches. Using the complex nonlinear relations among voxels in each region learned by deep neural networks, we extract a regional abnormality representation. We then make a final clinical decision by integrating the regional abnormality representations over a whole brain. It is noteworthy that the regional abnormality representations allow us to interpret and understand the symptomatic observations of a subject with AD or MCI by mapping and visualizing them in a brain space individually. We validated the efficacy of our method in experiments with baseline MRI dataset in the ADNI cohort by achieving promising performances in three binary classification tasks.

Original languageEnglish
Title of host publicationMachine Learning in Medical Imaging - 9th International Workshop, MLMI 2018, Held in Conjunction with MICCAI 2018, Proceedings
EditorsMingxia Liu, Heung-Il Suk, Yinghuan Shi
PublisherSpringer Verlag
Pages64-72
Number of pages9
ISBN (Print)9783030009182
DOIs
Publication statusPublished - 2018 Jan 1
Event9th International Workshop on Machine Learning in Medical Imaging, MLMI 2018 held in conjunction with the 21st International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2018 - Granada, Spain
Duration: 2018 Sep 162018 Sep 16

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11046 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other9th International Workshop on Machine Learning in Medical Imaging, MLMI 2018 held in conjunction with the 21st International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2018
CountrySpain
CityGranada
Period18/9/1618/9/16

Fingerprint

Magnetic resonance imaging
Brain
Voxel
Patch
Binary Classification
Efficacy
Template
Baseline
Integrate
Learning
Neural Networks
Denote
Experiments
Experiment
Deep neural networks

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Choi, J. S., Lee, E., & Suk, H-I. (2018). Regional abnormality representation learning in structural MRI for AD/MCI diagnosis. In M. Liu, H-I. Suk, & Y. Shi (Eds.), Machine Learning in Medical Imaging - 9th International Workshop, MLMI 2018, Held in Conjunction with MICCAI 2018, Proceedings (pp. 64-72). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11046 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-00919-9_8

Regional abnormality representation learning in structural MRI for AD/MCI diagnosis. / Choi, Jun Sik; Lee, Eunho; Suk, Heung-Il.

Machine Learning in Medical Imaging - 9th International Workshop, MLMI 2018, Held in Conjunction with MICCAI 2018, Proceedings. ed. / Mingxia Liu; Heung-Il Suk; Yinghuan Shi. Springer Verlag, 2018. p. 64-72 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11046 LNCS).

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

Choi, JS, Lee, E & Suk, H-I 2018, Regional abnormality representation learning in structural MRI for AD/MCI diagnosis. in M Liu, H-I Suk & Y Shi (eds), Machine Learning in Medical Imaging - 9th International Workshop, MLMI 2018, Held in Conjunction with MICCAI 2018, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11046 LNCS, Springer Verlag, pp. 64-72, 9th International Workshop on Machine Learning in Medical Imaging, MLMI 2018 held in conjunction with the 21st International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2018, Granada, Spain, 18/9/16. https://doi.org/10.1007/978-3-030-00919-9_8
Choi JS, Lee E, Suk H-I. Regional abnormality representation learning in structural MRI for AD/MCI diagnosis. In Liu M, Suk H-I, Shi Y, editors, Machine Learning in Medical Imaging - 9th International Workshop, MLMI 2018, Held in Conjunction with MICCAI 2018, Proceedings. Springer Verlag. 2018. p. 64-72. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-030-00919-9_8
Choi, Jun Sik ; Lee, Eunho ; Suk, Heung-Il. / Regional abnormality representation learning in structural MRI for AD/MCI diagnosis. Machine Learning in Medical Imaging - 9th International Workshop, MLMI 2018, Held in Conjunction with MICCAI 2018, Proceedings. editor / Mingxia Liu ; Heung-Il Suk ; Yinghuan Shi. Springer Verlag, 2018. pp. 64-72 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{4bb1f9f23ee945a08485bfa860f50801,
title = "Regional abnormality representation learning in structural MRI for AD/MCI diagnosis",
abstract = "In this paper, we propose a novel method for MRI-based AD/MCI diagnosis that systematically integrates voxel-based, region-based, and patch-based approaches in a unified framework. Specifically, we parcellate a brain into predefined regions by using anatomical knowledge, i.e., template, and find complex nonlinear relations among voxels, whose intensity denotes the volumetric measure in our case, within each region. Unlike the existing methods that mostly use a cubical or rectangular shape, we regard the anatomical shape of regions as atypical forms of patches. Using the complex nonlinear relations among voxels in each region learned by deep neural networks, we extract a regional abnormality representation. We then make a final clinical decision by integrating the regional abnormality representations over a whole brain. It is noteworthy that the regional abnormality representations allow us to interpret and understand the symptomatic observations of a subject with AD or MCI by mapping and visualizing them in a brain space individually. We validated the efficacy of our method in experiments with baseline MRI dataset in the ADNI cohort by achieving promising performances in three binary classification tasks.",
author = "Choi, {Jun Sik} and Eunho Lee and Heung-Il Suk",
year = "2018",
month = "1",
day = "1",
doi = "10.1007/978-3-030-00919-9_8",
language = "English",
isbn = "9783030009182",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "64--72",
editor = "Mingxia Liu and Heung-Il Suk and Yinghuan Shi",
booktitle = "Machine Learning in Medical Imaging - 9th International Workshop, MLMI 2018, Held in Conjunction with MICCAI 2018, Proceedings",

}

TY - GEN

T1 - Regional abnormality representation learning in structural MRI for AD/MCI diagnosis

AU - Choi, Jun Sik

AU - Lee, Eunho

AU - Suk, Heung-Il

PY - 2018/1/1

Y1 - 2018/1/1

N2 - In this paper, we propose a novel method for MRI-based AD/MCI diagnosis that systematically integrates voxel-based, region-based, and patch-based approaches in a unified framework. Specifically, we parcellate a brain into predefined regions by using anatomical knowledge, i.e., template, and find complex nonlinear relations among voxels, whose intensity denotes the volumetric measure in our case, within each region. Unlike the existing methods that mostly use a cubical or rectangular shape, we regard the anatomical shape of regions as atypical forms of patches. Using the complex nonlinear relations among voxels in each region learned by deep neural networks, we extract a regional abnormality representation. We then make a final clinical decision by integrating the regional abnormality representations over a whole brain. It is noteworthy that the regional abnormality representations allow us to interpret and understand the symptomatic observations of a subject with AD or MCI by mapping and visualizing them in a brain space individually. We validated the efficacy of our method in experiments with baseline MRI dataset in the ADNI cohort by achieving promising performances in three binary classification tasks.

AB - In this paper, we propose a novel method for MRI-based AD/MCI diagnosis that systematically integrates voxel-based, region-based, and patch-based approaches in a unified framework. Specifically, we parcellate a brain into predefined regions by using anatomical knowledge, i.e., template, and find complex nonlinear relations among voxels, whose intensity denotes the volumetric measure in our case, within each region. Unlike the existing methods that mostly use a cubical or rectangular shape, we regard the anatomical shape of regions as atypical forms of patches. Using the complex nonlinear relations among voxels in each region learned by deep neural networks, we extract a regional abnormality representation. We then make a final clinical decision by integrating the regional abnormality representations over a whole brain. It is noteworthy that the regional abnormality representations allow us to interpret and understand the symptomatic observations of a subject with AD or MCI by mapping and visualizing them in a brain space individually. We validated the efficacy of our method in experiments with baseline MRI dataset in the ADNI cohort by achieving promising performances in three binary classification tasks.

UR - http://www.scopus.com/inward/record.url?scp=85054479364&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85054479364&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-00919-9_8

DO - 10.1007/978-3-030-00919-9_8

M3 - Conference contribution

AN - SCOPUS:85054479364

SN - 9783030009182

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

SP - 64

EP - 72

BT - Machine Learning in Medical Imaging - 9th International Workshop, MLMI 2018, Held in Conjunction with MICCAI 2018, Proceedings

A2 - Liu, Mingxia

A2 - Suk, Heung-Il

A2 - Shi, Yinghuan

PB - Springer Verlag

ER -