Five-class differential diagnostics of neurodegenerative diseases using random undersampling boosting

Tong Tong, Christian Ledig, Ricardo Guerrero, Andreas Schuh, Juha Koikkalainen, Antti Tolonen, Hanneke Rhodius, Frederik Barkhof, Betty Tijms, Afina W. Lemstra, Hilkka Soininen, Anne M. Remes, Gunhild Waldemar, Steen Hasselbalch, Patrizia Mecocci, Marta Baroni, Jyrki Lötjönen, Wiesje van der Flier, Daniel Rueckert

Research output: Contribution to journalArticleScientificpeer-review

6 Citations (Scopus)

Abstract

Differentiating between different types of neurodegenerative diseases is not only crucial in clinical practice when treatment decisions have to be made, but also has a significant potential for the enrichment of clinical trials. The purpose of this study is to develop a classification framework for distinguishing the four most common neurodegenerative diseases, including Alzheimer's disease, frontotemporal lobe degeneration, Dementia with Lewy bodies and vascular dementia, as well as patients with subjective memory complaints. Different biomarkers including features from images (volume features, region-wise grading features) and non-imaging features (CSF measures) were extracted for each subject. In clinical practice, the prevalence of different dementia types is imbalanced, posing challenges for learning an effective classification model. Therefore, we propose the use of the RUSBoost algorithm in order to train classifiers and to handle the class imbalance training problem. Furthermore, a multi-class feature selection method based on sparsity is integrated into the proposed framework to improve the classification performance. It also provides a way for investigating the importance of different features and regions. Using a dataset of 500 subjects, the proposed framework achieved a high accuracy of 75.2% with a balanced accuracy of 69.3% for the five-class classification using ten-fold cross validation, which is significantly better than the results using support vector machine or random forest, demonstrating the feasibility of the proposed framework to support clinical decision making.
Original languageEnglish
Pages (from-to)613-624
Number of pages12
JournalNeuroImage: Clinical
Volume15
DOIs
Publication statusPublished - 1 Jan 2017
MoE publication typeA1 Journal article-refereed

Fingerprint

Neurodegenerative Diseases
Lewy Body Disease
Frontotemporal Dementia
Vascular Dementia
Dementia
Alzheimer Disease
Biomarkers
Clinical Trials
Learning
Therapeutics

Keywords

  • neurodegenerative diseases
  • differential diagnosis
  • MRI
  • dementia
  • imbalance learning
  • multi-class feature selection

Cite this

Tong, Tong ; Ledig, Christian ; Guerrero, Ricardo ; Schuh, Andreas ; Koikkalainen, Juha ; Tolonen, Antti ; Rhodius, Hanneke ; Barkhof, Frederik ; Tijms, Betty ; Lemstra, Afina W. ; Soininen, Hilkka ; Remes, Anne M. ; Waldemar, Gunhild ; Hasselbalch, Steen ; Mecocci, Patrizia ; Baroni, Marta ; Lötjönen, Jyrki ; Flier, Wiesje van der ; Rueckert, Daniel. / Five-class differential diagnostics of neurodegenerative diseases using random undersampling boosting. In: NeuroImage: Clinical. 2017 ; Vol. 15. pp. 613-624.
@article{56e0bcc78c864633815998e6c7e8c39c,
title = "Five-class differential diagnostics of neurodegenerative diseases using random undersampling boosting",
abstract = "Differentiating between different types of neurodegenerative diseases is not only crucial in clinical practice when treatment decisions have to be made, but also has a significant potential for the enrichment of clinical trials. The purpose of this study is to develop a classification framework for distinguishing the four most common neurodegenerative diseases, including Alzheimer's disease, frontotemporal lobe degeneration, Dementia with Lewy bodies and vascular dementia, as well as patients with subjective memory complaints. Different biomarkers including features from images (volume features, region-wise grading features) and non-imaging features (CSF measures) were extracted for each subject. In clinical practice, the prevalence of different dementia types is imbalanced, posing challenges for learning an effective classification model. Therefore, we propose the use of the RUSBoost algorithm in order to train classifiers and to handle the class imbalance training problem. Furthermore, a multi-class feature selection method based on sparsity is integrated into the proposed framework to improve the classification performance. It also provides a way for investigating the importance of different features and regions. Using a dataset of 500 subjects, the proposed framework achieved a high accuracy of 75.2{\%} with a balanced accuracy of 69.3{\%} for the five-class classification using ten-fold cross validation, which is significantly better than the results using support vector machine or random forest, demonstrating the feasibility of the proposed framework to support clinical decision making.",
keywords = "neurodegenerative diseases, differential diagnosis, MRI, dementia, imbalance learning, multi-class feature selection",
author = "Tong Tong and Christian Ledig and Ricardo Guerrero and Andreas Schuh and Juha Koikkalainen and Antti Tolonen and Hanneke Rhodius and Frederik Barkhof and Betty Tijms and Lemstra, {Afina W.} and Hilkka Soininen and Remes, {Anne M.} and Gunhild Waldemar and Steen Hasselbalch and Patrizia Mecocci and Marta Baroni and Jyrki L{\"o}tj{\"o}nen and Flier, {Wiesje van der} and Daniel Rueckert",
year = "2017",
month = "1",
day = "1",
doi = "10.1016/j.nicl.2017.06.012",
language = "English",
volume = "15",
pages = "613--624",
journal = "NeuroImage: Clinical",
issn = "2213-1582",
publisher = "Elsevier",

}

Tong, T, Ledig, C, Guerrero, R, Schuh, A, Koikkalainen, J, Tolonen, A, Rhodius, H, Barkhof, F, Tijms, B, Lemstra, AW, Soininen, H, Remes, AM, Waldemar, G, Hasselbalch, S, Mecocci, P, Baroni, M, Lötjönen, J, Flier, WVD & Rueckert, D 2017, 'Five-class differential diagnostics of neurodegenerative diseases using random undersampling boosting', NeuroImage: Clinical, vol. 15, pp. 613-624. https://doi.org/10.1016/j.nicl.2017.06.012

Five-class differential diagnostics of neurodegenerative diseases using random undersampling boosting. / Tong, Tong; Ledig, Christian; Guerrero, Ricardo; Schuh, Andreas; Koikkalainen, Juha; Tolonen, Antti; Rhodius, Hanneke; Barkhof, Frederik; Tijms, Betty; Lemstra, Afina W.; Soininen, Hilkka; Remes, Anne M.; Waldemar, Gunhild; Hasselbalch, Steen; Mecocci, Patrizia; Baroni, Marta; Lötjönen, Jyrki; Flier, Wiesje van der; Rueckert, Daniel.

In: NeuroImage: Clinical, Vol. 15, 01.01.2017, p. 613-624.

Research output: Contribution to journalArticleScientificpeer-review

TY - JOUR

T1 - Five-class differential diagnostics of neurodegenerative diseases using random undersampling boosting

AU - Tong, Tong

AU - Ledig, Christian

AU - Guerrero, Ricardo

AU - Schuh, Andreas

AU - Koikkalainen, Juha

AU - Tolonen, Antti

AU - Rhodius, Hanneke

AU - Barkhof, Frederik

AU - Tijms, Betty

AU - Lemstra, Afina W.

AU - Soininen, Hilkka

AU - Remes, Anne M.

AU - Waldemar, Gunhild

AU - Hasselbalch, Steen

AU - Mecocci, Patrizia

AU - Baroni, Marta

AU - Lötjönen, Jyrki

AU - Flier, Wiesje van der

AU - Rueckert, Daniel

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Differentiating between different types of neurodegenerative diseases is not only crucial in clinical practice when treatment decisions have to be made, but also has a significant potential for the enrichment of clinical trials. The purpose of this study is to develop a classification framework for distinguishing the four most common neurodegenerative diseases, including Alzheimer's disease, frontotemporal lobe degeneration, Dementia with Lewy bodies and vascular dementia, as well as patients with subjective memory complaints. Different biomarkers including features from images (volume features, region-wise grading features) and non-imaging features (CSF measures) were extracted for each subject. In clinical practice, the prevalence of different dementia types is imbalanced, posing challenges for learning an effective classification model. Therefore, we propose the use of the RUSBoost algorithm in order to train classifiers and to handle the class imbalance training problem. Furthermore, a multi-class feature selection method based on sparsity is integrated into the proposed framework to improve the classification performance. It also provides a way for investigating the importance of different features and regions. Using a dataset of 500 subjects, the proposed framework achieved a high accuracy of 75.2% with a balanced accuracy of 69.3% for the five-class classification using ten-fold cross validation, which is significantly better than the results using support vector machine or random forest, demonstrating the feasibility of the proposed framework to support clinical decision making.

AB - Differentiating between different types of neurodegenerative diseases is not only crucial in clinical practice when treatment decisions have to be made, but also has a significant potential for the enrichment of clinical trials. The purpose of this study is to develop a classification framework for distinguishing the four most common neurodegenerative diseases, including Alzheimer's disease, frontotemporal lobe degeneration, Dementia with Lewy bodies and vascular dementia, as well as patients with subjective memory complaints. Different biomarkers including features from images (volume features, region-wise grading features) and non-imaging features (CSF measures) were extracted for each subject. In clinical practice, the prevalence of different dementia types is imbalanced, posing challenges for learning an effective classification model. Therefore, we propose the use of the RUSBoost algorithm in order to train classifiers and to handle the class imbalance training problem. Furthermore, a multi-class feature selection method based on sparsity is integrated into the proposed framework to improve the classification performance. It also provides a way for investigating the importance of different features and regions. Using a dataset of 500 subjects, the proposed framework achieved a high accuracy of 75.2% with a balanced accuracy of 69.3% for the five-class classification using ten-fold cross validation, which is significantly better than the results using support vector machine or random forest, demonstrating the feasibility of the proposed framework to support clinical decision making.

KW - neurodegenerative diseases

KW - differential diagnosis

KW - MRI

KW - dementia

KW - imbalance learning

KW - multi-class feature selection

UR - http://www.scopus.com/inward/record.url?scp=85020929654&partnerID=8YFLogxK

U2 - 10.1016/j.nicl.2017.06.012

DO - 10.1016/j.nicl.2017.06.012

M3 - Article

VL - 15

SP - 613

EP - 624

JO - NeuroImage: Clinical

JF - NeuroImage: Clinical

SN - 2213-1582

ER -