Machine learning based hierarchical classification of frontotemporal dementia and Alzheimer's disease

Jun Pyo Kim, Jeonghun Kim, Yu Hyun Park, Seong Beom Park, Jin San Lee, Sole Yoo, Eun Joo Kim, Hee Jin Kim, Duk L. Na, Jesse A. Brown, Samuel N. Lockhart, Sang Won Seo, Jun Kyung Seong

Research output: Contribution to journalArticle

Abstract

Background: In a clinical setting, an individual subject classification model rather than a group analysis would be more informative. Specifically, the subtlety of cortical atrophy in some frontotemporal dementia (FTD) patients and overlapping patterns of atrophy among three FTD clinical syndromes including behavioral variant FTD (bvFTD), non-fluent/agrammatic variant primary progressive aphasia (nfvPPA), and semantic variant PPA (svPPA) give rise to the need for classification models at the individual level. In this study, we aimed to classify each individual subject into one of the diagnostic categories in a hierarchical manner by employing a machine learning-based classification method. Methods: We recruited 143 patients with FTD, 50 patients with Alzheimer's disease (AD) dementia, and 146 cognitively normal subjects. All subjects underwent a three-dimensional volumetric brain magnetic resonance imaging (MRI) scan, and cortical thickness was measured using FreeSurfer. We applied the Laplace Beltrami operator to reduce noise in the cortical thickness data and to reduce the dimension of the feature vector. Classifiers were constructed by applying both principal component analysis and linear discriminant analysis to the cortical thickness data. For the hierarchical classification, we trained four classifiers using different pairs of groups: Step 1 - CN vs. FTD + AD, Step 2 - FTD vs. AD, Step 3 - bvFTD vs. PPA, Step 4 - svPPA vs. nfvPPA. To evaluate the classification performance for each step, we used a10-fold cross-validation approach, performed 1000 times for reliability. Results: The classification accuracy of the entire hierarchical classification tree was 75.8%, which was higher than that of the non-hierarchical classifier (73.0%). The classification accuracies of steps 1–4 were 86.1%, 90.8%, 86.9%, and 92.1%, respectively. Changes in the right frontotemporal area were critical for discriminating behavioral variant FTD from PPA. The left frontal lobe discriminated nfvPPA from svPPA, while the bilateral anterior temporal regions were critical for identifying svPPA. Conclusions: In the present study, our automated classifier successfully classified FTD clinical subtypes with good to excellent accuracy. Our classifier may help clinicians diagnose FTD subtypes with subtle cortical atrophy and facilitate appropriate specific interventions.

Original languageEnglish
Article number101811
JournalNeuroImage: Clinical
Volume23
DOIs
Publication statusPublished - 2019 Jan 1

Fingerprint

Frontotemporal Dementia
Alzheimer Disease
Primary Progressive Aphasia
Semantics
Atrophy
Machine Learning
Frontal Lobe
Discriminant Analysis
Temporal Lobe
Principal Component Analysis
Noise
Magnetic Resonance Imaging

Keywords

  • Classification model
  • Frontotemporal dementia
  • Machine learning

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging
  • Neurology
  • Clinical Neurology
  • Cognitive Neuroscience

Cite this

Machine learning based hierarchical classification of frontotemporal dementia and Alzheimer's disease. / Kim, Jun Pyo; Kim, Jeonghun; Park, Yu Hyun; Park, Seong Beom; Lee, Jin San; Yoo, Sole; Kim, Eun Joo; Kim, Hee Jin; Na, Duk L.; Brown, Jesse A.; Lockhart, Samuel N.; Seo, Sang Won; Seong, Jun Kyung.

In: NeuroImage: Clinical, Vol. 23, 101811, 01.01.2019.

Research output: Contribution to journalArticle

Kim, JP, Kim, J, Park, YH, Park, SB, Lee, JS, Yoo, S, Kim, EJ, Kim, HJ, Na, DL, Brown, JA, Lockhart, SN, Seo, SW & Seong, JK 2019, 'Machine learning based hierarchical classification of frontotemporal dementia and Alzheimer's disease', NeuroImage: Clinical, vol. 23, 101811. https://doi.org/10.1016/j.nicl.2019.101811
Kim, Jun Pyo ; Kim, Jeonghun ; Park, Yu Hyun ; Park, Seong Beom ; Lee, Jin San ; Yoo, Sole ; Kim, Eun Joo ; Kim, Hee Jin ; Na, Duk L. ; Brown, Jesse A. ; Lockhart, Samuel N. ; Seo, Sang Won ; Seong, Jun Kyung. / Machine learning based hierarchical classification of frontotemporal dementia and Alzheimer's disease. In: NeuroImage: Clinical. 2019 ; Vol. 23.
@article{cf611b97d5c44de09e5f2eb711bd59c7,
title = "Machine learning based hierarchical classification of frontotemporal dementia and Alzheimer's disease",
abstract = "Background: In a clinical setting, an individual subject classification model rather than a group analysis would be more informative. Specifically, the subtlety of cortical atrophy in some frontotemporal dementia (FTD) patients and overlapping patterns of atrophy among three FTD clinical syndromes including behavioral variant FTD (bvFTD), non-fluent/agrammatic variant primary progressive aphasia (nfvPPA), and semantic variant PPA (svPPA) give rise to the need for classification models at the individual level. In this study, we aimed to classify each individual subject into one of the diagnostic categories in a hierarchical manner by employing a machine learning-based classification method. Methods: We recruited 143 patients with FTD, 50 patients with Alzheimer's disease (AD) dementia, and 146 cognitively normal subjects. All subjects underwent a three-dimensional volumetric brain magnetic resonance imaging (MRI) scan, and cortical thickness was measured using FreeSurfer. We applied the Laplace Beltrami operator to reduce noise in the cortical thickness data and to reduce the dimension of the feature vector. Classifiers were constructed by applying both principal component analysis and linear discriminant analysis to the cortical thickness data. For the hierarchical classification, we trained four classifiers using different pairs of groups: Step 1 - CN vs. FTD + AD, Step 2 - FTD vs. AD, Step 3 - bvFTD vs. PPA, Step 4 - svPPA vs. nfvPPA. To evaluate the classification performance for each step, we used a10-fold cross-validation approach, performed 1000 times for reliability. Results: The classification accuracy of the entire hierarchical classification tree was 75.8{\%}, which was higher than that of the non-hierarchical classifier (73.0{\%}). The classification accuracies of steps 1–4 were 86.1{\%}, 90.8{\%}, 86.9{\%}, and 92.1{\%}, respectively. Changes in the right frontotemporal area were critical for discriminating behavioral variant FTD from PPA. The left frontal lobe discriminated nfvPPA from svPPA, while the bilateral anterior temporal regions were critical for identifying svPPA. Conclusions: In the present study, our automated classifier successfully classified FTD clinical subtypes with good to excellent accuracy. Our classifier may help clinicians diagnose FTD subtypes with subtle cortical atrophy and facilitate appropriate specific interventions.",
keywords = "Classification model, Frontotemporal dementia, Machine learning",
author = "Kim, {Jun Pyo} and Jeonghun Kim and Park, {Yu Hyun} and Park, {Seong Beom} and Lee, {Jin San} and Sole Yoo and Kim, {Eun Joo} and Kim, {Hee Jin} and Na, {Duk L.} and Brown, {Jesse A.} and Lockhart, {Samuel N.} and Seo, {Sang Won} and Seong, {Jun Kyung}",
year = "2019",
month = "1",
day = "1",
doi = "10.1016/j.nicl.2019.101811",
language = "English",
volume = "23",
journal = "NeuroImage: Clinical",
issn = "2213-1582",
publisher = "Elsevier BV",

}

TY - JOUR

T1 - Machine learning based hierarchical classification of frontotemporal dementia and Alzheimer's disease

AU - Kim, Jun Pyo

AU - Kim, Jeonghun

AU - Park, Yu Hyun

AU - Park, Seong Beom

AU - Lee, Jin San

AU - Yoo, Sole

AU - Kim, Eun Joo

AU - Kim, Hee Jin

AU - Na, Duk L.

AU - Brown, Jesse A.

AU - Lockhart, Samuel N.

AU - Seo, Sang Won

AU - Seong, Jun Kyung

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Background: In a clinical setting, an individual subject classification model rather than a group analysis would be more informative. Specifically, the subtlety of cortical atrophy in some frontotemporal dementia (FTD) patients and overlapping patterns of atrophy among three FTD clinical syndromes including behavioral variant FTD (bvFTD), non-fluent/agrammatic variant primary progressive aphasia (nfvPPA), and semantic variant PPA (svPPA) give rise to the need for classification models at the individual level. In this study, we aimed to classify each individual subject into one of the diagnostic categories in a hierarchical manner by employing a machine learning-based classification method. Methods: We recruited 143 patients with FTD, 50 patients with Alzheimer's disease (AD) dementia, and 146 cognitively normal subjects. All subjects underwent a three-dimensional volumetric brain magnetic resonance imaging (MRI) scan, and cortical thickness was measured using FreeSurfer. We applied the Laplace Beltrami operator to reduce noise in the cortical thickness data and to reduce the dimension of the feature vector. Classifiers were constructed by applying both principal component analysis and linear discriminant analysis to the cortical thickness data. For the hierarchical classification, we trained four classifiers using different pairs of groups: Step 1 - CN vs. FTD + AD, Step 2 - FTD vs. AD, Step 3 - bvFTD vs. PPA, Step 4 - svPPA vs. nfvPPA. To evaluate the classification performance for each step, we used a10-fold cross-validation approach, performed 1000 times for reliability. Results: The classification accuracy of the entire hierarchical classification tree was 75.8%, which was higher than that of the non-hierarchical classifier (73.0%). The classification accuracies of steps 1–4 were 86.1%, 90.8%, 86.9%, and 92.1%, respectively. Changes in the right frontotemporal area were critical for discriminating behavioral variant FTD from PPA. The left frontal lobe discriminated nfvPPA from svPPA, while the bilateral anterior temporal regions were critical for identifying svPPA. Conclusions: In the present study, our automated classifier successfully classified FTD clinical subtypes with good to excellent accuracy. Our classifier may help clinicians diagnose FTD subtypes with subtle cortical atrophy and facilitate appropriate specific interventions.

AB - Background: In a clinical setting, an individual subject classification model rather than a group analysis would be more informative. Specifically, the subtlety of cortical atrophy in some frontotemporal dementia (FTD) patients and overlapping patterns of atrophy among three FTD clinical syndromes including behavioral variant FTD (bvFTD), non-fluent/agrammatic variant primary progressive aphasia (nfvPPA), and semantic variant PPA (svPPA) give rise to the need for classification models at the individual level. In this study, we aimed to classify each individual subject into one of the diagnostic categories in a hierarchical manner by employing a machine learning-based classification method. Methods: We recruited 143 patients with FTD, 50 patients with Alzheimer's disease (AD) dementia, and 146 cognitively normal subjects. All subjects underwent a three-dimensional volumetric brain magnetic resonance imaging (MRI) scan, and cortical thickness was measured using FreeSurfer. We applied the Laplace Beltrami operator to reduce noise in the cortical thickness data and to reduce the dimension of the feature vector. Classifiers were constructed by applying both principal component analysis and linear discriminant analysis to the cortical thickness data. For the hierarchical classification, we trained four classifiers using different pairs of groups: Step 1 - CN vs. FTD + AD, Step 2 - FTD vs. AD, Step 3 - bvFTD vs. PPA, Step 4 - svPPA vs. nfvPPA. To evaluate the classification performance for each step, we used a10-fold cross-validation approach, performed 1000 times for reliability. Results: The classification accuracy of the entire hierarchical classification tree was 75.8%, which was higher than that of the non-hierarchical classifier (73.0%). The classification accuracies of steps 1–4 were 86.1%, 90.8%, 86.9%, and 92.1%, respectively. Changes in the right frontotemporal area were critical for discriminating behavioral variant FTD from PPA. The left frontal lobe discriminated nfvPPA from svPPA, while the bilateral anterior temporal regions were critical for identifying svPPA. Conclusions: In the present study, our automated classifier successfully classified FTD clinical subtypes with good to excellent accuracy. Our classifier may help clinicians diagnose FTD subtypes with subtle cortical atrophy and facilitate appropriate specific interventions.

KW - Classification model

KW - Frontotemporal dementia

KW - Machine learning

UR - http://www.scopus.com/inward/record.url?scp=85064073581&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85064073581&partnerID=8YFLogxK

U2 - 10.1016/j.nicl.2019.101811

DO - 10.1016/j.nicl.2019.101811

M3 - Article

C2 - 30981204

AN - SCOPUS:85064073581

VL - 23

JO - NeuroImage: Clinical

JF - NeuroImage: Clinical

SN - 2213-1582

M1 - 101811

ER -