Advanced

Binary classification of F-18-flutemetamol PET using machine learning: Comparison with visual reads and structural MRI

Vandenberghe, Rik ; Nelissen, Natalie ; Salmon, Eric ; Ivanoiu, Adrian ; Hasselbalch, Steen ; Andersen, Allan ; Korner, Alex ; Minthon, Lennart LU ; Brooks, David J. and Van Laere, Koen , et al. (2013) In NeuroImage 64. p.517-525
Abstract
F-18-flutemetamol is a positron emission tomography (PET) tracer for in vivo amyloid imaging. The ability to classify amyloid scans in a binary manner as 'normal' versus 'Alzheimer-like', is of high clinical relevance. We evaluated whether a supervised machine learning technique, support vector machines (SVM), can replicate the assignments made by visual readers blind to the clinical diagnosis, which image components have highest diagnostic value according to SVM and how F-18-flutemetamol-based classification using SVM relates to structural MRI-based classification using SVM within the same subjects. By means of SVM with a linear kernel, we analyzed F-18-flutemetamol scans and volumetric MRI scans from 72 cases from the F-18-flutemetamol... (More)
F-18-flutemetamol is a positron emission tomography (PET) tracer for in vivo amyloid imaging. The ability to classify amyloid scans in a binary manner as 'normal' versus 'Alzheimer-like', is of high clinical relevance. We evaluated whether a supervised machine learning technique, support vector machines (SVM), can replicate the assignments made by visual readers blind to the clinical diagnosis, which image components have highest diagnostic value according to SVM and how F-18-flutemetamol-based classification using SVM relates to structural MRI-based classification using SVM within the same subjects. By means of SVM with a linear kernel, we analyzed F-18-flutemetamol scans and volumetric MRI scans from 72 cases from the F-18-flutemetamol phase 2 study (27 clinically probable Alzheimer's disease (AD), 20 amnestic mild cognitive impairment (MCI), 25 controls). In a leave-one-out approach, we trained the F-18-flutemetamol based classifier by means of the visual reads and tested whether the classifier was able to reproduce the assignment based on visual reads and which voxels had the highest feature weights. The F-18-flutemetamol based classifier was able to replicate the assignments obtained by visual reads with 100% accuracy. The voxels with highest feature weights were in the striatum, precuneus, cingulate and middle frontal gyrus. Second, to determine concordance between the gray matter volume- and the F-18-flutemetamol-based classification, we trained the classifier with the clinical diagnosis as gold standard. Overall sensitivity of the F-18-flutemetamol- and the gray matter volume-based classifiers were identical (85.2%), albeit with discordant classification in three cases. Specificity of the F-18-flutemetamol based classifier was 92% compared to 68% for MRI. In the MCI group, the F-18-flutemetamol based classifier distinguished more reliably between converters and non-converters than the gray matter-based classifier. The visual read-based binary classification of F-18-flutemetamol scans can be replicated using SVM. In this sample the specificity of F-18-flutemetamol based SVM for distinguishing AD from controls is higher than that of gray matter volume-based SVM. (C) 2012 Elsevier Inc. All rights reserved. (Less)
Please use this url to cite or link to this publication:
author
, et al. (More)
(Less)
organization
publishing date
type
Contribution to journal
publication status
published
subject
keywords
Amyloid imaging, Alzheimer's disease, Mild cognitive impairment, MRI, volumetry
in
NeuroImage
volume
64
pages
517 - 525
publisher
Elsevier
external identifiers
  • wos:000312504200050
  • scopus:84867468075
  • pmid:22982358
ISSN
1095-9572
DOI
10.1016/j.neuroimage.2012.09.015
language
English
LU publication?
yes
id
ed92f7f9-02cc-44ad-98f0-93b7e9384e49 (old id 3481316)
date added to LUP
2016-04-01 09:57:53
date last changed
2019-04-23 01:05:46
@article{ed92f7f9-02cc-44ad-98f0-93b7e9384e49,
  abstract     = {F-18-flutemetamol is a positron emission tomography (PET) tracer for in vivo amyloid imaging. The ability to classify amyloid scans in a binary manner as 'normal' versus 'Alzheimer-like', is of high clinical relevance. We evaluated whether a supervised machine learning technique, support vector machines (SVM), can replicate the assignments made by visual readers blind to the clinical diagnosis, which image components have highest diagnostic value according to SVM and how F-18-flutemetamol-based classification using SVM relates to structural MRI-based classification using SVM within the same subjects. By means of SVM with a linear kernel, we analyzed F-18-flutemetamol scans and volumetric MRI scans from 72 cases from the F-18-flutemetamol phase 2 study (27 clinically probable Alzheimer's disease (AD), 20 amnestic mild cognitive impairment (MCI), 25 controls). In a leave-one-out approach, we trained the F-18-flutemetamol based classifier by means of the visual reads and tested whether the classifier was able to reproduce the assignment based on visual reads and which voxels had the highest feature weights. The F-18-flutemetamol based classifier was able to replicate the assignments obtained by visual reads with 100% accuracy. The voxels with highest feature weights were in the striatum, precuneus, cingulate and middle frontal gyrus. Second, to determine concordance between the gray matter volume- and the F-18-flutemetamol-based classification, we trained the classifier with the clinical diagnosis as gold standard. Overall sensitivity of the F-18-flutemetamol- and the gray matter volume-based classifiers were identical (85.2%), albeit with discordant classification in three cases. Specificity of the F-18-flutemetamol based classifier was 92% compared to 68% for MRI. In the MCI group, the F-18-flutemetamol based classifier distinguished more reliably between converters and non-converters than the gray matter-based classifier. The visual read-based binary classification of F-18-flutemetamol scans can be replicated using SVM. In this sample the specificity of F-18-flutemetamol based SVM for distinguishing AD from controls is higher than that of gray matter volume-based SVM. (C) 2012 Elsevier Inc. All rights reserved.},
  author       = {Vandenberghe, Rik and Nelissen, Natalie and Salmon, Eric and Ivanoiu, Adrian and Hasselbalch, Steen and Andersen, Allan and Korner, Alex and Minthon, Lennart and Brooks, David J. and Van Laere, Koen and Dupont, Patrick},
  issn         = {1095-9572},
  language     = {eng},
  pages        = {517--525},
  publisher    = {Elsevier},
  series       = {NeuroImage},
  title        = {Binary classification of F-18-flutemetamol PET using machine learning: Comparison with visual reads and structural MRI},
  url          = {http://dx.doi.org/10.1016/j.neuroimage.2012.09.015},
  doi          = {10.1016/j.neuroimage.2012.09.015},
  volume       = {64},
  year         = {2013},
}