Deep-learning-based classification of FDG-PET data for Alzheimer's disease categories

Shibani Singh, Anant Srivastava, Liang Mi, Richard John Caselli, Kewei Chen, Dhruman Goradia, Eric M. Reiman, Yalin Wang

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

3 Citations (Scopus)

Abstract

Fluorodeoxyglucose (FDG) positron emission tomography (PET) measures the decline in the regional cerebral metabolic rate for glucose, offering a reliable metabolic biomarker even on presymptomatic Alzheimer's disease (AD) patients. PET scans provide functional information that is unique and unavailable using other types of imaging. However, the computational efficacy of FDG-PET data alone, for the classification of various Alzheimers Diagnostic categories, has not been well studied. This motivates us to correctly discriminate various AD Diagnostic categories using FDG-PET data. Deep learning has improved state-of-the-art classification accuracies in the areas of speech, signal, image, video, text mining and recognition. We propose novel methods that involve probabilistic principal component analysis on max-pooled data and mean-pooled data for dimensionality reduction, and multilayer feed forward neural network which performs binary classification. Our experimental dataset consists of baseline data of subjects including 186 cognitively unimpaired (CU) subjects, 336 mild cognitive impairment (MCI) subjects with 158 Late MCI and 178 Early MCI, and 146 AD patients from Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. We measured F1-measure, precision, recall, negative and positive predictive values with a 10-fold cross validation scheme. Our results indicate that our designed classifiers achieve competitive results while max pooling achieves better classification performance compared to mean-pooled features. Our deep model based research may advance FDG-PET analysis by demonstrating their potential as an effective imaging biomarker of AD.

Original languageEnglish (US)
Title of host publication13th International Conference on Medical Information Processing and Analysis
PublisherSPIE
Volume10572
ISBN (Electronic)9781510616332
DOIs
StatePublished - Jan 1 2017
Event13th International Conference on Medical Information Processing and Analysis, SIPAIM 2017 - San Andres Island, Colombia
Duration: Oct 5 2017Oct 7 2017

Other

Other13th International Conference on Medical Information Processing and Analysis, SIPAIM 2017
CountryColombia
CitySan Andres Island
Period10/5/1710/7/17

Fingerprint

Positron Emission Tomography
Positron emission tomography
Alzheimer's Disease
learning
positrons
tomography
impairment
biomarkers
Biomarkers
Diagnostics
Imaging
Neuroimaging
Imaging techniques
Binary Classification
Probabilistic Analysis
Pooling
Speech Signal
Feedforward neural networks
Text Mining
Feedforward Neural Networks

Keywords

  • Alzheimers
  • Cross Validation
  • Deep Learning
  • Dimensionality Reduction
  • Multilayer Perceptrons
  • Neural Networks
  • PET

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Condensed Matter Physics
  • Computer Science Applications
  • Applied Mathematics
  • Electrical and Electronic Engineering

Cite this

Singh, S., Srivastava, A., Mi, L., Caselli, R. J., Chen, K., Goradia, D., ... Wang, Y. (2017). Deep-learning-based classification of FDG-PET data for Alzheimer's disease categories. In 13th International Conference on Medical Information Processing and Analysis (Vol. 10572). [105720J] SPIE. https://doi.org/10.1117/12.2294537

Deep-learning-based classification of FDG-PET data for Alzheimer's disease categories. / Singh, Shibani; Srivastava, Anant; Mi, Liang; Caselli, Richard John; Chen, Kewei; Goradia, Dhruman; Reiman, Eric M.; Wang, Yalin.

13th International Conference on Medical Information Processing and Analysis. Vol. 10572 SPIE, 2017. 105720J.

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

Singh, S, Srivastava, A, Mi, L, Caselli, RJ, Chen, K, Goradia, D, Reiman, EM & Wang, Y 2017, Deep-learning-based classification of FDG-PET data for Alzheimer's disease categories. in 13th International Conference on Medical Information Processing and Analysis. vol. 10572, 105720J, SPIE, 13th International Conference on Medical Information Processing and Analysis, SIPAIM 2017, San Andres Island, Colombia, 10/5/17. https://doi.org/10.1117/12.2294537
Singh S, Srivastava A, Mi L, Caselli RJ, Chen K, Goradia D et al. Deep-learning-based classification of FDG-PET data for Alzheimer's disease categories. In 13th International Conference on Medical Information Processing and Analysis. Vol. 10572. SPIE. 2017. 105720J https://doi.org/10.1117/12.2294537
Singh, Shibani ; Srivastava, Anant ; Mi, Liang ; Caselli, Richard John ; Chen, Kewei ; Goradia, Dhruman ; Reiman, Eric M. ; Wang, Yalin. / Deep-learning-based classification of FDG-PET data for Alzheimer's disease categories. 13th International Conference on Medical Information Processing and Analysis. Vol. 10572 SPIE, 2017.
@inproceedings{2e71f7db735149e9be2617de3f80683c,
title = "Deep-learning-based classification of FDG-PET data for Alzheimer's disease categories",
abstract = "Fluorodeoxyglucose (FDG) positron emission tomography (PET) measures the decline in the regional cerebral metabolic rate for glucose, offering a reliable metabolic biomarker even on presymptomatic Alzheimer's disease (AD) patients. PET scans provide functional information that is unique and unavailable using other types of imaging. However, the computational efficacy of FDG-PET data alone, for the classification of various Alzheimers Diagnostic categories, has not been well studied. This motivates us to correctly discriminate various AD Diagnostic categories using FDG-PET data. Deep learning has improved state-of-the-art classification accuracies in the areas of speech, signal, image, video, text mining and recognition. We propose novel methods that involve probabilistic principal component analysis on max-pooled data and mean-pooled data for dimensionality reduction, and multilayer feed forward neural network which performs binary classification. Our experimental dataset consists of baseline data of subjects including 186 cognitively unimpaired (CU) subjects, 336 mild cognitive impairment (MCI) subjects with 158 Late MCI and 178 Early MCI, and 146 AD patients from Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. We measured F1-measure, precision, recall, negative and positive predictive values with a 10-fold cross validation scheme. Our results indicate that our designed classifiers achieve competitive results while max pooling achieves better classification performance compared to mean-pooled features. Our deep model based research may advance FDG-PET analysis by demonstrating their potential as an effective imaging biomarker of AD.",
keywords = "Alzheimers, Cross Validation, Deep Learning, Dimensionality Reduction, Multilayer Perceptrons, Neural Networks, PET",
author = "Shibani Singh and Anant Srivastava and Liang Mi and Caselli, {Richard John} and Kewei Chen and Dhruman Goradia and Reiman, {Eric M.} and Yalin Wang",
year = "2017",
month = "1",
day = "1",
doi = "10.1117/12.2294537",
language = "English (US)",
volume = "10572",
booktitle = "13th International Conference on Medical Information Processing and Analysis",
publisher = "SPIE",

}

TY - GEN

T1 - Deep-learning-based classification of FDG-PET data for Alzheimer's disease categories

AU - Singh, Shibani

AU - Srivastava, Anant

AU - Mi, Liang

AU - Caselli, Richard John

AU - Chen, Kewei

AU - Goradia, Dhruman

AU - Reiman, Eric M.

AU - Wang, Yalin

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Fluorodeoxyglucose (FDG) positron emission tomography (PET) measures the decline in the regional cerebral metabolic rate for glucose, offering a reliable metabolic biomarker even on presymptomatic Alzheimer's disease (AD) patients. PET scans provide functional information that is unique and unavailable using other types of imaging. However, the computational efficacy of FDG-PET data alone, for the classification of various Alzheimers Diagnostic categories, has not been well studied. This motivates us to correctly discriminate various AD Diagnostic categories using FDG-PET data. Deep learning has improved state-of-the-art classification accuracies in the areas of speech, signal, image, video, text mining and recognition. We propose novel methods that involve probabilistic principal component analysis on max-pooled data and mean-pooled data for dimensionality reduction, and multilayer feed forward neural network which performs binary classification. Our experimental dataset consists of baseline data of subjects including 186 cognitively unimpaired (CU) subjects, 336 mild cognitive impairment (MCI) subjects with 158 Late MCI and 178 Early MCI, and 146 AD patients from Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. We measured F1-measure, precision, recall, negative and positive predictive values with a 10-fold cross validation scheme. Our results indicate that our designed classifiers achieve competitive results while max pooling achieves better classification performance compared to mean-pooled features. Our deep model based research may advance FDG-PET analysis by demonstrating their potential as an effective imaging biomarker of AD.

AB - Fluorodeoxyglucose (FDG) positron emission tomography (PET) measures the decline in the regional cerebral metabolic rate for glucose, offering a reliable metabolic biomarker even on presymptomatic Alzheimer's disease (AD) patients. PET scans provide functional information that is unique and unavailable using other types of imaging. However, the computational efficacy of FDG-PET data alone, for the classification of various Alzheimers Diagnostic categories, has not been well studied. This motivates us to correctly discriminate various AD Diagnostic categories using FDG-PET data. Deep learning has improved state-of-the-art classification accuracies in the areas of speech, signal, image, video, text mining and recognition. We propose novel methods that involve probabilistic principal component analysis on max-pooled data and mean-pooled data for dimensionality reduction, and multilayer feed forward neural network which performs binary classification. Our experimental dataset consists of baseline data of subjects including 186 cognitively unimpaired (CU) subjects, 336 mild cognitive impairment (MCI) subjects with 158 Late MCI and 178 Early MCI, and 146 AD patients from Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. We measured F1-measure, precision, recall, negative and positive predictive values with a 10-fold cross validation scheme. Our results indicate that our designed classifiers achieve competitive results while max pooling achieves better classification performance compared to mean-pooled features. Our deep model based research may advance FDG-PET analysis by demonstrating their potential as an effective imaging biomarker of AD.

KW - Alzheimers

KW - Cross Validation

KW - Deep Learning

KW - Dimensionality Reduction

KW - Multilayer Perceptrons

KW - Neural Networks

KW - PET

UR - http://www.scopus.com/inward/record.url?scp=85038438092&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85038438092&partnerID=8YFLogxK

U2 - 10.1117/12.2294537

DO - 10.1117/12.2294537

M3 - Conference contribution

AN - SCOPUS:85038438092

VL - 10572

BT - 13th International Conference on Medical Information Processing and Analysis

PB - SPIE

ER -